SAJCH Vol 11, No 1 (2017)

CHILD HEALTH THE SOUTH AFRICAN JOURNAL OF

March 2017

• • • • • •

Volume 11

No. 1

First do no harm – medication errors in paediatrics Nutritional status of school children in Limpopo Oral health promotion in KwaZulu-Natal Selenium status and HIV infection Pubertal development in Nigerian children The role of HIV in severe acute malnutrition mortality

CHILD HEALTH THE SOUTH AFRICAN JOURNAL OF

MARCH 2017

Volume 11

No. 1

CONTENTS Editorial

2 The new Nelson Mandela Children's Hospital

P A Cooper

EDITOR J M Pettifor FOUNDING EDITOR N P Khumalo EDITORIAL BOARD Prof. M Adhikari (University of KwaZuluNatal, Durban) Prof. M Kruger (Stellenbosch University) Prof. H Rode (Red Cross War Memorial Children's Hospital, Cape Town) Prof. L Spitz (Emeritus Nuffield Professor of Paediatric Surgery, London) Prof. A Venter (University of the Free State, Bloemfontein) Dr T Westwood (Red Cross War Memorial Children's Hospital, Cape Town) Prof. D F Wittenberg (University of Pretoria)

3 The new Nelson Mandela Children's Hospital – a white elephant or an essential development for paediatric care in Johannesburg?

J M Pettifor

HEALTH & MEDICAL PUBLISHING GROUP:

Research

EXECUTIVE EDITOR Bridget Farham

5 Identifying medication errors in the neonatal intensive care unit and paediatric wards using a medication error checklist at a tertiaty academic hospital in Gauteng, South Africa

A Truter, N Schellack, J C Meyer

11 Nutritional status of children on the National School Nutrition Programme in Capricorn District, Limpopo Province, South Africa

F Malongane, X G Mbhenyane

16 The promotion of oral health within health-promoting schools in KwaZulu-Natal Province, South Africa

M Reddy, S Singh

21 Serum selenium status of HIV-infected children on care and treatment in Enugu, Nigeria

A C Ubesie, B C Ibe, I J Emodi, K K Iloh

26 Individual v. community-level measures of women's decision-making involvement and child survival in Nigeria

J O Akinyemi, S A Adedini, C O Odimegwu

33 Pubertal breast development in primary school girls in Sokoto, North-Western Nigeria M O Ugege, K I Airede, A Omar, O Pinhas-Hamiel, P K Ibitoye, U Chikani, A Adamu, K O Isezuo, F Jiya-Bello, J A Legbo

38 Short-term and sustained effects of a health system strengthening intervention to improve mortality trends for paediatric severe malnutrition in rural South African hospitals: An interrupted time series design

M Muzigaba, G Kigozi, T Puoane

46 Independent and interactive effects of HIV infection, clinical stage and other comorbidities on survival of children treated for severe malnutrition in rural South Africa: A retrospective multicohort study

M Muzigaba, T Puoane, B Sartorius, D Sanders

54 Parental satisfaction in the traditional system of neonatal intensive care unit services in a public sector hospital in North India

V Sankar, P Batra, M Saroha, J Sadiza

58

CPD Questions ublished by the Health and Medical Publishing Group, P Suite 11, Lonsdale Building, Lonsdale Way Pinelands 7405 apers for publication should be addressed to the Editor, P via the website: www.sajch.org.za Tel: 072 635 9825 E-mail: publishing@hmpg.co.za

Cover: Oyisa, Red Cross War Memorial Children's Hospital Primary School

©Copyright: Health and Medical Publishing Group (Pty) Ltd

CEO AND PUBLISHER Hannah Kikaya

MANAGING EDITORS Ingrid Nye Claudia Naidu TECHNICAL EDITORS Naadia van der Bergh Kirsten Morreira PRODUCTION MANAGER Emma Jane Couzens DTP AND DESIGN Travis Arendse Clinton Griffin CHIEF OPERATING OFFICER Diane Smith | Tel. 012 481 2069 Email: dianes@hmpg.co.za ONLINE SUPPORT Gertrude Fani | Tel. 072 463 2159 Email: publishing@hmpg.co.za FINANCE Tshepiso Mokoena HMPG BOARD OF DIRECTORS Prof. M Lukhele (Chair), Dr M R Abbas, Dr M J Grootboom, Mrs H Kikaya, Prof. E L Mazwai, Dr M Mbokota, Dr G Wolvaardt HEAD OFFICE Block F, Castle Walk Corporate Park, Nossob Street, Erasmuskloof Ext. 3, Pretoria, 0181 EDITORIAL OFFICE Suite 11, Lonsdale Building, Lonsdale Way, Pinelands, 7405 Tel. 021 532 1281 | Cell. 072 635 9825 Email: publishing@hmpg.co.za ISSN 1999-7671

Use of editorial material is subject to the Creative Commons Attribution – Noncommercial Works Licence. http://creativecommons.org/licenses/by-nc/4.0

EDITORIAL

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

The Nelson Mandela Children’s Hospital History

In 2005, Madiba expressed his wish for improved medical care for children. This was taken up by the Nelson Mandela Children’s Fund (NMCF) under the leadership of Mrs Sibongile Mkhabela, who has been central to the project since its inception. After numerous discussions, including those with members of the paediatric academic community in Gauteng, it was agreed that there was a great need for a dedicated children’s hospital that would provide specialised medical and surgical services for children. This was confirmed by detailed feasibility studies.

The Nelson Mandela Children’s Hospital (NMCH) Trust was established in 2009 and, after various sites were considered, it was agreed with the University of the Witwatersrand (Wits) that the hospital would be built on one corner of the Wits Education Campus. An international architectural firm together with a South African (SA) partner were appointed in 2010 and they had many meetings with various stakeholders, including clinicians from the various disciplines who would be most involved with the hospital. It was estimated that ZAR1 bn was needed for the construction and equipping of the hospital and a global awareness and fundraising campaign was initiated in 2012 for the largest civil-society-led capital campaign in SA. When sufficient funds had been raised to justify the initiation of the project, building commenced in March 2014 and, for many of us who passed the building site frequently, the rapidity with which the building took shape was remarkable. A commissioning team was appointed in early 2016 and set about equipping the hospital. By the end of 2016 the hospital was ready to open, and the official opening ceremony was held on 2 December.

Planned services

The National Department of Health has divided tertiary services into three categories: • T1: general specialist services • T2: central referral services usually provided by subspecialists • T3: national referral services provided by one or two centres in the country, sometimes referred to as quaternary services.

The NMCH will concentrate on providing T2 and T3 services primarily for Gauteng and surrounding provinces, and also for some from further afield including Southern African Development Community (SADC) countries. Currently these services in Gauteng are somewhat fragmented as a result of services that developed in previously racially segmented hospitals, and combining them into one centre will be important for developing critical mass in these areas and promoting research. Some public-sector services such as paediatric cardiothoracic surgery and renal transplantation may move in their entirety, but most would provide expanded and hopefully improved services. No child who qualifies for admission would be turned away. The main services planned for the NMCH are: • Cardiology and cardiothoracic surgery: A state-of-the-art cardiac catheterisation suite and cardiothoracic theatre will allow for a major reduction in the current backlog for cardiothoracic surgery, as well as provide for the expansion of non-invasive interventional cardiology and the development of cardiac magnetic resonance imaging.

• Nephrology and renal transplantation: The current shortage of renal haemodialysis services has limited the number of children that can be offered haemodialysis with a view to transplantation.

Expansion of these services will help greatly in this regard and should increase the number of successful transplants, the ultimate goal in the treatment of chronic renal failure.

2

SAJCH

• Oncology: Haematological and other malignancies requiring complex treatment resulting in immunological depression will benefit from the facilities at NMCH, with its specialised cubicles for treating such patients. Surgery for various malignancies with stateof-the-art theatre and intensive care units (ICUs) will also benefit greatly. A bone marrow transplant unit will also be developed.

• Paediatric and neonatal surgery: Complex surgery for both congenital and acquired conditions will be centralised at NMCH, again taking advantage of the theatre and ICU facilities.

• Neurology and neurosurgery: Video telemetry and other specialised neurological services will be developed. Complex neurosurgery will be centralised and epilepsy surgery will be developed.

• Organ transplantation: Solid-organ transplantation will be centralised at the NMCH, once again taking advantage of the theatre and ICU facilities.

Other paediatric subspecialities will be needed to support these services as well as develop and expand services in their specific fields. These include pulmonology, gastroenterology, endocrinology and infectious diseases, which will require additional subspecialists to be trained over time. Other surgical disciplines will also benefit greatly from the theatre and ICU facilities, including orthopaedic surgery, craniofacial surgery, ophthalmology, urology and otolaryngology. Central to all these services will be the major support services and facilities: • Paediatric and neonatal ICU: Public-sector hospitals in Gauteng currently suffer from an acute shortage of both paediatric and neonatal ICU beds, resulting in unacceptable delays in both acute and elective complex surgery. The hospital, when fully functional, will have 24 neonatal and 24 paediatric ICU and high-care beds, which will facilitate the existing surgical services and many of the medical disciplines, as well as providing for some new services such as bone marrow transplantation.

• Operating theatres: Dedicated state-of-the-art theatres for cardiothoracic, transplant , neuro-, and orthopaedic surgery and others will allow further expertise to be developed and, together with ICU facilities, will cut down on unacceptable delays in surgery.

• Anaesthesiology: Expertise that has already been developed at existing hospitals will be concentrated at NMCH. This will be expanded as the surgical services in the various specialties develop.

• Imaging: State-of-the-art magnetic resonance imaging, computed tomography and positron emission tomography-computed tomography scanners, as well as other imaging equipment are already in place to provide services for NMCH as well as helping to reduce waiting times at the other hospitals.

Additional support from the allied medical disciplines, laboratory services and pharmacy have all been provided for.

Staffing

Very few of the services to be provided will involve the transfer of existing staff to NMCH, as the vast majority of the services will be an expansion of existing services or the development of new services.

Training of general and specialised nurses, as well as fellows in medical and surgical disciplines has been in place for several years. However, staffing of the hospital will be an ongoing challenge and must be done carefully so as not to deplete staff at the existing public hospitals.

Conditions of service will be the same as those for the public sector.

Training of both doctors and nurses is ongoing and the interest shown in working at NMCH from those in SA and abroad indicates that, although it may take several years, the hospital can be fully staffed.

MARCH 2017 Vol. 11 No. 1

EDITORIAL Patient selection

Each discipline has already, or will define, which categories of patients will qualify for admission to NMCH, and all patients will be referred through defined channels of referral. Referral back to the referring hospital will take place when appropriate and followup will depend on the services needed. No accident and emergency service will be available, as that of Charlotte Maxeke Johannesburg Academic Hospital is less than 1 km away. Private fee-paying patients, who could make up to 20% of the patients treated, will be admitted provided their condition fits the criteria for admission.

Similarly, those from SADC countries or further afield will be admitted on the same basis once a suitable funding model has been worked out for them.

Training and academic affiliation

Once the NMCH is functional, an application will be made to the Health Professions Council for recognition as a training facility for nurses, registrars, fellows etc. Due to NMCH's proximity, the majority of trainees will probably be affiliated to Wits, but registrars and fellows from the University of Pretoria and Sefako Makgatho University will also rotate through NMCH. Supernumerary registrars and fellows from other countries will be supported. Specialists and subspecialists employed by NMCH will have the option of joint appointment with any of the medical faculties. Links with the University of Johannesburg and other tertiary institutions will also be encouraged.

Funding is the key

The NMCH will be run as an independent facility with a board consisting of members from all the relevant stakeholders. Most of the funding for the hospital will come from the National Treasury and, as required constitutionally, will be channeled through the Gauteng Department of Health, with additional funds from the GDH which will follow some services that move from existing hospitals. Additional funding will come through fee-paying or foreign patients. However, funding for the coming financial year will only allow a very limited service to be provided, consisting of imaging, haemodialysis and cardiac catheterisation, with some ICU back up. The resources that have gone into the physical facility and training of human capital cannot be utilised in the near future until funds allow for sufficient staff to be employed. Every effort should be made by all the relevant stakeholders to ensure that sufficient funding becomes available so that the hospital can be utilised to its full potential.

Peter A Cooper

MB ChB, FCPaed (SA), PhD Emeritus Professor Department of Paediatrics and Child Health, University of the Witwatersrand, Johannesburg, South Africa peter.cooper@wits.ac.za S Afr J Child Health 2017;11(1):2-3. DOI:10.7196/SAJCH.2017.v11i1.1357

The recently launched Nelson Mandela Children’s Hospital in Johannesburg, South Africa.

3

SAJCH

MARCH 2017 Vol. 11 No. 1

EDITORIAL

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

The new Nelson Mandela Children’s Hospital – a white elephant or an essential development for paediatric care in Johannesburg?

In December 2016, the new Nelson Mandela Children’s Hospital (NMCH) was opened by the minister of health, Dr Aaron Motsoaledi, on the University of the Witwatersrand’s (Wits) Education Campus across the road from Charlotte Maxeke Johannesburg Academic Hospital in which one of the three clinical academic Departments of Paediatrics attached to Wits is located. Although it has yet to admit its first patient, the new hospital is impressively equipped with wonderful state-of-the-art facilities for both sick children and their parents (cf. editorial in this issue of SAJCH).

A children’s hospital is not a novel venture for Johannesburg.

Those white paediatricians who trained in Johannesburg before the opening of the new Johannesburg Academic Hospital (now the Charlotte Maxeke Johannesburg Academic Hospital) in 1978 will probably remember with considerable fondness and nostalgia the Transvaal Memorial Hospital for Children (TMH) and its associated Queen Victoria Maternity Hospital just north of the Johannesburg Fort in Hillbrow. This 112-bed hospital was opened in 1923, following a decision in 1919 by the National Council of Women to raise funds to build the hospital to commemorate those South Africans who were killed in World War 1. TMH was thus built and opened some 33 years before the Red Cross War Memorial Children’s Hospital in Cape Town in 1956. Many paediatricians mourned the closure of TMH in 1978/9 and the move of paediatric facilities, both medical and surgical, into the then recently completed Johannesburg Academic Hospital, to become lost in a 1 000-bed general hospital.

There is no doubt that sick children need facilities which are very different from those required by adults. The ambience of the hospital should be different, the facilities should be child friendly and the staff appropriately trained to care for infants, children and adolescents with their very different psychological and physical needs. Why then is there so much discussion around the need for the NMCH among those caring for children within the three clinical paediatric departments in Johannesburg, when it should be clear that providing scarce skills and facilities for sick children can only benefit the paediatric community?

Those who are concerned about the effect that the NMCH will have on paediatric care in Johannesburg raise a number of issues, all of which relate to the possible increase in competition for limited resources available for the continued provision of care for children within the province. In this discussion, it should be emphasised that NMCH will probably have little effect on admission rates and acuity levels of children admitted to the public-sector hospitals as NMCH will be providing care that is currently frequently unavailable or inadequately provided in the public hospitals.

With the closure of a number of nursing colleges over the last 20 years, provincial hospitals are finding it more and more difficult to attract appropriately trained nurses to care for children, especially in areas requiring technical skills such as ICU, surgical theatres and neonatology. Even in the general paediatric wards, there are insufficient staff, especially at night, to ensure appropriate

4

SAJCH

feeding, monitoring and the provision of medication to admitted children. This nursing shortage within the public hospital sector is aggravated by an increasing demand for skilled staff from private hospitals. It is not just the availability of nursing staff that will likely be affected by the opening of NMHC, but also specialised allied medical disciplines such as radiographers and paediatric and surgical subspecialists, who will be attracted from the currently under-resourced provincial hospitals to NMCH with its state-ofthe-art equipment and facilities.

A further concern that has been raised is the challenge of funding the operational costs of the hospital. It is one thing to raise funds through donations to build and equip the hospital, but it is quite another to consistently raise the necessary funds to keep the ship afloat. It has been suggested that with its developing reputation, NMCH could attract sufficient numbers of privately funded patients to subsidise the treatment of public-sector children requiring care in a quaternary hospital. Further, the national minister of health has made it clear that the hospital is a national resource and therefore will be funded from the national purse. However, as has very recently been highlighted by the tragedy associated with the closure of psychiatric facilities in Gauteng, provincial funds available to care appropriately for the province’s steadily growing population (now estimated at 13.4 million) are insufficient. The concern of many paediatricians is the effect that the funding of NMHC operating costs from the provincial budget will have on the already tight budget allocated to the three academic teaching hospitals. Will the opening of a highly specialised quaternary hospital reduce the province’s ability to care for its sick children requiring less technically advanced facilities in secondary and tertiary hospitals?

It is important that we applaud the Nelson Mandela Children’s Fund in having the foresight to champion the building of a quaternary care children’s hospital in the most populous and wealthiest province in South Africa. I have no doubt that over the coming years it will define its own role in the province’s attempt to provide an holistic approach to the care of its children. In the meantime, we should develop services carefully and slowly in the NMCH so that the services currently provided in the secondary and tertiary hospitals are not destabilised. Further, the provision of a state-of-the-art children’s hospital does not relieve the state of its responsibility to upgrade the staffing and facilities at key hospitals such as Chris Hani Baragwanath Academic Hospital, which should not be allowed to physically deteriorate even further.

John M Pettifor

MB BCh, FCPaed (SA), PhD (Med), MASSAf Editor, South African Journal of Child Health john.pettifor@wits.ac.za S Afr J Child Health 2017;11(1):4. DOI:10.7196/SAJCH.2017.v11i1.1356

MARCH 2017 Vol. 11 No. 1

RESEARCH

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

Identifying medication errors in the neonatal intensive care unit and paediatric wards using a medication error checklist at a tertiary academic hospital in Gauteng, South Africa A Truter, BPharm, MPharm; N Schellack, BCur, BPharm, PhD (Pharmacy); J C Meyer, BPharm, MSc (Med), PhD (Pharmacy) Department of Pharmacy, Faculty of Health Sciences, School of Health Care Sciences, Sefako Makgatho Health Sciences University, Pretoria, South Africa Corresponding author: A Truter (archele.truter@live.com) Background. Paediatric patients are particularly prone to medication errors as they are classified as the most fragile population in a hospital setting. Paediatric medication errors in the South African healthcare setting are comparatively understudied.

Objectives. To determine the incidence of medication errors in neonatal and paediatric inpatients, investigate the origin of medication errors that occurred and describe and categorise the types of medication errors made in both the neonatal intensive care unit (NICU) and paediatric wards.

Methods. The study followed a prospective, quantitative design with a descriptive approach. A prospective record review of inpatients’ medication charts was undertaken to determine what was prescribed by the physician, dispensed by the pharmacy and administered by the nurses. The researcher also directly observed the preparation and administration techniques as performed by the nurses. A medication error checklist was used to collect the data.

Results. A total of 663 medication errors were detected in 227 patients over the study period of 16 weeks, of which 177 (78%) patients had one or more error(s). There were 338 (51%) administration errors and 309 (47%) prescribing errors. Incorrect dosing was the most frequent type of error (34%), followed by omission of medication (18.5%) and medication given at the incorrect time (12%). The causes of these medication errors were mostly due to miscalculation (26%), failure to monitor (15%) and procedures not followed (15%). Anti-infectives (43%) and analgesics (25%) had the most errors. In 118 (67%) patients the errors resulted in no harm to the patient, whereas in 59 (33%) patients the medication error resulted in some level of harm.

Conclusion. The incidence of medication errors in the NICU and paediatric wards at the teaching hospital was higher than values reported elsewhere globally. Most errors occur during prescribing and administration of medication. Dosing errors are a common problem in paediatrics. Therefore, a formalised system to record these errors should be introduced alongside regular discussions on preventive measures among the multidisciplinary team.

S Afr J Child Health 2017;11(1):5-10. DOI:10.7196/SAJCH.2017.v11i1.1101

Medication errors in paediatrics are relatively understudied in South Africa (SA). The National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) defines a medication error as ‘any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of the health care professional, patient, or consumer. Such events may be related to professional practice, healthcare products, procedures, and systems, including prescribing, order communication, product labelling, packaging, and nomenclature, compounding, dispensing, distribution, administration, education, monitoring, and use’.[1] Most medications used in paediatrics are used off-label and hence adult dosage forms are administered, which may increase opportunities for medication errors with subsequent risks for the patient.[2] The most error-prone step in the medication process is prescribing by the prescriber.[3] Dosing errors are the leading cause of medication errors, especially in the paediatric population and more so in the intensive care unit (ICU).[3] Paediatric patients require individualised prescribing of medication based on their age and body weight, together with the interpretation of the therapy outcome.[4] There are five stages of the medication process where medication errors can occur, called the origin of medication errors,[5,6] i.e.: prescribing of medication by a legitimate prescriber; transcribing of the order documentation by someone other than the prescriber for ordering and 5

SAJCH

processing; dispensing, whereby a pharmacist assesses a medication order and releases the product for use by another healthcare professional; administration, where the medication is administered to the patient, including administration of the correct medication to the correct patient at the prescribed time and labelling the current intravenous medication that the patient is receiving; and monitoring to evaluate the patient’s response to medication and record the findings.[5,6] The following types of medication errors are identified and described by various studies: incorrect medicine, incorrect dose, incorrect preparation, expired product, incorrect time, incorrect route, unauthorised medication, omission, wrong patient, mislabelling, incorrect dispensing, incorrect duration of treatment, extra dose, deteriorated medication and contraindication.[5,7-10] The classes of drugs mostly involved in paediatric medication errors include anti-infectives, sedatives, analgesics, bronchodilators and cardiovascular drugs.[11-13] The causes of medication errors in general clinical practice can be divided into human and system errors. Human errors consist of performance deficit, procedure or protocol not followed, miscommunication, inaccurate or omitted transcription, improper documentation, knowledge deficit, miscalculations, missing or misplaced zero and decimal points, use of non-standard abbreviations, lack of patient information and lack of patient understanding of their

MARCH 2017 Vol. 11 No. 1

RESEARCH therapy.[5] Human errors are compounded when wards are overcrowded, leading to an increase in workload, generation of more stress, tiredness and sleep deprivation in the healthcare professional, which facilitates the occurrence of more medication errors.[4,5] Factors that contribute to system errors include medications with similar names, complex or poorly designed electronic patient record technology, access to medication by non-pharmacy personnel, drug distribution system errors, computer entry error, lack of system safeguards, and workplace environmental problems that include the daily occupancy rate and the workload of the healthcare professionals.[5] The severity of medication errors can be categorised (A - I) based on the effect they have on the patient, as constructed by the NCC MERP.[1,6,14] Currently, in SA, there is no standardised medication error database for the identification and quantification of medication errors in hospitals. However, ongoing safety and effectiveness of medicines is ensured by voluntary reporting systems that form part of pharmacovigilance programmes.[15] These programmes aim to detect, assess, manage and prevent drug-related problems.[15,16] The importance of the presence of a clinical pharmacist in a paediatric unit to monitor drug treatment and prevent medication errors is well established in some countries.[7,9,17] Clinical pharmacists monitoring medication orders might prevent more than half (58%) of all errors, including 72% of potentially harmful errors.

Furthermore, they may improve doctor-pharmacist communication, preventing 47% of all errors.[17] The clinical pharmacist should be involved in double-checking all drug administrations to reduce medication errors. For preventive measures to be successful against medication errors there needs to be collaboration between all healthcare professionals.[13] This study emphasises that it is the responsibility of the healthcare provider to ensure that patient safety incidents are instantly identified and managed to minimise patient harm and suffering, that errors are routinely investigated and managed to prevent repetition and to learn from errors.[18] To this end the purpose of the study was to determine the incidence of medication errors in neonatal and paediatric inpatients, to investigate the origin of medication errors, and to describe and categorise the types of medication errors made in the neonatal intensive care unit (NICU) and paediatric wards in a tertiary academic hospital.

dosage form and on-time dispensing when a prescription was sent to the pharmacy. Direct observation occurred 5 - 7 hours per day, 5 days a week, for the 16-week study period.

A medication error checklist was designed according to recommendations from several studies.[5-10,14] The definitions of the various types of errors were adopted from previous studies and are defined in Table 1.[5,8,11,13,14] The medication error checklist was validated in a pilot study of 10 patients and minor amendments were made. The checklist was subsequently used to capture the following information, as outlined by the objectives: the origin of medication errors (e.g. prescription or administration), the type of medication errors that occurred (e.g. incorrect dose or inadequate preparation) and the cause of the medication errors (e.g. miscalculation or inexperienced staff), after Table 1. Classification of types of medication errors Type of error

Definition of error

Inadequate preparation of medication

When medication was prepared or manipulated incorrectly. This includes incorrect method of reconstitution or dilution, not shaking the suspension thoroughly and crushing of specially coated tablets.

Incorrect dose

Dose that was prescribed or administered was >10% above or below the correct dose based on the patient's weight.

Incorrect duration

Medication administered for a longer period of time than was prescribed, or prescribed medication that was not discontinued when indicated.

Incorrect frequency

Medication administered at incorrect intervals (e.g. 8-hourly instead of 6-hourly).

Incorrect medication

Administration of medication that was not prescribed, misread prescription, or medication administered to the wrong patient.

Incorrect time

There was >1 h difference between the scheduled time and time of administration.

Mislabelling

When reconstituted medication was kept in storage and had no label indicating the time of reconstitution and volume of diluent used. Infusion not labelled with the name or dose of medication that was being administered.

Omission

Failure to administer a prescribed medication, or medication that was being administered without noting that it had been dispensed.

Prescribing error (e.g. no route)

Elements of good prescribing practice were observed and each medication prescribed was evaluated for compliance with pharmacy legislation as stipulated in Good Pharmacy Practice, i.e. the correct name, dosage, units, route, frequency and duration of treatment.

Methods

Study design, study site and population

This study followed an observational, quantitative, descriptive design that was done prospectively. It was conducted in four paediatric wards over a 16-week period at a tertiary academic hospital, which has 28 clinical departments, and is one of four academic institutions in Gauteng Province. It provides a service to the surrounding populations of ~1 700 000 people. The hospital also receives referrals from Limpopo, North West and Mpumalanga provinces. In addition, this facility receives referrals from Southern African Development Community (SADC) countries, other tertiary academic hospitals, local specialists and general practitioners. The hospital has 1 650 active beds, 22 approved ICU beds, 60 high-care beds and 17 theatres. The four paediatric wards under study consisted of a 40-bed orthopaedics ward, a 40-bed surgical ward, a 20-bed oncology ward and a 55-bed NICU.

Data collection and data collection instruments

Neonatal and paediatric inpatient medical files for background of the patient’s disease and current condition and the patients’ medication charts were reviewed to evaluate what was prescribed by the physicians and documented as administered by the nurses.

The method of how and when nurses prepared (reconstituted) and administered medication to the patients was directly observed by the researcher. The dispensing process was evaluated for the correct drug, 6

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH middle childhood (39%, n=71), early adolescence (7%, n=12), and one late adolescence. The median age of the patients was 5 years (interquartile range (IQR) 2 - 8 years, range 1 month - 19 years) and the neonates’ mean age was 33 (standard deviation (SD) 3.89; range 25 - 40) weeks.

Table 2. Categorising medication errors Error and category

Harm and description

No error

No harm

Category A Error

Circumstances or events that have the capacity to cause harm

Patient diagnoses and medicines prescribed

No harm

Category B

Error occurred but did not reach the patient

Category C

Error occurred and reached the patient but did not cause harm

Category D

Error occurred and reached the patient – required monitoring to confirm that it resulted in no harm to the patient and/or required intervention to preclude harm

Error

Harm

Category E

Error occurred – need for treatment or intervention – temporary harm to patient

Category F

Error occurred – need for prolonged hospitalisation – temporary harm to patient

In total, 663 medication errors were detected among 227 patient files studied over 16 weeks, with one or more errors in 177 (78%) patients and no errors in 50 (22%) patients. The incidence of medication errors in the four paediatric wards was 2.9 per patient, per admission, over the 16-week observation period.

Death

A total of 715 medications were prescribed and studied over the 16-week study period. There were 309 (47%) prescribing errors documented, with an overall prescribing error rate of 43% (95% confidence interval (CI) 39.6 - 46.9). The nurses were responsible for 338 administration errors (51%), including labelling errors, giving an overall administration error rate of 47.3% (95% CI 43.6 - 50.9). The pharmacy was responsible for 16 dispensing errors (2%).

Error occured that resulted in patient death

Site of medication error

Error occurred that contributed to permanent patient harm

Category H

Error occurred that required intervention necessary to sustain life

Category I

Incidence and nature of medication errors

Origin of medication errors

Category G

Error

The main causes of hospitalisation were respiratory distress syndrome, neonatal sepsis, babies born to HIV-positive mothers, B-cell acute lymphoblastic leukaemia, neonatal jaundice, adenotonsillectomy, tuberculosis and pneumonia. The most commonly prescribed medications were paracetamol, tilidine, amikacin, allopurinol, amoxicillin-clavulanic acid, benzylpenicillin, cloxacillin, ibuprofen, metronidazole, nevirapine and piperacillin-tazobactam.

which it was categorised (Table 2) according to the NCC MERP for data analyses.[1]

Data analysis

The data from the medication error checklists were imported into Statistical Analysis System (SAS) release 9.3 (USA) for statistical analysis. Descriptive statistics were used to analyse and summarise data to obtain the frequency of occurrence of medication errors.

Incidence of medication errors was calculated as a percentage with 95% confidence intervals (CIs). The χ2 test and p-value were calculated for homogeneity between the four wards, and the relative risk to an error was evaluated.

Ethical considerations

Ethical approval was obtained from the university associated with the hospital. The Medunsa Research Ethics Committee approved the study (MREC/H/225/2014: PG). Permission to perform this study at the tertiary academic hospital was obtained from the hospital’s chief executive officer. All data were collected anonymously. Patient confidentiality was maintained by the allocation of a unique study number to each participant.

Results

Demographics

During the study period, 227 patient medication charts (47 neonates and 180 paediatric patients) were evaluated for medication errors. A total of 91 patients were female (40%) and 136 patients were male (60%). The ages of the patients were categorised according to paediatric terminology developed by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD).[19] Therefore, the study consisted of the following percentages: infants (8%, n=15), toddlers (17%, n=31), early childhood (28%, n=50), 7

SAJCH

Medication errors in the four paediatric wards were separately analysed (Table 3). The incidence of medication errors was tested using the χ2 test of homogeneity. It was used to determine whether frequency counts were distributed identically across the different paediatric wards. The results showed no homogeneity, meaning that the four wards differed in respect of the percentage of errors (p<0.0001). The relative risk for medication errors was calculated using the number of medications observed and the incidence of errors per ward. The orthopaedic ward had the highest risk for medication errors compared with the other three wards.

Types of medication errors

The types of medication errors observed are shown in Fig. 1. Incorrect dose was the most frequent type of error that occurred, giving an overall incorrect dosing error rate of 34% (95% CI 30.5 - 37.7) in 663 errors detected. This was followed by omission of medication 19% (95% CI 15.5 - 21.4), medication given at the incorrect time 12% (95% CI 9.2 - 14.1), incorrect frequency of administration 7.9% (95% CI 5.8 - 9.9) and inadequate preparation of medication 7.8% (95% CI 5.7 - 9.7). Other types of errors found were mislabelling (6.7%), incorrect medication (4%), no route of administration prescribed (3.2%), incorrect duration of therapy (1.7%) and no frequency prescribed (1%). Examples of medication errors observed, and possible causes or reasons are shown in Table 4.

Groups of medications with errors

The most commonly prescribed medications for the four wards included anti-infectives and analgesics. The number of errors associated against the number of medication orders with these two classes are displayed in Fig. 2. Anti-infectives with the most errors, accountable for 284 (43%) of the 663 errors, were: amoxicillinclavulanic acid, metronidazole, cloxacillin, cefuroxime, amikacin, benzylpenicillin, imipenem, piperacillin-tazobactam and gentamicin.

MARCH 2017 Vol. 11 No. 1

RESEARCH Table 3. Frequency of errors in the four paediatric wards Ward

Medications observed, n

Errors, n

Ď&#x2021;2

p-value

Relative risk

Orthopaedics

87

113

44.68

<0.0001*

1.299

Surgery

251

259

1.032

Oncology

241

185

0.768

NICU

136

106

0.779

Types medication T ypes oof fM edication Eerrors rrors

*Highly statistically significant.

Incorrect dose Omission Incorrect time Incorrect frequency Inadequate preparation Mislabelling Incorrect medication No route Incorrect duration No frequency 0

5

10

15

20

25

30

35

40

Incidence rate (%) with 95% confidence interval Fig. 1. Types of medication errors across the four paediatric wards.

No. ofOrders/errors, orders/ no. of enrrors

Medication orders Medication errors

284

300 250 200

214

195

166

150 100 50 0

Anti-infectives

Analgesics

Classes of medications with the most errors Fig. 2. Classes of medications with the number of times it was ordered and the number of medication errors observed.

Analgesics accountable for 166 (25%) of the errors were: paracetamol, tilidine and ibuprofen. The remaining 213 (32%) errors consisted of different medications from multiple classes. Paracetamol had the most medication errors, with 106 errors out of the total medication errors identified.

Categorising medication errors

Medication errors were classified into six categories (Aâ&#x20AC;&#x201A; - F) in the 177 patients with medication errors, according to severity (Tableâ&#x20AC;&#x201A; 2).[1] Just more than 60% (118) of the patients who had medication errors did not suffer any harm (categories A - C), whereas

59 (33%) patients had some level of harm (categories D - F). No permanent patient harm or deaths were recorded owing to a medication error.

Discussion

In this study, the largest proportion of patients was toddlers and children in their early and middle childhood. The majority of the neonates included in this study were premature. Patients were mostly hospitalised owing to respiratory distress syndrome (in neonates), adenotonsillectomy and tuberculosis (in paediatrics). Anti-infectives and analgesics were the most prescribed classes 8

SAJCH

MARCH 2017 Vol. 11 No. 1

of medication and correlates with similar studies that also identified these agents as commonly prescribed in paediatrics.[7,11,13,14] The majority of studies conducted in paediatrics reporting medication errors declared medication error frequencies of between 11% and 54%.[2,4,7,8,14] This is markedly lower than the frequency of medication errors (78%) reported in this study.

Although the frequency of medication errors in this study is high, a number of factors may have led to biased findings, which actually resulted in an overall conservative estimate of errors reported.

Firstly, although the doctors and nurses were unaware of the reasons for the observation, they were aware that they were being observed and might have changed their behaviour. Secondly, the research team tried to avoid as many interventions as they could to prevent observer-induced bias. However, in terms of the ethical principle of nonmaleficence, it would be unethical to allow medication administration errors to occur for the purpose of observation when it can cause direct, severe or irreversible harm.

The researchers therefore had to intervene and prevent some errors from happening in cases where an observed error could cause harm to the patient. Examples of interventions by the researchers included the doctor being asked to change the high dose of 1 g paracetamol prescribed for a 2-yearold, to the recommended 180 mg, and the nurse being stopped before administering 12 drops of tilidine to a 1-year-old instead of two drops as prescribed. Lastly, the lack of observation of medication reconstitution and medication administration by nurses, 24 hours per day, 7 days per week, probably reduced the number of errors detected.

In the evaluation of the origin of medication errors, administration of medicines by nurses caused just over half (51%) of all detected errors, with prescribing errors by doctors causing 47% of the medication errors. There is some variation in the literature with similar results reported by Khalili et al.[7] and lower percentages in various other studies that ranged between 11.7% and 19% administration errors, and 11% and 34% prescribing errors.[2,8,13,14] Dispensing errors amounted to 2% of the total errors, which may be due to the fact that most commonly prescribed medications are kept in the ward as ward stock.

The most prevalent type of error identified was incorrect dosing (34%), followed by omission (19%) and medication given at the incorrect time (12%). Possible reasons for omission and incorrect time could include nursing staff being tasked with a number of non-nursing related duties, e.g. cleaning, admixture of medications and insertion of IV lines. This is consistent with similar studies that have also reported dosing errors.[3,4,6,7,14]

RESEARCH Table 4. Examples of medication errors observed, possible causes and reasons Type of error

Frequency of error (n=663)

Incorrect dose

226

Possible causes or reasons (n)

Description of examples

Cause or reason, % (95% CI)

1 g paracetamol prescribed instead of 180 mg (researcher intervened)

Miscalculation (175), e.g. dose 26.4 (23.0 - 29.8) prescribed was incorrectly calculated based on patient's weight or protocol

5 mL paracetamol syrup given as 500 mg tablet

Dose to be administered was miscalculated due to limited strength available

150 mg metronidazole prescribed as 60 mg 2 drops tilidine given as 12 drops for 8-month-old (researcher intervened) Omission

Incorrect time

122

77

Common medications not given: paracetamol, tilidine and ibuprofen

Failure to monitor (97), e.g. prescription not checked to see what must be administered

The nurse forgot to order antibiotics from the phar-macy that is not part of ward stock, patients missed doses

Medication out of stock, no communication with doctor to prescribe an alternative

Drip not inserted, patient failed to receive intravenous (IV) medication

No documentation (64) Rule of thumb: if it was not signed it was not administered

Prescription sent later to the pharmacy, received antibiotics at the incorrect time â&#x20AC;&#x201C; administered at the wrong time

Increased workload (68), e.g. one nurse for 10 patients or one nurse for four babies

Medication given at 14h00 instead of 10h00

Responsible nurse too busy with administration and other duties

14.7 (11.9 - 17.3)

9.7 (7.4 - 11.9) 10.3 (8.0 - 12.6)

Not enough nursing staff in the ward One nurse for four babies â&#x20AC;&#x201C; (understaffed) increased workload (drips to be inserted, admixtures, nappy changes, cleaning, updating files, etc.)

Incorrect frequency

52

Cloxacillin given 8-hourly instead of 6-hourly, as prescribed

Procedure not followed (96), e.g. prescription not double-checked for frequency prescribed

14.5 (11.8 - 17.2)

Imipenem prescribed as 6-hourly for Antibiotic policy not followed all neonates instead of 12-hourly or when prescribing antibiotics in the 8-hourly when <21 days of age neonate Inadequate preparation/ incorrect technique

51

Oral amoxicillin-clavulanic acid mixed with tap water (sterile water available)

Inexperienced staff (67) e.g. nursing students and new nursing staff

Used one needle to reconstitute all IV medication for the ward

Does not consult the package insert or other reference sources ofinformation

10.2 (7.8 - 12.4)

Tired after working multiple 12hour shifts

Miller et al.[20] had remarkably high incidences in both omission of medication (42%) and the wrong time of administration (50%), corresponding to Chua et al.[13] and Hicks et al.[6] Inadequate knowledge, insufficient training and increased workload has been listed as major causes of medication errors.[4,6,9,13] Similar findings were identified as the most prominent causes of medication errors in this study, miscalculation of dosages (26%), failure to monitor therapy (15%) and procedures (e.g. antibiotic policy) not followed (15%).

The paediatric orthopaedic ward had the highest risk for medication errors, even though they had the lowest number of medications prescribed and administered. Reasons for this included the high number of analgesics prescribed for postoperative pain relief. Tilidine is frequently prescribed for pain relief in this ward. Medication errors 9

SAJCH

observed included omission of doses and incorrect dosing of the agent. Tilidine is administered using an oral medicine dropper, with the possibility of drops being administered inconsistently (i.e. too fast or too slow).[21] Small differences in the administration of this drug may result in either a lack of optimal pain relief (underdose) or harm (overdose).[21] In the class of analgesics, paracetamol, tilidine and ibuprofen were identified as agents that caused the most errors. The literature pertaining to paracetamol is contradictory, with one study listing only five doses of paracetamol with errors[13] and other studies listing many errors.[7,9,14] Slightly less than half (43%) of all medication errors were identified in the class of anti-infectives and is equal to studies that found numerous errors in antibiotics reported in paediatrics.[3,9-11,13]

MARCH 2017 Vol. 11 No. 1

RESEARCH The greatest majority (87%) of these medication errors that reached the patients did not cause harm according to the categories used by the NCC MERP.[1] Alternative studies showed similar results, except where circumstances or events that had the capacity to cause harm were not included.[6,14]

Study limitations and recommendations

Only those errors observed by the researchers were used for analysis. Therefore several errors may have gone undetected. Other study limitations included validation of the tool used to determine error. Our tool was developed using reputable references.

Observation per se may also affect practice and result in a decrease of medication error rate. Observer bias could not be ruled out since observational research requires a lot of resources and referencing, and is time consuming.

Using a medication error checklist is fundamental to improve patient safety. It is therefore crucial for healthcare providers to identify weak points in the healthcare system regarding medication prescribing, administering and dispensing in paediatrics. Measures that may be implemented include a clinical pharmacist actively working with other healthcare providers in paediatric wards.

This can improve the patient's healthcare outcomes and has been described in several studies. Reviewing patient medical charts and reporting medication therapy problems to the head of the medical team on a daily basis are part of the clinical pharmacist’s daily responsibilities. [2,7,9,12,13,17] Recommendations to reduce medication errors in the future include the implementation of training for paediatric nurses, doctors and pharmacists on aspects identified as part of the study, for e.g. steps to correct prescribing, how to reconstitute medication, and how to calculate dosages. Since medication preparations and administrations are the last line of defence against medication errors, an electronic medication error platform should be implemented to record and report medication errors.

Conclusion

This study showed that the frequency of medication errors in the paediatric ward of the teaching hospital was higher than that reported in similar studies elsewhere. Administration followed by prescription are the most error-prone steps in the paediatric medication process. Dosing errors are a common problem in paediatrics and, therefore, a formalised system to monitor errors should be introduced. Regular discussions on preventative measures among the multidisciplinary team should be initiated to further reduce the frequency of the errors. This study serves to create awareness and interest concerning medication safety in the paediatric population of SA.

10

SAJCH

1. National Coordinating Council for Medication Error Reporting and Prevention.

NCC MERP, 2014. http://www.nccmerp.org/ (accessed 28 January 2014).

2. Ghaleb MA, Barber N, Franklin BD, Wong ICK. The incidence and nature of prescribing and medication administration errors in paediatric inpatients. BMJ 2010;95(2):113-118. http://dx.doi.org/10.1136/adc.2009.158485 3. Jain S, Basu S, Parmar VR. Medication errors in neonates admitted in intensive care unit and emergency department. Indian J Med Sci 2009;63(4):145-151. http://dx.doi.org/10.4103/0019-5359.50763 4. Lerner RB, de Carvalho M, Vieira AA, Lopes JM, Moreira MA. Medication errors in a neonatal intensive care unit. Jornal de Pediatria 2008;84(4):166-170. http://dx.doi.org/10.1590/S0021-75572008000200013 5. Jhanjee A, Bhatia MS, Srivastava S. Medication errors in clinical practice. Delhi Psychiatry J 2011;14(2):205-210.

6. Hicks RW, Becker SC, Krenzicheck D, Beyea SC. Mediation errors in the PACU: A secondary analysis of MEDMARX findings. J PeriAnesth Nurs 2004;19(1):1828. http://dx.doi.org/10.1016/j.jopan.2003.11.007 7. Khalili H, Farsaei S, Rezaee H, Dashti-Khavidaki S. Role of clinical pharmacists’ interventions in detection and prevention of medication errors in a medical ward. Int J of Clin Pharm Pharm Care 2011;33(2):281-284. http://dx.doi. org/10.1007/s11096-011-9494-1 8. Otero P, Leyton A, Mariani G, Cernadas JMC. Medication errors in pediatric inpatients: Prevalence and results of a prevention programme. J Am Acad Pediatr 2008;122(3):737-743. http://dx.doi.org/10.1542/peds.2008-0014 9. Wong ICK, Wong LYL, Cranswick NE. Minimising medication errors in children. Arch Dis Child 2008;94(2):161-164. http://dx.doi.org/10.1136/ adc.2007.116442 10. Ross LM, Wallace J, Paton JY. Medication errors in a paediatric teaching hospital in the UK: Five years operational experience. BMJ 2000;83(6):492-497. http://dx.doi.org/10.1136/adc.83.6.492 11. Feleke Y, Girma B. Medication administration errors involving paediatric inpatients in a hospital in Ethiopia. Trop J Pharm Res 2010;9(4):401-407. http:// dx.doi.org/10.4314/tjpr.v9i4.58942 12. Clifton-Koeppel R. What nurses can do right now to reduce medication errors in the neonatal intensive care unit. Newborn Infant Nurs Rev 2008;8(2):72-82. http://dx.doi.org/10.1053/j.nainr.2008.03.008 13. Chua SS, Chua HM, Omar A. Drug administration errors in paediatric wards: A direct observation approach. Eur J Pediatr 2009;169(5):603-611. http://dx.doi.org/10.1007/s00431-009-1084-z 14. Martines-Anton A, Sanchez IJ, Casanueva L. Impact of an intervention to reduce prescribing errors in a pediatric intensive care unit. Intensive Care Med 2012;38(9):1532-1538. http://dx.doi.org/10.1007/s00134-012-2609-x 15. Mehta U, Dheda, M, Steel G, et al. Strengthening pharmacovigilance in South Africa. S Afr Med J 2014;104(2):104-106. http://dx.doi.org/10.7196/SAMJ.7517 16. Maigetter K, Pollock AM, Kadam A, Ward K, Weiss, MG. Pharmacovigilance in India, Uganda and South Africa with reference to WHO’s minimum requirements. Int J Health Pol Manag 2015;4(5):295-305. http://dx.doi. org/10.15171/ijhpm.2015.55 17. Simpson JH, Lynch R, Grant J, Alroomi L. Reducing medication errors in the neonatal intensive care unit. BMJ 2004;89(6):480-482. http://dx.doi. org/10.1136/adc.2003.044438 18. National Department of Health (NDoH). National Core Standards for Health Establishments in South Africa. Tshwane: NDoH, 2011. http://pmgassets.s3-website-eu-west-1.amazonaws.com/docs/120215abridge_0.pdf (accessed 10 October 2015).

19. Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD). Pediatric Terminology: Current efforts. 2015. https:// www.nichd.nih.gov (accessed 20 November 2015).

20. Miller MR, Robinson KA, Lubomski LH, Rinke ML, Pronovost PJ. Medication errors in paediatric care: A systemic review of epidemiology and an evaluation of evidence supporting reduction strategy recommendations. Qual Saf Health Care J 2007;16(2):116-126. http://dx.doi.org/10.1136/qshc.2006.019950 21. Bauters T, Claus B, Willems E, et al. What’s in a drop? Optimizing strategies for administration of drugs in paediatrics. Int J Clin Pharm 2012;34(5):679-681. http://dx.doi.org/10.1007/s11096-012-9670-y

MARCH 2017 Vol. 11 No. 1

RESEARCH

Nutritional status of children on the National School Nutrition Programme in Capricorn District, Limpopo Province, South Africa F Malongane,1 DT (HPCSA), MSc; X G Mbhenyane,2 DT (HPCSA), PhD Department of Life and Consumer Sciences, College of Agriculture and Environmental Sciences, University of South Africa, Pretoria, South Africa 2 Department of Interdisciplinary Health Sciences, Faculty of Medicine and Health Sciences, Stellenbosch University and Tygerberg Academic Hospital, Cape Town, South Africa 1

Corresponding author: F Malongane (malonf@unisa.ac.za)

Background. School feeding programmes are intended to alleviate short-term hunger, improve nutrition and cognition of children, and provide incomes to families.

Objectives. To assess the nutritional status of children receiving meals provided by the National School Nutrition Programme (NSNP) in Capricorn Municipality, Limpopo Province, South Africa.

Methods. The setting was 18 randomly selected schools on the NSNP in Capricorn District. The total sample comprised 602 randomly selected schoolchildren from grades 4 to 7, aged 10 (26.6%), 11 (35.4%) and 12 (35.4%). Socioeconomic characteristics, anthropometric measurements, dietary patterns and school attendance were determined. Children were interviewed to assess their nutritional status using a validated questionnaire. Descriptive statistics such as means, standard deviations (SDs) and ranges were used for socioeconomic parameters and dietary patterns, and z-scores for anthropometric data.

Results. The results showed that boys (9.5%) and girls (7.8% ) were underweight. The prevalence of stunting in the sample was 11.3% for boys and 7.4% for girls, whereas boys (3.6%) and girls (4.2%) were wasted, with a z-score of â&#x20AC;&#x201C;2 SD. School attendance was good.

Conclusion. The nutritional status of most subjects in the study was within the acceptable range as indicated by the assessment of growth using anthropometric measurements.

S Afr J Child Health 2017;11(1):11-15. DOI:10.7196/SAJCH.2017.v11i1.1124

School feeding programmes have been implemented in both developed and developing countries, where they are often implemented by national government organisations or nongovernmental organisations (NGOs). The largest provider in developing countries is the United Nations with its World Food Programme, which was operational in up to 78 countries in 2006. [1] Numerous other agencies and NGOs operate school feeding programmes at the national, regional and local levels.[1] The National School Nutrition Programme (NSNP), commonly known as the school feeding scheme, was introduced in South Africa (SA) by the government in 1994 to alleviate hunger and contribute to learning in schools by providing learners with nutritious meals.[2] There is evidence that education and learning depend on good nutrition. Studies done in Honduras showed that the academic performance and mental ability of learners with good nutritional status were significantly higher than those of learners with poor nutritional status, irrespective of the level of family income, school quality and teacher ability.[3] School feeding alleviates short-term hunger, thus enhancing active learning capacity, school attendance and punctuality among needy and hungry children. In Jamaica, providing breakfast to primary school learners significantly increased attendance and arithmetic scores.[4] A similar study in the USA showed the benefits of providing breakfast to disadvantaged primary school learners. This programme, which targeted children coming from low-income families, improved their test scores significantly and improved school attendance.[5] In SA, the NSNP is designed to identify and reach areas where poverty is most extreme, with the aim of providing one meal or snack a day by 10h00 through approved menu options.[6] The NSNP, as implemented in Limpopo Province, follows a warm menu which 11

SAJCH

includes beans, soya, samp and vegetables as well as bread and peanut butter. The minimum number of days for feeding was reported to be 156 out of 197 school days in a year.[6] The Limpopo Province initially aimed to provide state-funded primary schools with NSNP meals for grade R to grade 7 learners, with priorities given to those situated in rural areas, farms and informal settlements.[6] The programme was extended to secondary schools in 2009 and has progressively covered quintiles 1 - 3.[2] Limpopo is known to be one of the poorest provinces in SA, with 89.0% of the provincial population living in rural areas and 11.0% in urban/peri-urban areas, with a very high rate (38.9%) of unemployment.[7] There are more than 2 747 primary schools in the province, most of which are situated in rural and peri-urban areas.[7] The objective of this study was to assess the socioeconomic characteristics, anthropometric status, school attendance and dietary patterns of children receiving NSNP meals in Capricorn District Municipality of Limpopo.

Methods

A descriptive, cross-sectional survey design was used to obtain data on selected parameters of primary schoolchildren on the NSNP in Capricorn District, Limpopo Province, SA. Prior to the collection of data, written permission was obtained from the Limpopo Department of Education and local education authorities. Parents were requested to give written consent after an information leaflet had been sent to them. Ethical clearance was given by the University of Venda research ethics committee. The survey was conducted in 18 (10.7%) primary schools randomly selected from a total of 169 in the district. All children aged 9 - 13 years in grades 4 - 7 were recruited, and those whose parents consented to participation were

MARCH 2017 Vol. 11 No. 1

RESEARCH included in the study. The volunteers responsible for the preparation of school meals were consulted to provide information on the NSNP meals. Two volunteers per school were randomly selected on the day of data collection. Data collection was conducted from May 2007 to March 2008.

Variables measured included socioeconomic factors, anthropo­metric status, school attendance and usual dietary patterns. Socioeconomic factors affecting children’s dietary patterns were parents’ employment status, receipt of pension or children’s grants and number of household members. These data were collected by the researcher and two trained field workers, who were trained nutritionists, using a researcheradministered standardised question­ naire. The language used for interviews was Sepedi. School attendance refers to the number of days attended v. the number of school days. Anthropometric status for the purpose of this study included height and weight, from which heightfor-age (HAZ), weight-for-age (WAZ), weight-for-height (WHZ) and body mass index (BMI) z-scores were calculated. Height was recorded without shoes using a measuring board and read to the nearest 0.5 cm. The subjects were weighed using a calibrated electronic scale and the weight was recorded to the nearest 0.2 kg. All measurements were done by a trained 3rd-year nutrition student of the University of Venda and were monitored and checked by the researcher.

Usual dietary patterns and food from the NSNP meal were measured using selected food frequency and 24-hour recall instruments. The SA Medical Research Council dietary analysis software program was used to analyse the dietary data collected.[8] Dietary patterns included meal frequency, frequency of food consumption and consumption of breakfast, all of which were measured via a food frequency questionnaire. The 24-hour recall was provided by two volunteers per school working on the NSNP with the purpose of verifying the meal provided at the school a day before.

The reliability and validity of the questionnaires were tested in a pilot study. Necessary adjustments were made thereafter; the children involved in this pilot study were not included in the final survey. Statistical analysis was conducted to determine the mean weight and height of children as well as the z-score distribution of the anthropometric variables. The data were exported into Statistical Package for Social Sciences version 21 (IBM Corp., USA) for further analysis. The Centers for Disease Control/National Center for Health Statistics 2000 growth references/standards were used to determine z-score values during analysis. Dietary reference intakes were used to estimate nutrient adequacy of the subjects.[9]

Results

Socioeconomic factors

The study included 602 children from 18 primary schools. Their ages ranged from 9 to 13 years and females represented 63.3% of the total sample. Table 1 provides a summary of the demographic characteristics of the study population.

Anthropometric measurements

WAZ values of children The majority of children were of normal weight for age (Fig. 1).

HAZ values of children The interval for normal height had the highest number of children, both male and female. Of the children, <10% were stunted and <8% were tall for their age (Fig. 2).

BMI values of children The majority of the participants (67.2% of males, 71.5% of females) had normal BMI values for their age. Of the female participants, 20.5% were <–1 standard deviation (SD) while the same was true for 19.9% of the male participants. The difference was not regarded as significant.

Few children (8.6% of males, 12.3% of females) were >+1 SD. 12

SAJCH

Table 1. Demographic characteristics of the study population (N=602) Characteristic

n (%)

Age (years) 9

5 (0.8)

10

160 (26.6)

11

213 (35.4)

12

213 (35.4)

≥13

11 (1.8)

Sex Male

381 (36.7)

Female

221 (63.3)

Grade in school 4

131 (21.8)

5

191 (31.7)

6

220 (36.5)

7

60 (10.0)

Number of people in household 2-4

120 (19.9)

5-7

344 (57.1)

8 - 10

115 (19.1)

≥11

23 (3.8)

Income from social grants Not receiving any grant

182 (30.2)

Child grant

304 (50.5)

Pension fund

98 (16.3)

Both child grant and pension fund

18 (3.0)

Type of employment Unemployed

Father (%) Mother (%) 17.3

49.5

Service worker

18.9

24.0

Public servant

23.3

11.0

Retail

5.8

9.0

Self-employed

2.5

2.0

Technical

13.6

1.3

Professional

0.3

0.2

Unknown

3.2

0.2

Not applicable – deceased or did not know mother or father

15.1

2.8

School attendance

School attendance was also determined as a percentage, with 100% meaning all school days had been attended. The majority (74.9%) of participants attended all their school days, 19.1% missed 1 - 2 school days, 4.4% missed 3 - 4 days and 0.7% missed 5 - 8 days. Only 1.3% missed 9 - 16 school days.

Dietary patterns

The majority (77.2%) of the participants ate breakfast more than five times a week, while 22.5% did not have breakfast or did so only occasionally. Ninety-five percent of subjects reported eating three or four meals per day, with 93.5% consuming porridge and 74.8%

MARCH 2017 Vol. 11 No. 1

RESEARCH (56.6%), potato chips (54.3%) and ice cream (51.5%) one to three times a week. Dietary pattern data also included meals provided by the NSNP. The average intake of nutrients is presented in Table 2.

57.0 56.1

60

Subjects, %

50 40 27.0

30

Discussion

31.2

20 6.8 8.1

10

7.6

2.7

0.5 0.5

1<z≤2

2<z≤3

1.0 1.4

0

z<–3

–3≤z<–2

–2≤z<–1

–1<z≤1

z-score interval

Female

Male

Fig.1.WAZ values (N=602).

70 57.0

60

59.3

Subjects, %

50 40 30.7 29.0

30 20 7.1

10

10.4 4.2

0.3 0.9

0

z<–3

–3≤ z<–2

-2≤z<–1

–1<z≤ 1

0.8

0.5

1<z≤ 2

2<z≤ 3

z-score interval

Female

Male

Fig. 2. HAZ values (N=602).

80 67.2

70

71.5

Subjects, %

60 50 40 30 16.316.3

20 10 0

0.8 0.9 z<–3

11.3

3.5 2.7 –3≤z<–2

8.1 1

–2≤z<–1

–1<=z≤1

1<z≤2

0.5

2<z≤3

z-score interval

Female

Male

Fig. 3. BMI values (N=602).

eating bread four to seven times a week and 21.4% and 15.8% eating rice and potato, respectively. Fish (76.2%), chicken necks, feet and livers (70.1%) and chicken meat (63.5%) were the highest consumed protein products at one to three times a week. The majority of subjects reported consuming eggs (72.6%), yoghurt (59%) and milk (58.6%) one to three times a week. Furthermore, the majority of subjects reported consuming peanuts (75.4%), beans (69.4%) and peas (39.9%) one to three times a week.

More than half of the participants reported eating vegetables one to three times a week, while <18% were consuming vegetables regularly (four to seven times a week). Cabbage (59.6%), beetroot (59.3%), traditional vegetables (53.2%) and spinach (51.8%) were mostly consumed at least one to three times a week. Most of the participants reported consuming fruit one to three times a week, with only 4.5% not eating fruit at all. Apples and bananas were consumed by 26.7% and 18.9%, respectively, four to seven times a week. The majority of subjects reported drinking tea or coffee (66.4%), artificial juice (41.3%) and carbonated drinks (11.0%) four to seven times a week. Margarine was the most-used bread spread with 36.2% of participants using it four to seven times a week, while 14.7% and 10.8% reported using peanut butter and jam, respectively, four to seven times a week. The majority of participants reported eating maize chips (60.6%), sweets 13

SAJCH

The majority of the children were in grades 4 - 6. This is in contrast to a Jamaican study, which selected children in grades 3 and 4.[10] Similar grades were selected by an SA study that focused on grades 5 - 7.[11] The present study’s selection of grades was based on the ability of the children to recall information. Most of the participants were within the age range 10 - 12 years.

The unemployment rate in Limpopo was 38.9% in 2011.[7] In the present study, participants’ fathers had showed a lower unemployment rate of 17.3% compared with the national statistic of 29.8% and the provincial statistic of 38.9%, while close to 49.5% of participants’ mothers were unemployed. In addition, 24.0% of mothers occupied lower-paid jobs such as service workers compared with fathers (18.9%), and only 11.0% of mothers were public servants compared with 23.3% of fathers. It has been reported by Casale and Posel[12] that even though more women are represented in the labour force than previously, there is still a rising rate of female unemployment and low-paid and insecure forms of employment are becoming increasingly feminised. It is well known that the unemployment rate is high among women in SA and that this is worse in rural areas.[13] Subjects receiving child grants constituted 50.5% of the study population, meaning that they had passed the social development means test. This observation is in agreement with the report of the Department of Health and Social Development, which stated that over half of rural households nationally received government grants.[14] Such grants can be used to purchase food, thus leading to improved nutritional status. At the time of this study, the child grant was valid for children up to 14 years of age and the amount was ZAR180 per month per child. Approximately 69.8% of children in this study were receiving some form of government grant, supporting the characterisation of Limpopo as being largely rural.[15] The WAZ values in the present study showed that the majority of females (91.6%) and males (90.0%) were within the acceptable weight range. Male children were found to be most affected by underweight (<–2 SD) at 9.5% while the figure for females was at 7.8%, indicating low public health significance given that the prevalence of underweight was <10%.[16] The results were similar to those reported by the SA anthropometric survey, the National Food Consumption Survey (NFCS)-Fortification Baseline and the NFCS, which found 9.0, 9.3 and 10.0% of SA children to be underweight, respectively. [17-19] Compared with a Nigerian semi-urban population study that revealed 25.5% of children were underweight,[20] SA has a low prevalence of underweight children. The good nutritional status of SA children could be explained by the fact that their dietary patterns suggest that they were consuming adequate levels of all macro- and micronutrients. Their good school attendance also meant they received the school meal daily.

HAZ values were similar for both female and male children between 9 and 12 years. Similar results have been reported by Oninla et al.,[22] with boys being taller than girls until the age of 2 years and there being no statistical difference in height for age in both sexes thereafter. This study showed that 10.4% of male participants and 7.1% of females were moderately stunted (<–2 SD). The incidence of stunting was low compared with World Health Organization guidelines, which rate a country as moderately affected by stunting if 25 - 50% of its children are stunted.[16] Studies conducted in Nigeria[22] and on the tea-garden workers of Assam, India,[23] found a high level of stunting among children between 9 and 14 years at 27.6% and 53.9%, respectively. The results of the present study concur with those of a study carried out in Agincourt, a rural sub-district of Mpumalanga Province, SA, which reported the stunting rate

MARCH 2017 Vol. 11 No. 1

129.1 RDA = recommended dietary allowance; AI = adequate intake.

102.0 96.2

131.6 14.2

5.1 4.81

14.48 130.0

86.0 78.0

126.4 14.3

4.3 3.6

13.9 133.6 14.7

4.1

132.7 14.6 Vitamin E (mg/ d)

96.0 4.7 Vitamin D (µg /d)

82.0

85.7

102.6 89.0

86.9 38.6

307.9 267.14

39.1 80.4

91.1 88.6

118.4 36.2

273.2 265.9

53.3 96.6 43.5

273.7

91.0 41.4 Vitamin C (mg/d)

SAJCH

95.0 285.1 Folate (µg/d)

91.2

209.2 182.5 1 255.3 1 094.8 172.6 161.8 1 035.5 971.04 1 084.4

184.0 1 104.2 Vitamin A (µg/d)

180.7

154.8

96.3 98.8

121.6 185.8

7.7 7.9

1 46.92 33.3

185.0 92.5

20.3 39.91

14.8 7.4

24.3 52.3 62.8

7.3

14

61.3 73.6 Iodine (µg/d)

96.3 7.7 Zinc (mg/d)

91.3

141.3

24.6 25.5

136.3 11.3

320.5 331.6

10.9 126.3

23.7 24.5

127 10.1

308.5 318.1

10.2 131.2 10.5

314.1

141.3 11.3 Iron (mg/d)

25.1 325.9 Calcium (mg/d)

24.2

79.6

233.2 245.7

68.4 20.7

303.1 319.5

21.2 75.0

224.1 234.4

65.3 19.5

291.3 304.7

19.6 74.2 19.3

296.7

68.7 21.3 Fibre (g/d)

252.9 328.8 Carbohydrate (g/d)

228.2

-

173.2 178.2

50.2

58.9 60.6

52.4 -

165.3 171.2

51.5

56.2 58.2

50.4 51.6

57.8

54.8 Fat (g/d)

187.1 63.6 Protein (g/d)

170.0

96.2

Female Male

98.4 8 370.5 9 416.7 93.1 88.1 8 102.2 8 437.6 8 248.5

95.1 9 103.2 Energy (kJ/d)

94.8

Female (N=127) Male (N=87) Female Male Female (N=137) Male (N=77) Male Female (N=104) Male (N=57)

Nutrients

Female

% of RDA/AI % of RDA/AI % of RDA/AI

Age 10 Table 2. Nutrient intake compared with the RDA

Nutrient intake

Nutrient intake

Age 11

Nutrient intake

Age 12

RESEARCH to be 5.0% and 9.0% among children of 5 - 9 and 10 14 years of age, respectively.[24] The lower levels of stunting observed in the present study could be the result of a combination of factors, which may include direct or indirect nutritional interventions, e.g. child support grants, the NSNP and fortification of maize meal and bread. It was reported in a review study conducted on the impact of school feeding in developing countries that school-fed children gained 1.43 cm (in height) more than controls who were not given the school meal.[25] However, the present study did not include a comparative study or a before-or-after trial effect owing to a lack of assessment before the commencement of the NSNP in Limpopo.

There was a small difference in the BMI pattern of the two sex groups within the age range from 9 to 12 years.

Female participants were slightly heavier than males, contrary to what has been reported in Osun State, Nigeria, where wasting was higher in females (15.0%) than in males (13.0%).[26] In the current study the level of severe wasting (<–3 SD) was low for both males (0.9%) and females (0.8%). Only 3.4% of females and 2.7% of males had moderate wasting (<–2 SD). Studies conducted in Nigeria among primary schoolchildren reported the grade 1 level of wasting to be 77.3%, which indicated that wasting remains a problem in some African countries.[27] The present study concurs with the NFCS and the SA anthropometric survey that reported wasting to be at 3.7% and 2.6%, respectively. [17,18] Very few males (0.5%) and females (1%) were overweight (>+2 SD). The results for overweight were low compared with the results of a study conducted in five SA provinces, which showed a prevalence of overweight of 14.0% for boys and 17.9% for girls.[28] The present study did not include children from urban areas as their schools were excluded from receiving NSNP meals.

The average intake of all nutrients of all age groups and sexes was good except for females in age group 10, who reported 52.3% of the recommended dietary allowance (RDA) for iodine intake, and males in age group 11, who had 20.3% for iodine. All age groups had a calcium intake of ≤25.5% of the RDA and AI (adequate intake).

They also reported a moderate intake of fibre, ≤79.6% of the AI with the exception of males aged 11 years. The intakes of protein, carbohydrates, vitamin A and vitamin E were above the RDA/AI. Children who either missed or only had breakfast occasionally constituted 22.5% of the study population. This is in agreement with what has been reported[29] in another SA study, the Birth to Twenty Cohorts study, where only 76.4% of children aged 13 had breakfast. The recommendation of 400 g of fruits and vegetables a day,[30] the equivalent of five servings of 80 g each, was not met in this study group. The majority of participants consumed vegetables or fruits one to three times a week, including those provided by NSNP meals.

Similar results were reported in KwaZulu-Natal Province, SA, where consumption of fruits and vegetables was low, especially among the low living standard measure categories. [31] Good school attendance in this study could not be attributed to the school feeding programme since the researcher did not obtain retrospective data preceding the feeding programme.

Conclusion

The unemployment rate of mothers was high (49.5%) compared with that of fathers (17.3%), and the

MARCH 2017 Vol. 11 No. 1

RESEARCH majority of subjects (69.8%) depended on government grants. The anthropometric measurements revealed that most children were of normal weight, and normal height and had a normal weight for height. A small number of children showed stunting and wasting, which could have serious health implications and warrants attention.

Most children attended school and consumed the NSNP meals. Fruit and vegetable consumption was reported to be low in most subjects.

There is a need for more continuous nutritional assessment of learners receiving NSNP meals to allow for a proper evaluation of programme effectiveness. Further studies should explore the impact of the NSNP on the educational performance outcomes of the primary school children.

1. Briggs B. School feeding programs: Summary of the Literature and Best Practices. Village Hope Technical Report 6. Idaho, USA:Village Hope, 2008:3-4.

2. Department of Basic Education, South Africa. National School Nutrition Programme. http://www.education.gov.za/Programmes/NationalSchoolNutritionProgramme (accessed 24 June 2013).

3. Wilcox M, Israel R, Praun A. Lessons Learned from Honduras School Nutrition and Health Assessment Study. Newton, Mass., USA: Education Development Center, 1993:5-7.

4. Simeon DT, Grantham-McGregor S. Effects of missing breakfast on the cognitive functions of school children of differing nutritional status. Am J Clin Nutr 1989;49(4):646-653.

5. Meyers AF, Sampson AE, Weitzman M, Rogers BL, Kayne H. School breakfast program and school performance. Am J Dis Child 1989;143(10):1234-1239.

6. Kallman K. Food for Thought: A review of the National School Nutrition Programme. In: Leatt A, Rosa S, eds. Towards a Means to Live: Targeting Poverty Alleviation to Make Children’s Rights Real. Cape Town: Children’s Institute, University of Cape Town, 2005:12-13.

7. Statistics South Africa. Census 2001: Provinces at a Glance. Pretoria: SSA, 2012:1-43.

8. Food Finder 3. Dietary analysis software program, version 1.0.7. Cape Town: South African Medical Research Council, 2002.

9. Mahan KL, Escott Stump S. Krause’s Food Nutrition and Diet Therapy, 12th ed.

St. Louis, Mo,: Elsevier, 2008:39-135.

10. Grantham-McGregor SM, Chang S, Walker SP. Evaluation of school feeding programs: Some Jamaican examples. J Clin Nutr 1998;67(Suppl):S785-S789.

11. Faber M, Laubscher R, Berti C. Poor dietary diversity and low nutrient density of the complementary diet for 6- to 24-month-old children in urban and rural KwaZulu-Natal, South Africa. Matern Child Nutr 2014;5(5):1-18. http://dx.doi. org/10.1111/mcn.12146 12. Casale D, Posel D. The continued feminisation of the labour force in South Africa: An analysis of recent data and trends. S Afr J Econ 2002;70(1):156-184. http://dx.doi.org/10.1111/j.1813-6982.2002.tb00042.x 13. Kongolo M, Bamgose OO. Participation of rural women in development: A case study of Tsheseng, Thintwa, and Makhalaneng villages, South Africa. J Int Womens Stud 2002;4(1):79-92.

14. Department of Health and Social Development, South Africa. Socio-economic Impact of HIV and AIDS on Population and Development in Limpopo Province. Polokwane: Limpopo Provincial Government, 2006:4-5.

15

SAJCH

15. Neves D, du Toit A. Rural livelihoods in South Africa: Complexity, vulnerability and differentiation. J Agrar Change 2013;13(1):93-115. http://dx.doi. org/10.1111/joac.12009 16. De Onis M, Habicht JP. Anthropometric reference data for international use: Recommendations from a World Health Organization expert committee. Am J Clin Nutr 1996;64(4):650-658.

17. Labadarios D, Steyn N, Maunder E, et al. The National Food Consumption Survey (NFCS): Children aged 1 - 9 years, South Africa. Stellenbosch: Department of Health, Directorate of Nutrition, 2000:1-20.

18. Vorster HH, Oosthuizen W, Jerling JC, Veldman FJ, Burger HM. The Nutritional Status of South Africans: A Review of the Literature from 1975 - 1996. Durban: Health Systems Trust, 1997:7-10.

19. Kruger HS, Steyn NP, Swart EC, et al. Overweight among children decreased, but obesity prevalence remained high among women in South Africa, 1999- 2005. Public Health Nutr 2011;15(4):594-599. http://dx.doi.org/10.1017/S136898001100262X 20. Fetuga MB, Ogunlesi TA, Adekanmbi AF, Alabi AD. Nutritional status of semiurban Nigerian school children using 2007 WHO reference population. West Afr J Med 2011;30(5):331-336.

21. Botton J, Heude B, Maccario J, Ducimetière P, Charles MA, LFVS Study Group.

Postnatal weight and height growth velocities at different ages between birth and 5 y and body composition in adolescent boys and girls. Am J Clin Nutr 2008;87(6):1760-1768. https://dx.doi.org/10.1159/000362203 22. Oninla SO, Owa JA, Onayade AA, Taiwo O. Comparative study of nutritional status of urban and rural Nigerian school children. J Trop Pediatr 2007;53(1):3943. https://dx.doi.org/10.1093/tropej/fml051 23. Medhi GK, Barua A, Mahanta J. Growth and nutritional status of school age children (6 - 14 years) of tea garden worker of Assam. J Hum Ecol 2006;19(2):8385.

24. Kimani-Murage EW, Kahn K, Pettifor JM, et al. The prevalence of stunting, overweight and obesity, and metabolic disease risk in rural South African children.

BMC Public Health 2010;10(1):158-170. http://dx.doi.org/10.1186/1471-245810-158 25. Jomaa LH, McDonnell E, Probart C. School feeding programs in developing countries: Impacts on children’s health and educational outcomes. Nutr Rev 2011;69(2):83-98. http://dx.doi.org/10.1111/j.1753-4887.2010.00369.x 26. Falade OS, Otemuyiwa I, Oluwasola O, Oladipo W, Adewusi SA. School feeding programme in Nigeria: The nutritional status of learners in a public primary school in Ile-Ife, Osun State, Nigeria. Food Nutr Sci 2012;3(5):596-605. http:// dx.doi.org/10.4236/fns.2012.35082 27. Goon DT, Toriola AL, Shaw BS, et al. Anthropometrically determined nutritional status of urban primary schoolchildren in Makurdi, Nigeria. BMC Public Health 2011;11(1):769-777. http://dx.doi.org/10.1186/1471-2458-11-769 28. Armstrong MEG, Lambert MI, Sharwood KA, Lambert EV. Obesity and overweight in South African primary school children – the Health of the Nation Study. S Afr Med J 2006;96(5):439-444. https://dx.doi.org/10.1080/22 201009.2006.10872144 29. Feeley A, Musenge E, Pettifor JM, Norris SA. Changes in dietary habits and eating practices in adolescents living in urban South Africa: The birth to twenty cohort. Nutrition 2012;28(7-8):e1-e6. http://dx.doi.org/10.1016/j. nut.2011.11.025 30. World Health Organization, Food and Agricultural Organization. Diet, nutrition, and the prevention of chronic diseases: Report of a joint/FAO expert consultation. Geneva: WHO, 2003;95-101.

31. Faber M, Laubscher R, Laurie S. Availability of, access to and consumption of fruits and vegetables in a peri-urban area in KwaZulu-Natal, South Africa. Matern Child Nutr 2013;9(3):409-424. http://dx.doi.org/10.1111/j.1740-8709.2011.00372.x

MARCH 2017 Vol. 11 No. 1

RESEARCH

The promotion of oral health in health-promoting schools in KwaZulu-Natal Province, South Africa M Reddy, PhD; S Singh, PhD Discipline of Dentistry, School of Health Sciences, University of KwaZulu-Natal, Durban, South Africa Corresponding author: M Reddy (reddym@ukzn.ac.za) Background. Oral health promotion is a cost-effective strategy that can be implemented at schools for the prevention of oral diseases. The importance and value of school-based interventions in children has been identified in South Africa (SA). Although oral health strategies include integrated school-based interventions, there is a lack of published evidence on whether these strategies have been translated into practice and whether these programmes have been evaluated.

Objective. To assess the efficiency and sustainability of the toothbrushing programme implemented at health-promoting schools in KwaZulu-Natal Province, SA. Methods. A mixed-methods approach was used for this study, conducted at 23 health-promoting schools in KwaZulu-Natal using focus group discussions. Triangulation was used for evaluation.

Results. The intervention implemented had created awareness of oral health for learners, educators and parents. Findings in this study indicate that although there were benefits obtained from this school-based intervention, many challenges, such as time constraints, large classes and a lack of adequate resources and funding, affected the sustainability of the programme.

Conclusion. The school setting has the potential to deliver integrated preventive and promotive programmes provided they are supported by adequate funding and resources.

S Afr J Child Health 2017;11(1):16-20. DOI:10.7196/SAJCH.2017.v11i1.1132

Oral health promotion has been identified as a cost-effective strategy to reduce the burden of oral diseases in local communities in South Africa (SA).[1] Although most oral diseases are preventable, they are irreversible once established and affect function and quality of life. [2] Oral health promotion strategies therefore support preventive interventions. Lifestyle behaviours such as consuming foods and drinks high in sugars, smoking and drinking alcohol can affect oral health.[3] These behaviours are controllable in school settings through school policies, adjusting the physical environment and implementing education in oral health.[2] Schools, attended by over a billion children worldwide, have been identified as the most creative and cost-effective way of improving oral health and thereby quality of life through school-based interventions. These interventions provide the foundations for healthy patterns of behaviour that follow into adulthood.[3,4] International reviews conducted by The Cochrane Collaboration were inconclusive on the effectiveness of school-based interventions.[5] However, studies conducted in China, Indonesia, Brazil and Iran show positive results.[6-8] The importance and value of the impact of school-based interventions on children has been identified in SA.[9] SA oral health policies and strategies have therefore prioritised school-based preventive programmes. [10] The Oral Health Ten Point Plan 2011 - 2015 for KwaZulu-Natal Province includes integrated school-based toothbrushing, fissure sealant, screening and education programmes.[11] However, there is a lack of published evidence on whether these strategies have been translated into practice and whether the programmes have been evaluated. Evaluation generates information that can be utilised by stakeholders responsible for the improvement of interventions which ensures effective interventions, highquality practice, maximised use of limited resources, provision of feedback to all participants and informed policy development and implementation.[12,13] Evaluation, which includes both process and outcome data, should be a key component in planning school oral health promotion programmes.[14] Documenting and publishing 16

SAJCH

these interventions enables the sharing of knowledge globally.[15] However, international reviews indicate that current evaluation outcome measures are inappropriate and of poor quality.[16] Process evaluation data inform future planning and delivery, while outcome data assess the short-, medium- and long-term effects of the intervention.[17] This study evaluated the short-term effects of the intervention for efficiency and sustainability. The aim was to evaluate an implemented toothbrushing programme at health-promoting schools in KwaZulu-Natal to test for efficiency and sustainability.

This study formed part of a bigger study that examined the viability of incorporating oral health promotion into the Health-Promoting School Initiative in KwaZulu-Natal. The study was divided into three phases, namely assessment (phase 1), implementation (phase 2) and review (phase 3).

Methods

This study was conducted at 23 health-promoting primary schools, randomly selected from a total of 154, in the 11 districts of KwaZuluNatal. Fig. 1 illustrates details of school participation.

A situational and needs analysis was conducted in phase 1 using interviews, questionnaires, a data capture sheet and the World Health Organization Decayed Missing Filled Teeth Tool (WHODMFT Tool). In phase 2, appointments were made telephonically with the school health teams of the 20 consenting schools to report on phase 1 of the study, formulate and implement interventions based on learners’ needs and sign a memorandum of understanding for interventions. Discussions included: • a toothbrushing programme • instructions on toothbrushing technique, toothpaste application and toothbrush storage • parental involvement • outsourcing supplies • incorporating oral health education into school curriculum and at parent meetings

MARCH 2017 Vol. 11 No. 1

RESEARCH • sugary snacks control by vendors and tuck shops • healthier lunches.

Toothbrushes and a 3-month supply of toothpaste were provided to commence the programme. A mobile messenger application was set up between the researcher and school health teams for ease of communication and sharing of ideas between schools.

This article reports briefly on phase 1 and focuses on the evaluation of the toothbrushing programme using focus group discussions in phase 3 of the study. Twenty schools were given appointments for focus group discussions; however, only 13 schools were visited owing to rains (n=2) and unavailability for scheduled appointments (n=5). Focus group discussions, lasting 30 - 45 minutes and recorded with participants’ permission, were conducted 6 months after the implemented programme for insight into their programme experience. Discussions focused on opportunities, challenges, benefits and support available for this intervention. Qualitative responses were transcribed verbatim, examined for broad categories and then further categorised into phenomena using open coding.

Links were then formed between categories (axial coding), after which selective coding was used to create core categories. [18] Quantitative responses were analysed using SPSS version 21.0 (IBM Corp., USA).

A concurrent mixed-methods approach with both qualitative and quantitative data was therefore used. To ensure validity, data source triangulation, which entails gathering evidence from diverse sources and drawing conclusions based on data collected, was used for evaluation.[19,20] Gatekeeper permission was obtained from the Department of Health and the Department of Basic Education. The study was approved by the Humanities and Social Sciences Research Ethics Committee of the University of KwaZulu-Natal (UKZN) (ref. no. HSS/0509/013D). The UKZN ethical guidelines were used to ensure confidentiality, consent to conduct interviews and data management.

Results

Twenty-three schools participated in phase 1 of the study. Quantitative responses in the questionnaire indicated that 55% of the schools had toothbrushing programmes. However, further investigations of school records and qualitative questionnaire responses established that these programmes did not occur regularly.[21] Data capture sheet responses indicated that health messages formed part of the curriculum in the majority (96%) of the schools. However, staff indicated that they were not confident in conducting oral health promotion programmes owing to a lack of basic knowledge about oral health.[21] Results obtained from the WHODMFT Tool (Table 1) indicated that only 27% of the learners were caries free, with the majority (90%) requiring preventive care.[22] A total of 2 065 grade 1 learners from 20 schools participated in the toothbrushing programme. This article reports only on feedback received from 13 schools that participated in phase 3 of the study.

Table 2 illustrates the frequency of the toothbrushing programmes per week that were conducted at these schools.

One school (8%) was not able to continue with the programme owing to large classes, which subsequently impacted on contact time. Problems were also experienced with storage of toothbrushes and cups to rinse owing to only one basin and tap being available.

Educators therefore sent the toothbrushes and toothpaste home for learners to use; however, this created challenges as some learners did not use them at home or lost them. Three (23%) schools conducted the programme twice a week and one (8%) thrice a week. Daily toothbrushing was conducted by 8 (61%) schools, although they did highlight time constraints and lack of resources as a problem. It was alarming to note that learners at 12 (92%) schools did not have toothbrushes and paste to brush their teeth at home.

17

SAJCH

PHASE 1

PHASE 2

PHASE 3

ASSESSMENT

IMPLEMENTATION

REVIEW

13 participated

20 participated

5 unavailable for appointment 2 could not be reached due to rains and roads being inaccessible

23 participated 2 refused 1 incomplete questionnaire Fig. 1. Details of school participation in phases 1, 2 and 3.

Table 1. Caries experience and treatment needs of primary teeth of 6-year-olds in KwaZulu-Natal Percentage (%) Caries experience/incidence

73

Needs Preventive/caries arresting

90

Surface fillings

35

Extractions

5

Table 2. Frequency of toothbrushing programmes per week Number of schools (N=13)

Frequency

1

0

3

2 times/week

1

3 times/week

8

Daily

Table 3. Awareness of oral health Focus group response (%) (N=13)

Target population

100

Learners

23

Educators

23

Parents

Three salient themes, namely awareness, support for interventions and resources, emanated from the data. The benefits and challenges are embedded in the identified themes for reporting.

Awareness

Awareness of oral health as reported in focus group discussions is illustrated in Table 3. Study findings indicated that the importance of oral health awareness was created for learners, educators and parents.

Responses from all (100%) focus group participants emphasised that the impact of the interventions had created awareness of oral health for their learners: ‘Learners now know the importance of brushing their teeth.’

‘There is great improvement in learners' oral hygiene.’

Participants at 3 (23%) schools further indicated that this programme had created awareness of oral health for educators: ‘It was beneficial to educators – an eye opener – they are now aware of the importance of oral health.’

MARCH 2017 Vol. 11 No. 1

RESEARCH Furthermore, parent awareness had been created at 3 (23%) schools by learners who asked their parents why they did not brush their teeth at home. Learners had also become increasingly aware of eating correctly at 2 (15%) schools by being particular about what they ate and correcting friends eating unhealthy foods or lunches. One participant attributed this marked improvement to awareness created by the programme: ‘A learner, offered a sweet at her dentist, refused knowing this was not good for her teeth.’

Participants at 4 (31%) schools indicated that appropriate awareness of the correct brushing technique was created: ‘Learners not familiar with the toothbrushing routine learnt how to brush their teeth.’

Responses further indicated learners’ awareness of the importance of brushing their teeth twice a day.

Challenges and limitations

Challenges and limitations experienced with the intervention are illustrated in Table 4.

Responses from 3 (23%) schools indicated that learners chewed their toothbrushes and ate toothpaste, resulting in depleted supplies which impacted on the programme. Replacing supplies was unaffordable as parents depended largely on social grants for income. The majority (70%) of school participants identified time constraints as a limitation to the intervention: ‘Only what is relevant to the curriculum is done due to time constraints. It’s impossible to manage large numbers of learners for oral health promotion activities.’

One participant (8%) also specified that low staff numbers resulted in higher workloads. Seemingly, participants at 2 (15%) schools managed: ‘Coped with time – procedure took five minutes.’

Support for interventions

Support for interventions is illustrated in Table 5. Study findings indicated that participants at 5 schools (39%) had some support for oral health interventions, while 7 (54%) had none. Support for interventions received from Colgate World of Care and the provincial Department of Health was not continuous and was dependent upon the availability of supplies: ‘Oral health personnel only visit the school when supplies are available.’

‘Colgate World of Care visits once a year.’

Evidence of water shortages in the Sisonke District was brought to the researchers’ attention by educators who conveyed that children walked 5 km daily to fetch water from the river when there was no rain, resulting in learners missing valuable contact time. Participants also indicated a lack of basins and cups for rinsing: ‘There are insufficient cups for rinsing and only one basin to forty learners.’

Learners consequently left the space untidy. This had to be cleaned, resulting in time wastage. Another key challenge for the programme was the hygienic storage of toothbrushes and toothpaste supplies. This challenge was reported by all participants. Sponsorships for supplies could not be secured, resulting in programme discontinuation: ‘There are no sponsors to replace depleted supplies’.

Hygienic toothbrush storage was a challenge, especially in larger classes. Furthermore, labelling and distribution of toothbrushes was time consuming.

Discussion

Interventions in the school setting have been identified as the most creative and cost-effective way for improvement of health, oral health and, in turn, quality of life.[23] Integrated, school-based preventive and promotive oral health programmes are prioritised in KwaZulu-Natal. [11] However, there is a paucity of information on their implementation and effectiveness. Study findings indicated that the toothbrushing programmes were not implemented regularly. Therefore, knowledge gained from this study could inform future school-based preventive programmes. Although benefits were identified, many challenges affected the sustainability of this intervention.

The effectiveness of brushing daily with fluoride toothpaste is supported and reinforced by clinical trials.[24,25] Additionally, schools are often used as a platform for supervised toothbrushing Table 4. Challenges and limitations Focus group response (%) (N=13)

Challenge/limitation

23

Learners chew toothbrush/eat toothpaste

70

Time constraints

8

Low staff numbers/higher workloads

Table 5. Support for interventions

However, one quintile 1 school had support: ‘Nurses come almost every week and advise children. Colgate gives support.’

Focus group response (%) (N=13)

Support

39

Some support

54

No support

All (100%) participants indicated that the programme was dis­ continued once supplies were depleted as there was no budget for oral health from the provincial Department of Basic Education. Although educators recommended the purchasing of supplies by parents, this did not occur owing to affordability. The toothbrushing programme had to be discontinued once supplies were depleted.

8

Support

100

No support – budget

Resources

Challenges faced in terms of resources are shown in Table 6.

All schools (100%) identified resource availability as a challenge to programme success. Participants at 5 (38%) schools highlighted water access as a major barrier to the implementation and success of the intervention: ‘Toothbrushing at school is difficult because there is no running water.’

18

SAJCH

Table 6. Challenges for resources Focus group response (%) (N=13)

Challenge/barrier

100

Resource availability

38

Water access

100

Toothbrush and toothpaste supplies

100

Toothbrush storage

MARCH 2017 Vol. 11 No. 1

RESEARCH programmes.[26] It is evident from this study that awareness of the importance of daily brushing had been created for learners, educators and parents, especially those from rural areas. Moreover, the programme successfully inculcated correct brushing techniques.

It was further noted that the majority of learners did not brush at home because they did not have toothbrushes or toothpaste. This is supported by the high (73%) caries incidence noted in phase 1.

Interview responses suggested that this could be due to affordability, as many parents were poor, unemployed or dependent on social grants. The study findings support WHO recommendations for oral health promotion through schools.[27] Saied-Moallemi et al.[28] also argued for parental awareness of interventions at schools.

Evidently, educators also benefited from this programme through self-reflection.

The intervention further highlighted the importance of correct eating habits, thus ensuring informed choices about lunches and tuck shop and vendor purchases by learners. These findings demonstrate the modification of oral health risk behaviours in learners through oral health promotion. Similar findings were noted in Tanzania.[29] Although some success was reported, educators faced many challenges with programme implementation. The majority of schools encountered time constraints. Educators found the programme time consuming especially with larger classes. Staff shortages, resulting in higher workloads, and a demanding curriculum further impacted on the programme. Additional activities undoubtedly added to existing workloads.[30] Evidently, however, some participants coped.

The researcher observed that programme success was dependent upon educators’ commitment. Similar observations were noted in school-based brushing programmes in southern Thailand.[31 The study findings indicated that the majority of schools received no or intermittent support for oral health promotion interventions.

Although investigations revealed partnerships between Colgate World of Care, the Department of Basic Education and the Department of Health, only one mobile unit was operational in KwaZulu-Natal. Supplies distributed by school health nurses were inconsistent and not delivered to all districts. Numerous schools were unaware of toothbrush and toothpaste supplies offered by Colgate World of Care and the Department of Health. Hence, it can be assumed that supplies in districts are largely dependent upon staff and resource availability, and initiatives undertaken by school health nurses and oral health personnel in their respective districts.

Availability of funding for programme sustainability is imperative. [29] Although SA has school-based oral health intervention strategies, implementation is dependent upon the availability of funding and materials for programme sustainability. Study findings indicated that the Department of Education did not have a dedicated budget for health and oral health promotion at schools. This was confirmed with baseline data from phase 1. These findings are further corroborated by Peterson and Kwan,[29] who claimed that limited national budgets in countries worldwide impacted on the implementation of integrated health promotion. However, a recent global survey conducted by the WHO indicated that school-based oral health programmes were frequently subsidised by national and provincial governments.[32] This was not evident in KwaZuluNatal. Considering the prioritising of funding for the high burden of disease in KwaZulu-Natal, attempts should be made to secure funding outside the public sector in SA.[33] Moreover, with financial constraints, the focus on effective evidence-based interventions is imperative. This can be achieved by incorporating oral health into general health to ensure positive gains from invested resources.[32] Reviews of the context of school oral health service delivery in KwaZulu-Natal depict an inequity in resource allocations. All schools in this study identified challenges with securing resources for their programmes. Proper access to water, required for the toothbrushing programme, posed an obstacle for some schools, as water availability 19

SAJCH

continues to be a challenge, especially in Umkhanyakude, Sisonke and Umzinyati districts.[34] Study findings indicated that educators had difficulty storing toothbrushes hygienically and complained about the lack of proper cups and basins for rinsing. Labelling and distributing toothbrushes was time consuming, thereby impacting on teaching time. For a successful programme, careful thought must be given to providing adequate resources to address these problems to ensure educators’ willingness to conduct programmes.

This study has demonstrated that school-based interventions could have a positive impact on oral health for learners and communities by providing an opportunity for a holistic approach to healthy lifestyles and environments.[32] The literature suggests that schoolbased toothbrushing programmes have made a positive impact on children’s oral health.[13] This is evident in a study conducted in Scotland where long-term outcome data obtained over 2 years demonstrated a positive effect on learners by showing a decrease in the prevalence of caries.[35]

Conclusions

The results of this study suggest that the school setting has the potential to deliver integrated, preventive and promotive oral healthcare programmes. The interventions implemented in this study created awareness of oral health for learners, educators and parents in the short term. Evidence for the benefits of toothbrushing programmes is indisputable. To ensure long-term positive impacts on the oral health of communities in Kwazulu-Natal, factors affecting implementation in schools with limited resources warrant careful consideration.

Acknowledgements. This research project was supported by research grants from the University of KwaZulu-Natal and the National Research Foundation. Toothbrushes and toothpaste were supplied by Colgate World of Care.

1. National Department of Health, South Africa. National Oral Health Strategy (confidential draft for comment only). 2010;1-15.

2. Kwan SYL, Petersen PE, Pine CM, et al. Health-promoting schools: An opportunity for oral health promotion. Bull World Health Organ 2005;83(9):677-685.

3. Petersen P. The World Oral Health Report 2003. Continuous improvement of oral health in the 21st century: The approach of the WHO Global Oral Health Programme. Community Dentistry and Oral Epidemiology 2003;31(Suppl 1):3-24. https://dx.doi.org/10.1046/j..2003.com122.x 4. Tones K, Thornes STN. Health Promotion. Effectiveness, Efficiency and Equity.

3rd ed. Cheltenham: Nelson Thornes, 2001:524.

5. Cooper AM, O’Malley LA, Elison SN, et al. Primary school-based behavioural interventions for preventing caries. CDSR 2011;10. https://dx.doi. org/10.1002/14651858.cd009378 6. Amalia R, Schaub RM, Widyanti N, et al. The role of school-based dental programme on dental caries experience in Yogyakarta Province, Indonesia. Int J Paediatr Dent 2012;22(3):203-210. https://dx.doi.org/10.1111/j.1365-263x.2011.01177.x 7. Peterson P. Effect of a school-based oral health education programme in Wuhan City, People’s Republic of China. Int Dent J 2004;54(1):33-41. https://dx.doi. org/10.1111/j.1875-595x.2004.tb00250.x 8. Yazdani R, Vehkalahti MM, Nouri M, et al. School-based education to improve oral cleanliness and gingival health in adolescents in Tehran, Iran. Int J Paediatr Dent 2009;19(4):274-281. https://dx.doi.org/10.1111/j.1365-263x.2009.00972.x 9. National Department of Health, South Africa. National Guidelines For The Development of Health Promoting Schools/Sites in South Africa (Draft 4).

Pretoria: NDoH, 2000;1-42 10. National Department of Health. National Oral Health Strategy. Pretoria: NDoH, 2004. http://www.doh.gov.za/docs/index.html (accessed 30 April 2014).

11. Department of Health, KwaZulu-Natal. Annual Report 2011-12. http://www.kznhealth.gov.za/1112report/partA.pdf (accessed 22 August 2013).

12. World Health Organization (WHO). Health Promotion Evaluation: Recommendations to Policymakers. Copenhagen: WHO, 1998.

13. Petersen P, Kwan S. Evaluation of community-based oral health promotion and oral disease prevention - WHO recommendations for improved evidence in public health practice. Community Dent Health 2004;21(Suppl):S319-S329.

14. Lee A, Cheng FF, St Leger L. Evaluating health-promoting schools in Hong Kong: Development of a framework. Health Promot Int 2005;20(2):177-186. https://dx.doi.org/10.1093/heapro/dah607 15. World Health Organization. Concept Paper: Evaluation of Oral Health Promotion Interventions. Geneva: WHO, 2003.

MARCH 2017 Vol. 11 No. 1

RESEARCH 16. Sprod AJ, Anderson A, Treasure ET. Effective Oral Health Promotion: Literature Review. Technical Report 20. Cardiff: Health Promotion Wales, 1996.

17. Nutbeam D. Evaluating health promotion â&#x20AC;&#x201C; progress, problems and solutions.

Health Promot Int 1998;13(1):27-44. https://dx.doi.org/10.1093/heapro/13.1.27 18. Strauss A, Corbin J. Basics of Qualitative Research. 2nd ed. CA: Thousand Oaks, 1998:1-312.

19. Green J, Tones K. Towards a secure evidence base for health promotion. J Pub Health Med 1999;21(2):133-139. https://doi.org/10.1093/pubmed/21.2.133 20. Torrance H. Triangulation, respondent validation, and democratic participation in mixed methods research. J Mix Methods Res 2012;6(2):111-123. https:// dx.doi.org/10.1177/1558689812437185 21. Reddy M, Singh S. Viability in delivering oral health promotion activities within the health-promoting schools initiative in KwaZulu-Natal. S Afr J Child Health 2015;9(3):93-97. https://dx.doi.org/10.7196/sajch.7944 22. Reddy M and Singh S. Dental caries status in six-year-old children in KwaZuluNatal, South Africa. S Afr Dent J 2015;70(9):396-401.

23. Petersen P. Challenges to improvement of oral health in the 21st century - the approach of the WHO Global Oral Health Programme. Int Dent J 2004;54(S6):329-343. https://dx.doi.org/10.1111/j.1875-595x.2004.tb00009.x 24. Ellwood R, Cury J. How much toothpaste should a child under the age of 6 years use? Eur Arch Paediatr Dent 2009;10(3):168-174. https://dx.doi.org/10.1007/ bf03262679 25. Marinho V, Higgins J, Sheiham A, et al. Flouride tips for preventing dental caries in children and adolescents. CDSR, 2003.

26. Petersen PE, Phantumvanit P. Perspectives in the Effective use of fluoride in Asia.

J Dent Res 2012;91(2):119-121. https://dx.doi.org/10.1177/0022034511429347

20

SAJCH

27. World Health Organization (WHO). Oral Health Through Schools. Document 11. Geneva: WHO, 2003:1-64.

28. Saied-Moallemi Z, Virtanen JI, Tehranchi A, et al. School-based intervention to promote preadolescentsâ&#x20AC;&#x2122; gingival health: A community trial. Community Dent Oral Epidemiol 2009;37(6):518-526. https://dx.doi.org/10.1111/j.16000528.2009.00491.x 29. Petersen P, Kwan S. The 7th WHO Global Conference on health promotiontowards integration of oral health (Nairobi, Kenya 2009). Community Dent Health 2010;27(Suppl 1):S129-S136.

30. Swanepoel C. A comparison between the views of teachers in South Africa and six other countries on involvement in school change. S Afr J Educ 2009;29:461-74.

31. Pithpornchaiyakul W, Pitpornchaiyakul S, Thitasomakul S, et al. Toothbrushing activities and related factors among primary schools in Songkhla, Thailand. J Dent Assoc Thailand 2009;59:190-199.

32. Jurgensen N, Peterson PE. Promoting oral health of children through schools - Results from a WHO global survey 2012. Community Dent Health 2013;30:204-218.

33. Department of Health, KwaZulu-Natal. Annual Report 2013/2014 - Vote 7.

Pietermaritzburg: DOH, 2014.

34. Department of Co-Operative Governance and Traditional Affairs Province of KwaZulu-Natal. Provincial Infrastructure and Services Backlogs (Electricity, Water and Sanitation). Pietermaritzburg: Department of Co-Operative Governance and Traditional Affairs Province, KZN, 2013.

35. Curnow M, Pine C, Burnside G, et al. A randomised controlled trial of the efficacy of supervised toothbrushing in high-caries-risk children. Caries Res 2010;36(4):294-300. https://dx.doi.org/10.1159/000063925

MARCH 2017 Vol. 11 No. 1

RESEARCH

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

Serum selenium status of HIV-infected children on care and treatment in Enugu, Nigeria A C Ubesie,1,2 MBBS, MPH, FMCPaed, FWCAP; B C Ibe,1,2 MBBS, FMCPaed, FWCAP; I J Emodi,1,2 MBBS, FMCPaed, FWCAP; K K Iloh,2 MBBS, FMCPaed, FWCAP 1 2

Department of Paediatrics, College of Medicine, University of Nigeria, Nsukka, Nigeria Department of Paediatrics, College of Medicine, University of Nigeria Teaching Hospital, Ituku/Ozalla, Enugu, Nigeria

Corresponding author: A C Ubesie (agozie.ubesie@unn.edu.ng)

Objective. To compare the selenium status of HIV-infected and HIV-uninfected children.

Methods. This was a hospital-based comparative study using a structured questionnaire in the quantitative research domain at the University of Nigeria Teaching Hospital, Ituku/Ozalla, Enugu, Nigeria. Seventy-four HIV-infected children were compared with 74 nonHIV-infected children (35 males and 39 females in each group). The outcome measure was the selenium status of the study participants.

Results. The mean (standard deviation (SD)) weight-for-height z-score among the subjects was –0.18 (1.53) compared with 0.05 (1.68) among the controls (p=0.457). The mean (SD) height-for-age z-score among the subjects was –1.16 (1.44) compared with 0.06 (1.06) among the controls (p<0.001). Eighteen subjects (24.3%) compared with eight controls (11.4%) were selenium deficient (odds ratio 2.49; 95% confidence interval 1.00 - 6.18; p=0.044). Median CD4 counts of selenium-deficient and non-deficient subjects were 765.5 (range 409 1 489) and 694.0 (range 85 - 2 196) cells/μL, respectively (p=0.321). The proportions of selenium deficiency were 26.4% and 22.2% in the highly active antiretroviral therapy (HAART) and pre-HAART groups, respectively (p=0.565).

Conclusion. There was a significant difference in the proportion of HIV-infected children who were selenium deficient compared with their uninfected counterparts.

S Afr J Child Health 2017;11(1):21-25. DOI:10.7196/SAJCH.2017.v11i1.1134

Worldwide, an estimated 2.5 million children under 15 years of age are living with HIV, and more than 2.3 million of them live in subSaharan Africa.[1] Infections and malnutrition have been shown to be associated with increased HIV mortality.[2] HIV-related malnutrition involves both micro- and macronutrient deficiencies.[3] There is a compelling association of micronutrient deficiencies in HIV-infection with immune deficiency, rapid disease progression and mortality.[4] Micronutrient supplements can delay HIV disease progression and reduce mortality in HIV-infected persons not receiving highly-active antiretroviral therapy (HAART).[5] Although the use of HAART has revolutionised the management of HIV infection, micronutrient deficiencies still occur among HIV-infected patients on HAART.[6] The provision of simple, inexpensive micronutrient supplements as an adjunct to HAART may therefore have several cellular and clinical benefits, such as a reduction in mitochondrial toxicity and oxidative stress and an improvement in immune reconstitution.[5] One such micronutrient is selenium, an essential trace element with antioxidant properties.[7] In humans and animals, selenium increases immune function and is required for growth and reproduction.[8] It also exerts antiviral activities by inhibiting reverse transcriptase enzyme in RNA-virus-infected animals.[9,10] Supplemental selenium can potentially prevent the replication of HIV and retard the development of AIDS in newly infected subjects.[9] Ensuring selenium sufficiency among HIVinfected children, especially in settings with a high burden of malnutrition, may improve survival. However, studies that have evaluated the prevalence of selenium deficiency among HIV-infected children are limited. This study set out to evaluate the prevalence of selenium deficiency among HIVinfected children compared with uninfected controls at the University of Nigeria Teaching Hospital (UNTH), Enugu State.

Study area and design

This was a hospital-based comparative study of selenium in HIVinfected and HIV-uninfected children carried out between October 21

SAJCH

2013 and August 2014 at the UNTH. The hospital serves as a referral centre to primary and secondary healthcare facilities within and outside south-eastern Nigeria. It is among the first generation of tertiary hospital facilities in the country. HIV-infected children are managed at the paediatric HIV clinic, which runs once a week.

Study population

The subjects were HIV-infected children aged 6 - 180 months (15 years) enrolled between October 2013 and February 2014. The control group was non-HIV-infected children, matched for age, sex and socioeconomic status, who were recruited from the children’s outpatient clinic of the teaching hospital. Socioeconomic index scores were assigned to the occupations and educational attainments of the parents or caregivers of subjects and controls using the Oyedeji socioeconomic classification scheme, which grades subjects from I to V.[11] The socioeconomic classification for each study participant was obtained by finding the mean score of his or her parents. If either of the parents was dead, the score of the surviving parent was used, and if both parents were dead, the score of the caregiver was used. Classes I and II were regarded as upper social class, III as middle and IV and V as lower social class.

Inclusion criteria

The subjects were confirmed HIV-infected children aged 6 - 180 months (15 years), and the controls were HIV-uninfected children on follow-up visits at the children’s outpatient clinic matched for age, sex and socioeconomic status with the subjects.

Exclusion criteria

Children aged ≥7 years who refused to assent to the study or whose caregivers refused consent, as well as those with a history of micronutrient supplementation in the past 3 months, were excluded.

MARCH 2017 Vol. 11 No. 1

RESEARCH Consent

Thumb-printed and/or signed informed consent was obtained from the parents or caregivers, while assent was obtained from study participants aged ≥7 years.

Recruitment of study participants

Children who met the inclusion criteria were enrolled consecutively until the desired sample size for the subjects and controls was achieved. Seventy-four HIV-infected children served as the subjects, while nonHIV-infected children matched for age, sex and socioeconomic status served as the controls. A structured questionnaire was used to collect the following data from the subjects and controls: age in months, date of birth, date of interview, and the highest educational attainment and occupation of parents or caregivers. Data on HAART regimen and duration of treatment were retrieved from the medical records of the HIV-infected children. The controls were screened for the presence of HIV antibodies using the national algorithm for HIV testing.

The study participants were examined for the presence of any clinical signs of illness. An infant weighing scale (Hospibrand ZT-120, UK) was used to measure the weights of children under 2 years to the nearest 0.1 kg, while a standing scale was used for children aged 2 years and above. Weight measuring instruments were set to zero point before use and standardised at weekly intervals using known weights. Length was measured using an infantometer (Seca, Germany) for children under 2 years while height was measured for children 2 years and above using a stadiometer to the nearest 0.1 cm.

the Safranin O method. The following are recommended normal selenium levels:[14,15] • <18 months: 30 - 50 μg/L (0.38 - 0.63 μmol/L) • 18 months - 4 years: 45 - 90 μg/L (0.57 1.14 μmol/L) • 5 - 16 years: 55 - 115 μg/L (0.70 - 1.46 μmol/L) • adults (>16 years): 70 - 130 μg/L (0.89 1.65 μmol).

between CD4 count/selenium levels and independent variables. The odds ratio (ORs) of selenium deficiency between subjects and controls was calculated, and 95% confidence interval (CIs) reported. All analyses were done at the 5% level of significance and p<0.05 was considered statistically significant.

Ethical approval

The hospital’s Health Research and Ethics Committee approved the study (ref. no.

NHREC/05/01/2008B-FWA00002458IRB00002323).

In this study, therefore, selenium deficiency was defined as follows: • <18 months: <30 μg/L (0.38 μmol/L) • 18 months - 4 years: <45 μg/L (0.57 μmol/L) • 5 - 15 years: <55 μg/L (0.70 μmol/L).

Results

Study participants

Data analysis

One hundred and forty-eight participants (74 subjects, 74 controls) were included in the study. The sociodemographic characteristics of the study population are shown in Table 1.

A blood sample for selenium analysis was available in 74 subjects and 70 controls. The median ages of the subjects and controls were 94.8 and 84.0 months, respectively (range 7 180 months). Forty (54.1%) of the subjects and 37 (50.0%) of the controls were from the middle social class (p=0.777).

Data analysis was carried out using the Statistical Package for Social Sciences (SPSS) version 19.0 (IBM Corp., USA). The χ2 and Fisher’s exact tests were used to test for the significant association of categorical variables. Fisher’s exact test was used if the expected number in a cell of a two-by-two table was less than five, and Yates’ correction if a cell contained zero. The quantitative data were tested for normality using the ShapiroWilk normality test. A Student t-test was used to compare the mean WAZ and HAZ between the subjects and controls. MannWhitney U- and Kruskal-Wallis tests were used to test for a significant association

CD4 counts and antiretroviral (ARV) regimen

The median CD4 count of the subjects was 741.5 (interquartile range (IQR) 472.0 - 1

Table 1. Sociodemographic characteristics of the study population Sociodemographic characteristics

Subjects, n (%)

Controls, n (%)

χ2

df

p-value

0.5 - 5.0

25 (33.8)

24 (32.4)

0.147

2

0.923

5.1 - 10.0

32 (43.2)

31 (41.9)

10.1 - 15.0

17 (23.0)

19 (25.7)

Male

35 (47.3)

35 (47.3)

0.0

1

1.0

Female

39 (52.7)

39 (52.7)

Upper

10 (13.5)

13 (17.6)

0.508

2

0.777

Middle

40 (54.1)

37 (50.0)

Lower

24 (32.4)

24 (32.4)

Age group (years)

Nutritional assessment

Height-for-age z-score (HAZ), weight-forheight z-score (WHZ) and body mass indexfor-age z-score (BMIZ) values were calculated using the 2005 World Health Organization (WHO) AnthroPlus version 1.0.4 software calculator (Switzerland).[12] Acute malnutrition (wasting) was defined as WHZ and BMI z-scores ≤–2 while chronic malnutrition (stunting) was defined as HAZ ≤–2.[13]

Sex

Socioeconomic class

Laboratory tests

Two aliquots of 5 mL of blood were collected from the antecubital fossa of the subjects.

The first aliquot for CD4 estimation was collected in an ethylenediaminetetraacetic acid (EDTA) bottle. The second aliquot for selenium was collected in plain bottles.

Similarly, a 5 mL aliquot of blood was collected in plain bottles from the antecubital fossa of controls for selenium estimations.

The blood samples for CD4 estimation were analysed using the Partec CyFlow machine (Germany). Serum selenium was measured by the spectrophotometric method using

Table 2. Comparison of the mean anthropometric parameters of the subjects and controls

22

Anthropometric parameters

Subjects, mean (SD)

Controls, mean (SD)

t-value

p-value

Weight (kg)

24.3 (9.6)

26.1 (13.8)

­­–0.91

0.364

Height (cm)

118.4 (21.6)

121.5 (25.0)

–0.80

0.426

WHZ

0.2 (1.5)

–0.1 (1.7)

0.718

0.457

HAZ

–1.2 (1.4)

0.1 (1.1)

–5.677

<0.001

BMIZ

–1.5 (15.3)

–0.2 (1.9)

–0.706

0.481

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH 202.8 cells/μL. Fifty-six of the 74 subjects (75.7%) were on HAART, while 10 (17.9%) of the 56 children on HAART had been switched to a second-line regimen. Zidovudine (AZT), Lamivudine (3TC) and Nevirapine (NVP) were the first-line combination therapy in 46 (82.1%) of the 56 children on HAART.

Table 3. Distribution of selenium-deficient study participants by sex HIV-infected (n=74)*

Serum selenium levels of study participants

The mean (SD) serum level of selenium among the subjects was 82.1 (6.3) µg/L compared with 97.3 (7.1) µg/L among the controls (p=0.11). The median serum selenium levels for subjects and controls were 76.9 (IQR 50.8 - 98.1) µg/L and 83.30 (IQR 61.6 - 113.8) µg/L (p=0.085).

Selenium deficiency

Eighteen (24.3%) of the 74 subjects compared with eight (11.4%) of 70 controls were selenium deficient (p=0.044). The OR of selenium deficiency among the subjects was 2.49 times higher among the subjects than the controls (OR 2.49; 95% CI 1.00 6.18; z=1.9; p=0.049).

Age and serum selenium levels

The median ages of selenium-deficient and non-deficient study participants were 92.4 (IQR 62.4 - 123.6) and 90 (IQR 49.2 - 124.8) months, respectively (p=0.728). Seven (28.0%) of 25 subjects aged 6 - 60 months were selenium deficient (p=0.863). Six (20.7%) of 29 children aged 61 - 120 months were selenium deficient among the controls (p=0.093).

Sex and serum selenium levels

Twelve (34.3%) of 35 males compared with 6 (15.4%) of 39 females were selenium deficient among the subjects (p=0.058). Six (18.7%) of 32 males compared with 2 (5.3%) of 38 females were selenium-deficient among the controls (p=0.077) (Table 3).

Risk factors for selenium deficiency

Nutritional status One (20%) of the five acutely malnourished subjects was selenium deficient, compared

Yes (n %)

No (n %)

Yes (n %)

No (n %)

12 (34.3)

23 (65.7)

6 (18.7)

26 (81.3)

6 (15.4)

33 (84.6)

2 (5.3)

36 (94.7)

18 (22.5)

52 (77.5)

8 (11.4)

62 (88.6)

Sex Male

Nutritional status

The mean (standard deviation (SD)) weight and height of the subjects and controls were 24.3 (9.5) v. 26.1 (13.8) kg, and 118.4 (21.6) v. 121.5 (25.0) cm, respectively (Table 2). The mean WHZ among the subjects was –0.2 (1.5) compared with –0.1 (1.7) among the controls (p=0.457). The mean HAZ among the subjects were –1.2 (1.4) compared with 0.1 (1.1) among the controls (p<0.001) (Table 2).

Five (6.8%) of the 74 subjects and 5 (6.8%) of the 74 controls had acute malnutrition (Fisher’s exact test, p=1). Conversely, 18 (24.3%) subjects compared with none of the controls had chronic malnutrition (p<0.001).

HIV-uninfected (n=70)†

Female Total 2

*(χ =3.58, df=1, p=0.058) †

2

(χ =3.12, df=1, p=0.077)

with 17 (24.6%) of 69 without acute malnu trition (Fisher’s exact test, p=1.0). The mean WHZ and BMIZ among the seleniumdeficient subjects were 0.2 (0.8) and –6.9 (31.0), respectively, compared with –0.3 (1.4) and 0.2 (1.3) among non-deficient subjects.

Four (22.2%) of the 18 subjects with chronic malnutrition compared with 14 (25%) of 56 subjects without chronic malnutrition were selenium deficient (Fisher’s exact test, p=1).

The mean HAZ among selenium-deficient subjects was –1.06 (1.61), compared with –1.09 (1.36) among the non-selenium-deficient subjects (p=0.943).

CD4 and HAART The median CD4 counts of the 18 seleniumdeficient and 56 non-deficient subjects were 765.5 (range 409 - 1 489) and 694.0 (range 85 - 2 196) cells/μL of blood, respectively (p=0.321). Fourteen (26.4%) of 53 subjects on HAART compared with four (22.2%) of 21 ART-naive subjects and eight (11.4%) of 70 controls were selenium deficient (p=0.086).

There was no statistically significant difference between HAART and pre-HAART subjects who were selenium deficient (p=0.565). The mean (SD) duration of HAART for seleniumdeficient subjects was 41.1 (30.8) months, compared with 43.4 (29.9) months among selenium-sufficient subjects (p=0.807).

Discussion

This study showed that a significantly higher proportion of HIV-infected children compared with their non-infected counterparts were selenium deficient. The OR of selenium deficiency was 2.5 times higher among the subjects than the controls. Lower serum selenium levels among HIV-infected indivi duals have been linked to excessive utilisation of selenoproteins by the virus.[16] It is this increased utilisation of the selenoproteins in HIV-infected individuals that results in selenium depletion. A study in Ife, south-west Nigeria, also reported a significantly higher rate of selenium deficiency among HIV-infected children compared with the uninfected controls who were matched for age and sex with the subjects.[17] However, the reported 23

SAJCH

MARCH 2017 Vol. 11 No. 1

rates of selenium deficiency among the subjects and controls were higher than the findings in the present study. The group in the Ife, Nigeria, study included only ARV-naive subjects, and this may explain their reported higher proportion of selenium deficiency among HIVinfected children. The present study included both ARV-naive subjects and subjects on HAART. It is possible that HAART would have slowed down viral replications, reduced the need for selenoprotein synthesis and ultimately lowered the proportion of selenium-deficient subjects.

In contrast to the present study, Henderson et al.[18] reported lower selenium deficiency rates among 38 HIV-infected and -uninfected children in the USA. The lower rates found

by Henderson et al.[18] can be explained by their small sample size of 38 subjects (28 HIV-infected and 10 HIV-uninfected), which limits the generalisability of their findings.

The present study had a larger sample size of 148. Additionally, Henderson et al. [18] conducted their study in a country where it has been shown that adequate amounts of selenium are consumed, with an average daily intake from foods for those aged ≥2 years reaching 108.5 µg. Bunupuradah et al.[20] reported no deficiency in baseline selenium levels among 141 HIV-infected Thai children aged 1 - 12 years. The reason for their reported zero prevalence can be explained by their very low cut-off definition for selenium deficiency (<0.1 μmol/L or 8 µg/L) in these children.

Age and sex

The difference in the proportion of selenium deficiency among the three age groups of subjects in this study was not statistically significant. This agrees with the findings of Kouna et al.[21] in their study of 318 children aged 7 - 10 years. A possible explanation for this lack of significant difference may be that a wide variety of foods are rich in selenium, including seafoods, organ meats, grains and dairy products, which are consumed across age groups.[17,24] The findings of the northeast Thailand study by Krittaphol et al.,[22] however, disagree with those of the present

RESEARCH study. The authors reported that children under 9 years of age had a significantly lower mean serum selenium concentration than those over 9 years of age. The basis for dichotomising the children into under and over 9 by the authors was rather arbitrary and unclear.

Among the subjects and controls, the proportion of selenium deficiency was not significantly different between the sexes. There are conflicting reports in the literature regarding selenium levels and sex. Jones et al.[23] and Kouna et al.[21] reported no significant difference in the proportion of selenium deficiency between the sexes, which finding agrees with the present study. Rousseau et al.,[24] however, reported significantly lower selenium in males than females among 30 HIV-infected individuals. Safaralizadeh[25] also reported that the mean serum selenium levels were significantly lower in male than female children aged 1 - 16 years. Studies reporting lower selenium among males have failed to offer plausible explanations for the disparity. In contrast, Amare et al.[26] reported a higher rate of selenium deficiency in females than males. Krittaphol et al.[22] also reported that females had a significantly lower mean serum selenium concentration than males in their sample. Studies reporting lower selenium values among females have postulated a sex-linked hormonal influence onthe serum level of selenium.[25,26]

Nutritional status

There was no significant difference in the proportion of subjects with acute and chronic malnutrition who were selenium deficient. Nhien et al.[27] reported no significant difference in serum concentration of selenium with regard to underweight, stunting and wasting. Amare et al.[28] also reported no significant correlation between the levels of selenium and the anthropometric variables of schoolchildren. These findings suggest that selenium status in children is independent of their macronutrient status. This finding is not surprising because micronutrient deficiencies, or hidden hunger, remain a public health challenge among apparently healthy children, especially in developing countries.

Immunological status and HAART

There was no significant difference in the median CD4 count of selenium-deficient and non-deficient subjects in this study.

Although Anyabolu et al.[17] in the Ife study reported higher mean serum selenium levels in subjects with a CD4 count of ≥350 cells/ µL compared with those with <350 cells/µL , they failed to explain the rationale for grouping these children based on the CD4 counts of ≥350 and <350 cells/μL. The difference in the approach to statistical analysis of CD4 and serum selenium between their study and the present study makes it difficult to draw a meaningful comparison between the two studies.

The median serum selenium level among subjects on HAART did not differ significantly from that of those who were yet to commence HAART. There was also no significant difference in the proportion of subjects on HAART who were selenium deficient compared with the pre-ART subjects in this study. Similarly, when the proportion of selenium deficiency in the three groups of HAART, pre-HAART and controls was compared, no significant difference was found. To the best of the authors’ knowledge, this is the first study that compares selenium status between HIV-infected children on HAART and those who are not. Akinola et al.[6] in a study involving HIV-infected adults reported no significant difference in the serum levels of selenium of their subjects on HAART and those who were yet to commence HAART, which agrees with the finding of the present study.

Study limitations

The study did not assess dietary intakes of selenium among the study participants. The exclusion of study participants based on a history of micronutrient supplements was inadequate, since there could have been recall bias.

24

SAJCH

Conclusion

Although selenium deficiency was significantly higher among the subjects than controls, no significant difference was noted between the sexes. Nutritional status, CD4 cell count and use of HAART were not significantly associated with selenium levels in this study.

1. Heeren GA, Jemmott JB, Sidloyi L, Ngwale Z, Tyler JC. Disclosure of HIV diagnosis to HIV-infected children in South Africa: Focus groups for intervention development. Vulnerable Child Youth Study 2012;7(1):47-54. https://doi.org/10.1080/17450128.2012.656733 2. Munyagwa M, Baisley K, Levin J, Brian M, Grosskurth H, Maher D. Mortality of HIV infected and uninfected children in a longitudinal cohort in rural southwest Uganda during 8 years of follow-up. Trop Med Int Health 2012;17(7):836843. https://doi.org/10.1111/j.1365-3156.2012.03000.x 3. Ndeezi G, Tyllerskar T, Ndugwa CM, Tumwine JK. Effect of multiple micronutrient supplementation on survival of HIV-infected children in Uganda: A randomized, controlled trial. J Int AIDS Soc 2010;13:1-9. https:// doi.org/10.1186/1758-2652-13-18 4. Singhal N. A clinical review of micronutrients in HIV infection. J Int Assoc Provid AIDS Care 2002;1(2):63-75. https://doi.org/10.1089/apc.1998.12.249 5. Drain PK, Kupka R, Mugusi F, Fawzi WW. Micronutrients in HIV-positive persons receiving highly active antiretroviral therapy. Am J Clin Nutr 2007;85(2):333-345. https://doi.org/10.1097/qad.0b013e32826fb6c7 6. Akinola FF, Akinjinmi AA, Oguntibeju OO. Effect of combined antiretroviral therapy on selected trace elements and CD4+ T-cell count in HIV-positive persons in an African setting. J AIDS Clinic Res 2012;3(10):185. https://doi. org/10.4172/2155-6113.1000185 7. Townsend A, Featherstone A, Chéry CC, Vanhaecke F, Kirby J, Krikowa F. Increased selenium concentration in seronorm trace elements serum (level 2).

Clin Chem 2004;50(8):1481-1482. https://doi.org/10.1373/clinchem.2004.034579 8. Fan AM, Kizer KW. Selenium: Nutritional, toxicologic, and clinical aspects.

Western J Med 1990;153(2):160-167.

9. Schrauzer GN, Sacher J. Selenium in the maintenance and therapy of HIVinfected patients. Chem Biol Interact 1994;91(2-3):199-205. https://doi. org/10.1016/0009-2797(94)90040-x 10. Stone CA, Kawai K, Kupka R, Fawzi WW. The role of selenium in HIV infection. Nutr Rev 2010;68(11):671-681. https://doi.org/10.1111/j.17534887.2010.00337.x 11. Oyedeji GA. Socio-economic and cultural background of hospitalized children in Ilesha. Nig J Paediatr 1985;12(4):111-117.

12. World Health Organization. WHO Anthro (version 3.2.2) and macros. Geneva: WHO, 2011. http://www.who.int/childgrowth/software/en/ (accessed 26 February 2013).

13. World Health Organization Multicentre Growth Reference Study Group. WHO Child Growth Standards: Length/height-for-age, Weight-for-age, Weightfor-length, Weight-for-height and Body Mass index-for-age: Methods and Development. Geneva: WHO, 2006.

14. Supra-regional Assay Service Trace Element Laboratories. In: Walker AW, ed. Clinical and Analytical Handbook. 4th ed. Guildford, UK: Department of Clinical Biochemistry, Royal Surrey County Hospital, 2006 15. National Health Service Trust. Recommended protocol for monitoring copper, zinc and selenium. 2011. http://www.cityassays.org.uk/downloads/ CopperZincMagnesium%201.01.pdf (accessed 12 November 2013).

16. Taylor EW, Cox AG, Zhao L, et al. Nutrition, HIV, and drug abuse: The molecular basis of unique role for selenium. J Acquir Immune Defic Syndr 2000;25(1):S53-S61.

17. Anyabolu HC, Adejuyigbe EA, Adeodu OO. Serum micronutrient status of HAART-naive, HIV-infected children in south-western Nigeria: A case controlled study. AIDS Res Treat 2014;2014:351043. https://doi. org/10.1155/2014/351043 18. Henderson RA, Talusan K, Hutton N, Yolken RH, Caballero B. Serum and plasma markers of nutritional status in children infected with the human immunodeficiency virus. J Am Diet Assoc 1997;97(12):1377-1381. https://doi. org/10.1016/s0002-8223(97)00333-7 19. National Institutes of Health. Selenium: Dietary supplement fact sheets. 2013. http://ods.od.nih.gov/factsheets/Selenium-HealthProfessional/ (accessed 7 December 2014).

20. Bunupuradah T, Ubolyam S, Hansudewechakul R, Kosalaraksa P, Ngampiyaskul C, Kanjanavanit S. Correlation of selenium and zinc levels to antiretroviral treatment outcomes in Thai HIV-infected children without severe HIV symptoms. Eur J Clin Nutr 2012;66(8):900-905. https://doi.org/10.1038/ ejcn.2012.57 21. Kouna P, Mashavave G, Kandawasvika GQ, et al. Serum selenium levels and nutritional status of school children from an HIV prevention programme in Zimbabwe. J Trop Dis Pub Health 2014;2:134. https://doi.org/10.4172/2329891X.1000134 22. Krittaphol W, Bailey KB, Pongcharoen T, Winichagoon P, Thomson C, Gibson RS. Primary school children from northeast Thailand are not at risk of selenium deficiency. Asia Pac J Clin Nutr 2006;15(4):474-481. https://www. ncbi.nlm.nih.gov/pubmed/17077062 (accessed July 4, 2013).

23. Jones CY, Tang AM, Forrester JE, et al. Micronutrient levels and HIV disease status in HIV-infected patients on highly active antiretroviral therapy

MARCH 2017 Vol. 11 No. 1

RESEARCH in the nutrition for healthy living cohort. J Acquir Immune Defic Syndr 2006;43(4):475-482. https://doi.org/10.1097/01.qai.0000243096.27029.fe 24. Rousseau MC, Molines C, Moreau J, Delmont J. Influence of highly active antiretroviral therapy on micronutrient profiles in HIV-infected patients. Ann Nutr Metab 2000;44(5-6):212-216. https://doi.org/10.1159/000046686 25. Safaralizadeh R, Kardar GA, Pourpak Z, Moin M, Zare A, Teimourian S. Serum concentration of selenium in healthy individuals living in Tehran. Nutr J 2005;4:32. http://www.nutritionj.com/content/4/1/32 (accessed 13 May 2013).

25

SAJCH

26. Amare B, Tafess K, Ota F, et al. Serum concentration of selenium in diarrheic patients with and without HIV/AIDS in Gondar, Northwest Ethiopia. J AIDS Clin Res 2011;2:128. https://doi.org/10.4172/2155-6113.1000128 27. Nhien NV, Khan NC, Ninh NX, et al. Micronutrient deficiencies and anemia among preschool children in rural Vietnam. Asia Pac J Clin Nutr 2008;17(1):48-55.

28. Amare B, Moges B, Fantahun B, et al. Micronutrient levels and nutritional status of school children living in Northwest Ethiopia. Nutr J 2012;11:108. https://doi. org/10.1186/1475-2891-11-108

MARCH 2017 Vol. 11 No. 1

RESEARCH

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

Individual v. community-level measures of women’s decision-making involvement and child survival in Nigeria J O Akinyemi,1,2 PhD; S A Adedini,1,3 PhD; C O Odimegwu,1 PhD Demography and Population Studies Programme, University of the Witwatersrand, Johannesburg, South Africa Department of Epidemiology and Medical Statistics, Faculty of Public Health, University of Ibadan, Ibadan, Nigeria 3 Department of Demography and Social Statistics, Obafemi Awolowo University, Ile Ife, Nigeria 1 2

Corresponding author: J O Akinyemi (joshua.akinyemi@wits.ac.za) Background. Although decision-making authority is associated with maternal healthcare utilisation, the evidence on the relative importance of individual-level v. community-level decision-making participation for child survival in sub-Saharan Africa is limited.

Objectives. To assess the net effects of individual- and community-level measures of decision-making involvement (DMI) on under-5 mortality in Nigeria.

Methods. Data on a nationally representative sample of 31 482 children in the 2013 Nigeria Demographic and Health Survey were analysed.

Mothers who reported involvement in decision-making on own healthcare, major household purchases and visits to friends and relatives were categorised as having high DMI. Community-level measures of DMI were derived by aggregating the individual measures at the cluster level. Kaplan-Meier estimates of childhood mortality rates were computed. Multilevel discrete-time hazard models were employed to investigate the net effect of individual- and community-level DMI on childhood mortality.

Results. Childhood mortality, at 59 months, was higher among children of women with low DMI (120 per 1 000) compared with those with high DMI (84 per 1 000). The full multilevel model showed that there was no difference in the risk of childhood death between children whose mothers had high v. low DMI (hazard ratio (HR) 1.01, CI 0.90 - 1.12). However, mortality risk was found to be lower among children in communities with medium DMI (HR 0.84, CI 0.74 - 0.96). Maternal age at child’s birth, education, household wealth index and preceding birth interval were significantly associated with under-five mortality.

Conclusion. Besides socioeconomic and biodemographic characteristics, community- and not individual-level DMI was associated with under-5 mortality. Women’s empowerment programmes targeting maternal and child health outcomes should also focus on communities.

S Afr J Child Health 2017;11(1):26-32. DOI:10.7196/SAJCH.2017.v11i1.1148

Due to the difficulty in empirical measurement of women’s empower ment, most earlier studies have used proxy measures such as level of education and proportion (of women) who owned land. [1] Undoubtedly, such indicators reflect socioeconomic status more than empowerment, although the two are positively related. For operational/practical purposes, in this study, empowerment refers to the ability to make choices or take decisions.[2] Subsequent to developing a framework for relationships between sex, gender, population/health/nutrition outcomes, the Demographic and Health Survey (DHS) programme introduced four sets of genderrelated questions into the women’s core questionnaire in 1999.[3] These questions captured information on women’s participation in household decision-making, gender-related hurdles in accessing healthcare and women’s acceptance of norms that justify men’s control over women. Consequently, relationships between these decisionmaking (empowerment) indicators and reproductive health concepts (such as fertility intention, use of contraceptives and maternal health services utilisation) have been explored in some developing countries.[4,5] Studies have shown that women’s empowerment is positively associated with contraceptive use despite control for education level and other sociodemographic characteristics.[4] Other studies have investigated the link between women’s empowerment and use of antenatal care (ANC), skilled/facility delivery and postnatal care services. However, the eveidence was mixed and depended on the type of maternal healthcare service under investigation. The 26

SAJCH

common pattern was that women’s empowerment increased the uptake of ANC and postnatal services.[6-8] However, the influence of women’s empowerment on skilled delivery care was weak in Ethiopia/ Eriteria[9] and Nairobi, Kenya[10] but significant in Nigeria. [11,12] Aside from maternal healthcare utilisation, Adhikari et al.,[13] and Desai and Johnson[14] reported that women’s empowerment was also a significant independent predictor of child mortality. In both studies, infants born to women with decision-making authority were less likely to die.

Apart from the fact that fewer studies have been devoted to child health outcomes, the results also depend on the measures of empowerment. Furthermore, a large proportion of previous studies investigated empowerment at the level of individual women rather than at community level. Women’s empowerment at the community level refers to the prevailing community norm with respect to the ability of women to make decisions on matters concerning them, their children and/or their households. Context or the place where a woman lives plays a significant role in her ability to take decisions. If the general belief or attitude in a community is that a woman is free to take certain decisions, women in such communities would enjoy more decision-making autonomy than those in other communities with opposing views.

Meanwhile, there is increasing evidence of the role of context or community as far as maternal and child health outcomes are concerned.[6,15,16] These studies showed that contextual variables such as residence (rural v. urban), community level of education,

MARCH 2017 Vol. 11 No. 1

RESEARCH poverty, prenatal care and ethnic diversity were associated with utilisation of prenatal care, skilled attendance at delivery, postnatal care and childhood mortality. Beyond these factors there is a need to establish the relative importance of individual- v. communitylevel effects of women’s decision-making involvement (DMI) on child health out­comes.

Nigeria has the highest childhood mortality rate in sub-Saharan Africa. Alhough progress has been made in recent years, the slow pace has been attributed to poor progress in child survival interventions.[17] Interventions aimed at promoting child survival are targeted at mothers who are the primary caregivers of their children. The uptake or adoption of modern child healthcare practices and any survival intervention will depend on women’s decision-making autonomy. Therefore, this study was aimed at investigating the relationship between maternal DMI and childhood mortality in Nigeria. The study distinguished between individual- and community-level measures of decision-making and also determined their net effects while controlling for other background characteristics.

was a child’s death, which was treated as an event history outcome. Survival time was age at death, while children who were alive by the survey date were censored at their current age. There were two primary independent variables: 1.

Individual-level maternal DMI: This was an indicator variable coded as 1 (high involvement) or 0 (low involvement). It was coded 1 if a woman participated in all of the following decisions: own healthcare, major household purchases and visits to family or relatives. Decision on own healthcare refers to the choice of where to go and what to do for any health problem.

Major household purchases implied the purchase of items other than those for daily needs in the households. Decisions on visits to family/friends reflect a woman’s freedom of movement and association.

2. Community-level DMI: This captures the community norms about women’s DMI. The variable was created from individuallevel DMI by aggregating at the cluster (community) level. We first obtained the

proportion of women in each community who participate in decision-making using the egen command in Stata (StataCorp, USA). The proportion was then divided into tertiles categorised as low, medium and high. To reduce the problem of multicollinearity between the individual and community measures of DMI, community measures were computed for all women in the cluster irrespective of whether they had children or not.

Other individual- and community-level var­ iables that are known from the literature to correlate with women’s DMI and child mortality were controlled in the analyses (Table 1).

Household wealth index was calculated from the assets possessed by each household using principal component analysis.[19] Apart from region and place of residence, other community-level variables were community level of education and poverty. Derivation of community-level education and poverty followed the same procedure used for the community-level DMI.

Table 1. Definitions of variables

Methods

Variable

The data analysed for this study were extracted from the children’s recode data file of the 2013 Nigeria Demographic and Health Survey (NDHS). The NDHS 2013 is the fifth round of a national survey conducted to monitor population and reproductive health among Nigerians.

For administrative purposes, Nigeria is divided into 36 states and a Federal Capital Territory (FCT). Details of the sample design and survey findings are available in the published report.[18] Specifically, in the NDHS 2013, a stratified three-stage cluster design was used for the selection of respondents. The primary sampling units, referred to as clusters (communities) in this study were enumeration areas selected from a sampling frame prepared based on the 2006 Nigeria population and housing census. With a fixed sample intake of 45 households per 904 clusters (rural = 532, urban = 372), a total of 40 680 households were selected and 38 948 women aged 15 - 49 years were interviewed successfully. Data for 31 482 children nested in 896 clusters, and born within 5 years before the survey, were analysed. These data were based on the reproductive history of the women. As such, they contained variables on women’s characteristics as well as detailed information on the birth, sex and survival status of each under-five child.

Description

Community-level variables Residence

Type of residence (urban - 1; rural - 2). Reference category was urban.

Region

Region of residence. Each of the 36 states and FCT in Nigeria are grouped into 6 geopolitical regions as follows: North Central - 1; North East - 2; North West - 3; South East - 4; South - 5; South West - 6. South West was used as the reference.

Community poverty level

Proportion of poor households in the community. This was divided into tertiles and categorised as low, medium or high.

Community-level education

Proportion of women with secondary or higher education in the community. Also divided into tertiles and categorised as low, medium or high.

Community-level DMI

Proportion of women in the community who were involved in decision-making on own health, large household purchases and visits to friends/relatives. It was divided into tertiles and categorised as low, medium or high.

Individual-level variables

Variables and measurement

Maternal age at child’s birth

Age of mother at birth of the child (in years). It was categorised as: <20 (reference); 20 - 35; and >35 years.

Maternal education

The highest level of formal education attained by the mother, grouped into: none (reference), primary, secondary and higher.

Wealth quintile

This is the household wealth index. It was divided into 5 quintiles (poorest, poor, middle, rich, and richest but was recoded as poor (reference), middle and rich.

Religion

Religion of the respondent categorised into three groups: Christianity (reference), Islam and Others.

Preceding birth interval

This was the interval between the index and preceding birth. The following categories were used: first birth, ≤24 months, 25 - 35 months and ≥36 months (reference).

Children constituted the unit of analysis in this study. Therefore, the dependent variable 27

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH Statistical analysis

Frequency distribution was generated to summarise the variables, while the KaplanMeier (KM) method was employed to estimate childhood mortality rates. For multivariate analysis, multilevel discrete-time hazard models were fitted. A distinction was made between individual- and community-level effects of DMI on childhood mortality. Apart from the investigation of community-level effects, this modelling approach also ensured that standard errors were appropriately estimated to account for the hierarchical structure of the data. In order to implement the

discrete-time hazard model, each record was transformed into person-time records with each new record (in the transformed dataset) corresponding to a unit of time. For this study, a two-level structure was considered: children (level 1) nested in communities (level 2). The model equation is as follows: Logit(htij) = αt + β1kX1kij + β2kX2kj + uj + eij where: htij = hazards (risk) of death by child i in cluster j αt = baseline hazard of death at time t β1k = coefficients for the individual-level variables

Table 2. Background characteristics of under-5 children and survival status, NDHS 2013

Xkij = individual-level covariates (educa­ tion, age group, wealth index, etc.) for child i in community j β2k = coefficients for the community-level variables X2kj = c ommunity-level covariates (i.e. community-level DMI, residence, region, etc.) uj = community-level random effect eij = error terms for the individual-level model.

In essence, two-level random intercept discrete-time hazards models were fitted, with the assumption that individual-level covariates have similar effects across all communities. Error terms were assumed to be normally distributed with zero mean and constant variance at both individual (σe2) and community (σu2) levels. Fixed effects of individual-level variables were presented as hazard ratio (HR) with 95% confidence interval (CI). Random effects that represent the community effects were summarised using the variance (and standard errors) from which the intracluster correlation (ICC) was estimated. The ICC captures the extent to which children’s deaths are correlated in the community or the extent to which children in the same community are exposed to the same characteristics associated with the risk of death.

All children, n (%)

Deceased children, n (%)

Own healthcare

10 874 (34.5)

812 (7.5)

Large household purchases

10 683 (33.9)

811 (7.6)

Visit to relatives/friends

13 566 (43.1)

1 083 (8.0)

High

8 595 (27.3)

637 (9.8)

Low

22 887 (72.7)

2 249 (7.4)

<20

3 840 (12.2)

467 (12.2)

20 - 35

22 126 (70.3)

1 874 (8.5)

>35

5 516 (17.5)

545 (9.9)

None

14 762 (46.9)

1 657 (11.2)

Primary

6 432 (20.4)

596 (9.3)

Secondary/higher

10 288 (32.7)

633 (6.2)

Poor

14 462 (46.0)

1 722 (11.9)

Middle

6 272 (19.9)

502 (8.0)

Rich

10 748 (34.1)

662 (6.2)

Christianity

12 654 (40.2)

957 (7.6)

Islam

18 354 (58.3)

1 887 (10.3)

Models were fitted in four stages in order to explore the influence of individualand community-level DMI on childhood mortality. First, the effect of each covariate was assessed using univariate models with one independent variable at a time. The covariates in model I were individual- and communitylevel DMI. Model II combined model I with other individual-level covariates while model III consisted of model I and communitylevel variables. In model IV, all variables were included in order to adjust for confounding variables and determine the net effect of the key explanatory variables. All analyses were done using Stata SE version 12.0.

Others

474 (1.5)

42 (8.9)

Ethical considerations

1

6 109 (19.4)

581 (9.5)

2-3

10 074 (32.0)

791 (7.9)

4-5

7 380 (23.4)

610 (8.3)

≥6

7 919 (25.2)

904 (11.4)

First birth

6 181 (19.6)

607 (9.8)

≤24 months

6 668 (21.2)

891 (13.4)

25 - 36 months

9 572 (30.4)

815 (8.5)

>36 months

9 061 (28.8)

573 (6.3)

Variable Maternal DMI

Overall maternal DMI

Maternal age at child's birth (years)

Maternal education

Household wealth index

Maternal religion

Birth order

Preceding birth interval

28

SAJCH

MARCH 2017 Vol. 11 No. 1

Modelling procedure

This study involved a secondary analysis of anonymised data from the NDHS 2013. The survey was approved by the National Health Research Ethics Committee in Nigeria (ref. no. NHREC/01/01/2007). Informed consent was obtained from respondents during the data collection process. Formal approval to use the data was obtained from the NDHS programme.

Results

Table 2 shows the frequency distribution of individual-level characteristics of all the children, including those who have died.

RESEARCH The proportions of children whose mothers were involved in decision-making on their healthcare, large household purchases and visits to friends/relatives were 34.5%, 33.9% and 43.1%, respectively. Overall, 27.3% of 31 482 children had mothers who were involved in decision-making. Close to half (46.9%) belonged to mothers with no formal education, while 32.7% of mothers attained secondary or higher education. In terms of household wealth index and religion, 46.0% of children lived in poor households while 58.3% had Muslim mothers. First births constituted 19.4% among the under-fives.

Distribution of the preceding birth interval revealed that 21.2% were born <24 months after a previous birth.

Table 3 shows that 41.3% and 29.5% of children lived in communities with low and high maternal DMI, respectively. Similarly, the largest proportion of children (43.3%) lived in communities with a high poverty level while 44.9% resided in communities with low level of education. About two- thirds (67.1%) were rural dwellers. Regional distribution showed that about one-third (31.5%) were from the North-West region of Nigeria.

Table 3. Community-level characteristics of under-5 children and survival status, NDHS 2013 Deceased children, n (%)

Low

12 988 (41.3)

1 461 (11.3)

Medium

9 223 (29.3)

749 (8.1)

High

9 271 (29.5)

676 (7.3)

Low

7 829 (24.9)

446 (5.7)

Medium

10 011 (31.8)

833 (8.3)

High

13 642 (43.3)

1 607 (11.8)

Low

14 131 (44.9)

1 635 (11.6)

Medium

9 949 (31.6)

820 (8.2)

High

7 402 (23.5)

431 (5.8)

Urban

10 351 (32.9)

666 (6.4)

Rural

21 131 (67.1)

2 220 (10.5)

North Central

4 614 (14.7)

328 (7.1)

North East

6 517 (20.7)

661 (10.1)

North West

9 906 (31.5)

1 146 (11.6)

South East

2 816 (8.9)

263 (9.3)

South

3 747 (11.9)

249 (6.7)

South West

3 882 (12.3)

239 (6.2)

Community-level DMI

Community poverty level

Community level of education

Residence

Region

Maternal involvement in decisionmaking and childhood mortality

High

Low 0.14 0.12

Probability of death

Fig. 1 illustrates the KM curve for the probability of death according to individuallevel maternal DMI. At different ages, mortality was higher among children of women with low DMI. The under-five mortality rate was estimated as 120 per 1 000 live births (low DMI) and 84 per 1 000 live births (high DMI).

Fig. 2 presents the KM curve according to community-level involvement in decisionmaking. By the 59th month, the mortality rates were 138 per 1 000 live births (low involvement), 97 per 1 000 live births (medium involvement) and 82 per 1 000 live births (high involvement).

The results of multilevel discrete-time hazard models fitted to further explain the relationship between childhood mortality and maternal involvement in decisionmaking are shown in Table 4. Panel 1 of Table 4 shows the HR or the univariate models. Children of women with high involvement in decision-making were less likely to die before age 5 (HR 0.81, CI 0.74 0.89). Middle and rich household wealth quintiles also showed enhanced child survival. Children with a preceding birth interval ≤24 months were two times more likely to die than those with an interval of at least 36 months.

At the community level, high DMI (HR 0.62, CI 0.54 - 0.70) and high education (HR 0.49, CI 0.43 - 0.56) were protective against child death. Significant regional differences in childhood mortality risk were

All children, n (%)

Variable

0.10 0.08 0.06 0.04 0.02 0.00 0

4

8

12

16

20

24

28

32

36

40

44

48

52

56

60

Age (months)

Fig. 1. KM probability of childhood death according to individual women’s DMI, NDHS 2013.

observed in North East (HR 1.65, CI 1.37 1.99), North West (HR 1.94, CI 1.63 - 2.31) and South East Nigeria (HR 1.56, CI 1.26 1.93) compared with the South Western region.

29

SAJCH

MARCH 2017 Vol. 11 No. 1

Model I (Table 4, panel 2) shows that individual DMI was no longer statistically significant, but the community-level measure was. Children in low (HR 0.69, CI 0.61 - 0.79) and high (HR 0.64, CI 0.56 -

RESEARCH Table 4. Effects of decision-making-involvement and other selected variables on childhood mortality in Nigeria Individual-level variables

Univariate, Panel 1, OR (95% CI)

Model I, Panel 2 HR (95% CI)

Model II, Panel 3 Model III, Panel HR (95% CI) 4 HR (95% CI)

Model IV, Panel 5 HR (95% CI)

Maternal DMI, high v. low

0.81 (0.74 - 0.89)*

0.93 (0.84 - 1.04)

1.00 (0.89 - 1.11)

0.98 (0.88 - 1.09)

1.01 (0.90 - 1.12)

Maternal age at child’s birth (years)

<20

1.00

1.00

1.00

20 - 35

0.72 (0.64 - 0.79)*

0.81 (0.71 - 0.92)*

0.82 (0.72 - 0.93)*

>35

0.85 (0.75 - 0.97)*

0.91 (0.76 - 1.08)

0.92 (0.77 - 1.09)

Maternal education None

1.00

1.00

1.00

Primary

0.85 (0.76 - 0.94)*

1.01 (0.89 - 1.13)

1.01 (0.89 - 1.14)

Secondary/higher

0.54 (0.52 - 0.65)*

0.78 (0.67 - 0.89)*

0.80 (0.68 - 0.93)*

Household wealth index Poor

1.00

1.00

1.00

Middle

0.67 (0.59 - 0.75)*

0.74 (0.65 - 0.83)*

0.77 (0.67 - 0.88)*

Rich

0.52 (0.47 - 0.58)*

0.66 (0.58 - 0.75)*

0.78 (0.66 - 0.92)*

Maternal religion

Christianity

1.00

1.00

1.00

Islam

1.32 (1.19 - 1.46)*

0.92 (0.81 - 1.04)

0.94 (0.81 - 1.10)

Others Preceding birth interval (months)

1.04 (0.74 - 1.45)

0.79 (0.57 - 1.11)

0.80 (0.57 - 1.13)

First birth

1.62 (1.44 - 1.83)*

11.09 (6.91 - 17.76)*

11.09 (6.92 - 17.75)*

≤24

2.09 (1.87 - 2.33)*

2.09 (1.87 - 2.33)*

2.07 (1.85 - 2.31)*

25 - 36

1.32 (1.18 -1.48)*

1.32 (1.18 - 1.47)*

1.31 (1.17 - 1.46)*

>36

1.00

1.00

1.00

1.00

1.00

1.00

1.00

Community-level variables Community-level DMI Low

1.00

Medium

0.69 (0.61 - 0.78)*

0.69 (0.61 - 0.79)*

0.84 (0.74 - 0.96)*

0.83 (0.73 - 0.95)*

0.84 (0.74 - 0.96)*

High

0.62 (0.54 - 0.70)*

0.64 (0.56 - 0.74)*

0.83 (0.72 - 0.97)*

0.88 (0.75 - 1.03)

0.89 (0.76 - 1.04)

Community poverty level Low

1.00

1.00

1.00

Medium

1.48 (1.29 - 1.70)*

1.19 (0.99 - 1.41)

1.11 (0.93 - 1.33)

High

2.13 (1.88 - 2.43)*

1.36 (1.09 - 1.70)*

1.09 (0.85 - 1.39)

Community level of education Low

1.00

1.00

1.00

Medium

0.71 (0.63 - 0.79)*

0.95 (0.81 - 1.11)

1.02 (0.86 - 1.21)

High

0.49 (0.43 - 0.56)*

0.77 (0.61 - 0.97)*

0.91 (0.71 - 1.17)

Residence Urban v. rural Region

0.61 (0.55 - 0.68)*

0.77 (0.67 - 0.88)*

0.79 (0.69 - 0.91)*

North Central

1.15 (0.94 - 1.41)

0.85 (0.69 - 1.05)

0.84 (0.68 - 1.03)

North East

1.65 (1.37 - 1.99)*

0.96 (0.77 - 1.19)

0.89 (0.72 - 1.12)

North West

1.94 (1.63 - 2.31)*

1.08 (0.87 - 1.35)

1.03 (0.82 - 1.29)

South East

1.56 (1.26 - 1.93)*

1.47 (1.18 - 1.83)*

1.32 (1.06 - 1.65)*

South

1.09 (0.88 - 1.35)

0.90 (0.73 - 1.12)

0.84 (0.68 - 1.05)

South West

1.00

1.00

1.00

Continued ...

30

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH Table 4. (continued) Effects of decision-making involvement and other selected variables on childhood mortality in Nigeria Univariate, Panel 1, OR (95% CI) Random effects: community level

Model I, Panel 2 HR (95% CI)

Model II, Panel 3 Model III, Panel HR (95% CI) 4 HR (95% CI)

Model IV, Panel 5 HR (95% CI)

Variance (SE)

0.4899 (0.0304)

0.4686 (0.0303)

0.3906 (0.0310)

0.3746 (0.0311)

0.3636 (0.0314)

ICC (%)

7.80*†

6.25*

4.43*

4.09*

3.86*

Log likelihood

–11 554.149

–11 530.16

–11 328.60

–11 469.507

–11 310.098

*p<0.05 †

For a null model.

Low

High

Medium

0.14

Probability of death

0.12 0.10 0.08 0.06 0.04 0.02 0.00 0

4

8

12

16

20

24

28

32

36

40

44

48

52

56

60

Age (months)

Fig. 2. KM probability of childhood death according to community-level women’s DMI, NDHS 2013.

0.74) decision-making communities were less likely to die before age five.

In model II (Table 4, panel 3), other individual-level variables were controlled to assess the effect on DMI. Low (HR 0.84, CI 0.74 - 0.96) and medium (HR 0.83, CI 0.72 0.97) community-level DMI were associated with lower risk of childhood mortality.

The individual-level variables maintained their direction of effect as in the univariate models.

Model III (Table 4, panel 4) shows the extent to which other community-level measures explained the effect of individual/ community-level decision-making on under-5 mortality. The HRs for DMI remained the same as in model II. With the exception of the South East region (HR 1.47, CI 1.18 - 1.83), the statistical significance of the differentials between South West and other regions (North East and North West) has disappeared.

Panel 5 of Table 4 shows the results of model IV, in which all individual- and comunity-level variables were controlled.

There was no difference in the risk of under5 death between children whose mothers

had low v. high DMI. However, communitylevel decision-making was significant, with lower risks of death among children in communities with medium maternal DMI (HR 0.84, CI 0.74 - 0.96). Other variables associated with lower risk of under-5 death included: maternal age 20 - 35 years (HR 0.82, CI 0.72 - 0.93); secondary/higher education (HR 0.80, CI 0.68 - 0.93); rich household wealth quintile (HR 0.78, CI 0.66 0.92); and urban residence (HR 0.79, CI 0.69 0.91). Factors found to increase the hazard of under-5 death were being a first birth (HR 11.09, CI 6.92 - 17.75), and birth interval <24 months (HR 2.07, CI 1.85 - 2.31).

Discussion

The status of a woman is directly related to her ability to seek medical and nutritional care for herself and her children. In a patriarchal society such as Nigeria, the ability of women (as direct targets for child survival programmes) to translate knowledge of interventions to child care will depend on their decision-making authority. This study was therefore conducted to explore the 31

SAJCH

MARCH 2017 Vol. 11 No. 1

effects of individual- and community-level DMI of mothers on child mortality rates in Nigeria.

The risk of childhood mortality was higher among children of women with low DMI. This agrees with results from Nepal, [13] which showed that children of women involved in decision-making on their healthcare were less likely to die before 5 years. The findings also indicate that community-level DMI was associated with a lower risk of child death. This is similar to reports by Desai and Johnson,[14] which revealed that the effects of women’s DMI on child health outcomes were stronger at the community than individual level in selected Asian and sub-Saharan African countries (excluding Nigeria). These findings suggest that an individually empowered woman may have limited DMI if she lives in a community where women have limited decisionmaking power. The implication is that community norms and attitudes about decision-making may be more important for child survival efforts than individual decision-making authority.[2] The lack of relationship between individual women’s DMI and childhood mortality is similar to reports from four Asian countries (India, Malaysia, the Philippines and Thailand) in which the association between individual women’s autonomy and child survival was found to be weak.[14] It is likely that women are not able to translate their individual decision-making authority into child survival advantages. Alternatively, it could be that appropriate measures of empowerment that adequately capture the type of individual DMI that influences child survival are not being used.

Development of empowerment measures that are culturally sensitive in different settings has been advocated, but this would also suffer from global standardisation, which is necessary for intercountry and regional comparisons.[4] The results further reaffirmed the strong influence of education, type of residence (rural or urban), maternal age and wealth

RESEARCH quintile on child survival. The childhood mortality differentials according to these variables have been previously described. [16,17] The random component in the multilevel model showed significant intracommunity correlation in the risk of child death.

This correlation justified the analytical method used in the present study because children in the same community are exposed to similar circumstances, cultural norms and practices, and thus they would not have been completely independent of one another as far as the underfive death rate was concerned. Childcare practices and health-seeking behaviours are greatly influenced by community norms and these also affect child health outcomes.[20]

Study limitations

One limitation in the analyses is the inability to explore contextual variables that capture community norms and practices related to child survival. Such variables were not collected in the NDHS. The modelling procedure, however, adequately controlled for these unobserved variables. Also, there was no measure of decisionmaking about child healthcare. Questions related to decisions about child healthcare would have afforded an assessment of the influence of decision-making (in this regard) on child healthcare. Regardless, the analyses were based on nationally representative data. Therefore, the findings are applicable to current policies and programmes on women’s empowerment and child healthcare.

Conclusion

Under-5 mortality is negatively associated with maternal DMI in Nigeria.

The effect of community-level women’s DMI on the risk of childhood death is greater than that of the individual level, even though the former is partly explained by individual- and community-level education and wealth status. Empowerment programmes should not only target individual women but should also address community norms and beliefs about the ability of women as primary caregivers to take initiative and make timely decisions, especially on matters related to personal health, as well as that of their children. Given the importance of child survival in sustainable development, questions about decision-making on children’s health should be included in the next round of demographic and health surveys. This will make it possible to directly assess the extent to which women have autonomy in child care.

Acknowledgements. We acknowledge the ICF Macro International and other implementing partners for granting access to the NDHS data.

We also appreciate the University of the Witwatersrand, Johannesburg, South Africa, for providing a research-friendly environment to conduct this study. This research was supported by the Consortium for Advanced Research Training in Africa (CARTA). CARTA is jointly led by the African Population and Health Research Center and the University of the Witwatersrand and funded by the Wellcome Trust (UK) (Grant no. 087547/Z/08/Z), the Carnegie Corporation of New York (Grant no.​ B8606.R02), and Sida (Grant no. 4100029). The statements made and views expressed are solely the responsibility of the authors.

32

SAJCH

1. Bloom SS, Wypij D, das Gupta M. Dimensions of women’s autonomy and the influence on maternal health care utilization in a North Indian city.

Demography 2001;38(1):67-78. https://dx.doi.org/10.2307/3088289 2. Kabeer N. Gender equality and women’s empowerment: A critical analysis of the third millennium development goal. Gend Dev 2005;13(1):13-24. https:// dx.doi.org/10.1080/13552070512331332273 3. Sunita K. A focus on gender: Collected papers on gender using DHS data.

Maryland, USA: ORC Macro, 2005:1-13.

4. Ahmed S, Creanga AA, Gillespie DG, Tsui AO. Economic status, education and empowerment: Implications for maternal health service utilization in developing countries. PLoS ONE 2010;5(6):1-6. https://dx.doi.org/10.1371/ journal.pone.0011190 5. Shimamoto K, Gipson JD. The relationship of women’s status and empowerment with skilled birth attendant use in Senegal and Tanzania. BMC Pregnancy Childbirth 2015;15(1):1-11. https://dx.doi.org/10.1186/s12884-015-0591-3 6. Babalola S, Fatusi A. Determinants of use of maternal health services in Nigeria – looking beyond individual and household factors. BMC Pregnancy Childbirth 2009;9(1):1-13. https://dx.doi.org/10.1186/1471-2393-9-43 7. Hou X, Ma N. The effect of women’s decision-making power on maternal health services uptake: Evidence from Pakistan. Health Policy Plan 2013;28(2):176184. https://dx.doi.org/10.1093/heapol/czs042 8. Sado L, Spaho A, Hotchkiss DR. The influence of women’s empowerment on maternal health care utilization: Evidence from Albania. Soc Sci Med 2014;114(1):169-177. https://dx.doi.org/10.1016/j.socscimed.2014.05.047 9. Woldemicael G. Do women with higher autonomy seek more maternal and child health-care? Evidence from Ethiopia and Eritrea. Stockholm: Stockholm University, 2007:1-28.

10. Fotso JC, Ezeh AC, Essendi H. Maternal health in resource-poor urban settings: How does women’s autonomy influence the utilization of obstetric care services?

Reprod Health 2009;6(1):1-8. https://dx.doi.org/10.1186/1742-4755-6-9 11. Corroon M, Speizer IS, Fotso JC, et al. The role of gender empowerment on reproductive health outcomes in urban Nigeria. Matern Child Health J 2014;18(1):307-315. https://dx.doi.org/10.1007/s10995-013-1266-1 12. Ononokpono DN, Odimegwu CO. Determinants of maternal health care utilization in Nigeria: A multilevel approach. Pan Afr Med J 2014;17(2):5-11. https://dx.doi.org/10.11604/pamjs.supp.2014.17.1.3596 13. Adhikari R, Sawangdee Y. Influence of women’s autonomy on infant mortality in Nepal. Reprod Health 2011;8(1):1-7. https://dx.doi.org/10.1186/1742-4755-8-7 14. Desai S, Johnson K. Women’s decision-making and child health: Familial and social hierarchies. Maryland, USA: ORC Macro, 2005:55-67.

15. Ononokpono DN, Odimegwu CO, Imasiku EN, Adedini SA. Does it really matter where women live? A multilevel analysis of the determinants of postnatal care in Nigeria. Matern Child Health J 2014;18(4):950-959. https:// dx.doi.org/10.1007/s10995-013-1323-9 16. Adedini SA, Odimegwu C, Imasiku EN, Ononokpono DN, Ibisomi L. Regional variations in infant and child mortality in Nigeria: A multilevel analysis. J Biosoc Sci 2014;47(2):165-187. https://dx.doi.org/10.1017/s0021932013000734 17. Akinyemi JO, Bamgboye EA, Ayeni O. New trends in under-five mortality determinants and their effects on child survival in Nigeria: A review of childhood mortality data from 1990-2008. Afr Popul Stud 2013;27(1):25-42. https://dx.doi.org/10.11564/27-1-5 18. National Population Commission (NPC). Nigeria Demographic and Health Survey 2013. Abuja, Nigeria, and Rockville, Maryland, USA: NPC and ICF International, 2014;377-396.

19. Rutstein SO, Staveteig S. Making the demographic and health surveys wealth index comparable. Maryland, USA: ICF International, 2014:1-59.

20. Fayehun O, Omololu O. Ethnicity and child survival in Nigeria. Afr Popul Stud 2011;25(1):92-112. https://dx.doi.org/10.11564/25-1-258

MARCH 2017 Vol. 11 No. 1

RESEARCH

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

Pubertal breast development in primary school girls in Sokoto, North-Western Nigeria M O Ugege,1 MBBS, FWACP, FESPE; K I Airede,2 MBBS, MPH, FWACP; A Omar,3 MD, MMed (Paed), FPE; O Pinhas-Hamiel,4 MD; P K Ibitoye,1 MBBS, FWACP; U Chikani,5 MBBS, FMCPaed, FESPE; A Adamu,1 MBBS, FWACP; K O Isezuo,1 MBBS, FWACP; F Jiya-Bello,1 MBBS, FMCPaed; J A Legbo,1 MBBS; M Sanni,1 MBBS Department of Paediatrics, Usmanu Danfodiyo University Teaching Hospital, Sokoto, Nigeria Faculty of Clinical Sciences, University of Abuja, Federal Capital Territory, Abuja, Nigeria 3 Kenyatta National Hospital, Nairobi, Kenya 4 Paediatric Endocrine and Diabetics Unit, Edmund and Lily Safra Children’s Hospital, Sheba Medical Center, Sheba, Israel 5 University of Nigeria Teaching Hospital, Enugu, Nigeria 1 2

Corresponding author: M O Ugege (shallyben@yahoo.com) Background. There is wide variation in normal pubertal timing among various populations.

Objectives. To determine the mean age of pubertal stages of breast development and menarche, and the influence of nutrition and ethnicity on pubertal onset in primary school girls in Sokoto, North-Western Nigeria.

Methods. A cross-sectional study using a multistage random sampling design was conducted on 994 primary school girls in grades 3 - 6.

Weight and height measurements and Tanner breast staging were done. Body mass index (BMI) was calculated, and a BMI-for-age percentile was used to categorise nutritional status. There were four major ethnic groups. P≤0.05 was taken as showing statistical significance.

Results. The participants’ mean age was 10.23 years (standard deviation (SD) 1.70, range 6 - 15 years). Of the 994 girls, 628 (63.2%) were pre-pubertal, and 366 (36.8%) were pubertal. Of the latter, 158 (15.9%) were in breast stage 2, while 112 (11.3%), 70 (7.0%) and 26 (2.6%) were in breast stages 3, 4 and 5, respectively. The mean ages (SD; range) of pubertal onset and menarche were 10.50 (1.33; 8 - 13), and 12.67 (1.65; 11 - 15), years, respectively. The overnourished (overweight/obese) and Igbo ethnic group girls had early-normal pubertal onset (p=0.006 and p=0.001, respectively).

Conclusion. The mean ages of Tanner breast stages 1 - 5 and menarcheal age of girls in Sokoto, North-Western Nigeria, were within the age ranges reported worldwide. Pubertal onset was influenced by nutrition.

S Afr J Child Health 2017;11(1):33-37. DOI:10.7196/SAJCH.2017.v11i1.1199

Puberty is a period of maturation during which secondary sexual characteristics appear and reproductive capability is attained. Its onset requires an intact hypothalamic-pituitary-gonadal axis, the reactivation of the secretion of pulsatile gonadotrophin-releasing hormone from its stage of childhood quiescence, and the stimulation of luteinising hormone and follicle-stimulating hormone secretion, which in turn activate the production of the gonadal sex steroid, oestradiol, in girls.[1-3] The external sign of puberty onset in girls is the appearance of breast buds, because breast tissue is the primary target of oestradiol.

This usually occurs between 10 and 11 years of age.[4] Menarche occurs in middle or late puberty, usually within 2 years of pubertal onset.[2] However, there is wide variation in the normal pubertal timing among various populations separated in time and space.[4,5] Many different factors contribute to this variation, including genetic factors,[2] nutrition,[6,7] ethnicity[8,9] and socioeconomic status.[6,10] Previous studies have shown a decreasing mean age of pubertal breast development over time.[6-11] Most have documented the mean age as being between 8 and 9 years. This has been attributed to the secular trend, defined as the continuous movement of a variable in a somewhat consistent way over a long period of time. The secular trend of puberty therefore refers to a decreasing average age of puberty over time, explained by the decline in frequency and severity of illnesses, and the better health and nutritional status of the general population.[6,12] Studies have shown that obesity and overweight are important contributing factors to the earlier onset of puberty in girls.[3,7,10] Most of these studies were conducted in America, Europe

33

SAJCH

and Asia. The studies conducted on black people were carried out on African Americans. Few studies on pubertal onset have been conducted on African children living in Africa.[13,14] The influence of environment and geographical location on pubertal timing has been previously documented.[15,16] Most Nigerian studies have focused on age at menarche and influencing factors.[17,18] To the best of our knowledge the only similar Nigerian study was conducted on girls (10 - 20 years old) from the Igbo ethnic group.[19] There is a paucity of literature regarding pubertal breast development in the black African population, and utilising normative data from white girls or African Americans may not be truly representative of the Nigerian or African population as a whole. The aim of this study was to determine the mean age of pubertal stages of breast development and menarche, and to determine the influence of nutrition and ethnicity on pubertal onset, in primary school girls living in Sokoto, Nigeria.

Methods

Study design

A cross-sectional study was conducted between December 2014 and March 2015 on primary school girls in grades 3 - 6 in both public and private primary schools.

Study location

The study was carried out in Sokoto, the capital city of Sokoto state, in the North-Western geopolitical zone of Nigeria. Nigeria has more than 250 ethnic groups, but the three major ethnic groups

MARCH 2017 Vol. 11 No. 1

RESEARCH are the Hausa-Fulani, Igbo and Yoruba, which together constitute 71% of the population.[20] In Sokoto most Nigerian ethnic groups are well represented, but the majorities are the indigenous Hausa and Fulani. For the purpose of this study, ethnicity was stratified to five groups: Hausa, Fulani, Igbo, Yoruba and others, which consist of a heterogeneous group of minor tribes found in Nigeria and neighbouring countries. This was in an attempt to reflect genetic background, and the sociocultural practices peculiar to the major ethnic groups in Nigeria, some of which are related to type of diet, nutrition and lifestyle. A cross-section of all socioeconomic classes in the metropolis was well represented.

The data were processed using the Statistical Package for Social Sciences version 20 (IBM Corp., USA). SD scores (or z-scores) were generated from the BMI, and these were converted to BMI percentiles using a z-score percentile converter.[24] The BMI percentiles were used to categorise nutritional status. The age at onset of puberty was categorised as precocious (<8 years), normal (8 - 13 years) or delayed (>13 years). Statistical analysis involved Fisher’s exact test, analysis of variance (ANOVA), Student t-tests and logistic regression, as appropriate. All statistical analysis was done at a p≤0.05 level of significance.

Ethical approval and parental consent

General characteristics of the study population

The study protocol was approved by the Ethical Committee of the Usmanu Danfodiyo University Teaching Hospital (UDUTH), Sokoto (ref. no. UDUTH/HREC/2014/No 265). Written permission was obtained from the Ministry of Education, and written parental informed consent was obtained from the parents or guardians, as well as assent from all participants.

Recruitment and data collection

Questionnaires were used to record the subjects’ biological and sociodemographic data, including age, ethnicity/tribe (the ethnic group of the father was taken as each subject’s ethnic group), the marital status of parents/guardians and their occupations. The age at menarche was obtained by self-report if the subject was postmenarcheal at the time of the interview. Weight was measured to the nearest 0.1 kg using an electronic portable weighing scale (Omron digital personal scale HN283, Japan). The subjects were dressed in lightweight school uniforms. Height was measured using a portable stadiometer (Prevenar 13 Kidmeter, Pfizer, Spain) to the nearest 0.1 cm. These measurements were taken by properly trained research assistants. Body mass index (BMI) was calculated by dividing the weight in kilograms by the square of the height in metres. Nutritional status was classified using BMI-for-age percentiles according to National Centre for Health Statistics[21] as follows: underweight <5th percentile, normal weight 5th - <85th percentile, overweight: 85th <95th percentile, obese: ≥95th.

Pubertal breast staging

Pubertal breast staging was assessed by the principal researcher according to the pubertal staging described by Tanner.[4] Breast stages 1 - 5 were assessed by inspection, and palpation where necessary, for instance, to properly distinguish a visible breast bud from simple adipose tissue. Where the two breasts of an individual were not at the same stage of development, the stage of the more advanced side was recorded. The onset of puberty was accepted as Tanner breast stage 2 (B2 – appearance of breast bud). In cases of uncertainty between two adjacent stages, the subject was additionally evaluated by a second researcher and a consensus decision was reached.

Statistics

Power analysis showed that a sample size of 600 girls would have 80% power to detect a difference of 1 month, assuming that the common standard deviation (SD) is 1 with a 0.05 two-sided significance level.[22,23] Multistage random sampling was utilised in the selection of three local government areas (LGAs) out of five, and 20 (30%) of the total number of schools in the Sokoto metropolis. Both public and private schools were selected in proportion to their distribution in the three LGAs. Consequently, we selected 12 private and 8 public schools.

However, only 70% of the selected schools (n=14: 8 private and 6 public) permitted the study (21% of the total number of schools in the metropolis).

All the primary schools were co-educational, and all the girls whose parents consented in grades 3 - 6 were included as a cluster.

34

SAJCH

Results

The study cohort comprised 994 girls, 506 (51%) from private schools and 488 (49%) from public schools. Their mean (SD) age was 10.23 (1.7) years, range 6 - 15 years. Most participants were between the ages of 8 and <10 or 10 and <12 years (34.9% and 39.6%, respectively). The 6 - <8 years group were the minority (4.5%). The 12 - <13 and 13 - 15-year-olds stood at 10.6% and 10.4%, respectively.

Age was unknown in 14 (1.4%) subjects.

Ethnic distribution

The majority (n=486; 49%) of the participants were Hausa and a minority (n=70; 7.1%) were Igbo. Ninety-six (9.7%) were Yoruba, 84 (8.5%) were Fulanis, and 256 (5.8%) were classified as ‘others’, a heterogeneous group consisting of minor tribes from within the six geopolitical zones of Nigeria, and tribes from neighbouring countries such as Niger and Benin Rebublic. Two participants did not know their ethnic group (0.2%).

The nutritional status of the study population using BMI-for-age percentiles

The majority of the subjects (888 (89.3%)) were of normal weight.

Only 2 (0.2%) were underweight. There were 40 overweight and 64 obese subjects (4% and 6.4%, respectively).

Biodemographic, anthropometric and pubertal characteristics of the study population

Table 1 shows the distribution of the study population by age, weight, height and BMI, median BMI-SDs and type of school, categorised according to the Tanner stage and menarche.

The majority (628 (63.2%)) of the study population was prepubertal. Only 366 (36.8%) were pubertal, of whom 158 (15.9%) were in breast stage 2, and 112 (11.3%), 70 (7.0%) and 26 (2.6%), were in breast stages 3, 4 and 5, respectively. Fourteen (1.4%) had attained menarche. Mean (SD) age of pubertal onset (B2) was 10.50 (1.33) years. The mean (SD) (age) of menarche was 12.67 (1.65) years. The mean age of Tanner breast stages significantly increased as the stages progressed (p<0.001), except for Tanner 4. The mean duration from pubertal onset to menarche was 2.17 years. The mean weight, height and BMI of the study population increased with progressing Tanner stage from B1 to B5. This generally observed increase was statistically significant (p<0.001). There were 1.3 times more girls in B2 (pubertal onset), and 2.5 times more post-menarcheal girls in private schools than public schools.

Comparison of mean age of pubertal onset, weight, height and BMI of girls in public and private schools

Table 2 shows the mean age of pubertal onset and menarche, weight, height and BMI of subjects in Tanner stage B2 in private and public schools. The girls in private schools had a lower mean age of pubertal onset than the girls from public schools (p=0.04). Similarly, girls in private schools had a higher mean weight, height and BMI than those in public schools (p=0.03, 0.001 and 0.04, respectively). The age ranges of pubertal onset and menarche of subjects in private schools were 8 -

MARCH 2017 Vol. 11 No. 1

RESEARCH Table 1. Distribution of study population by type of school, and the mean age, weight, height, BMI and median BMI-SDS according to the Tanner stage and menarche* TS B1 B2 B3 B4 B5 M

Pu, N (%) 310 (63.5) 68 (13.9) 64 (13.1) 30 (6.1) 16 (3.3) 4 (0.82)

Pr, N (%) 318 (62.8) 90 (17.8) 48 (9.5) 40 (7.9) 10 (2.0) 10 (1.97)

Total, N (%) 628 (63.2) 158 (15.9) 112 (11.3) 70 (7.0) 26 (2.6) 14 (1.4)

Age (years), mean (SD) 9.63 (1.47) 10.50 (1.33) 11.58 (1.52) 11.58 (1.08) 13.72 (1.33) 12.67 (1.65)

Weight (kg), mean (SD) 26.21 (4.93) 31.51 (5.44) 33.87 (6.40) 37.50 (7.25) 48.24 (5.26) 50.06 (6.15)

Height (cm), mean (SD) 131.0 (7.05) 138.3 (5.68) 144.4 (7.48) 145.7 (7.10) 152.8 (4.30) 152.4 (4.97)

BMI (kg/m2), mean (SD) 15.22 (1.87) 16.52 (2.82)† 16.16 (1.90)‡ 17.59 (2.79) 21.55 (3.19) 21.54 (2.54)

BMI-SDS (median) –0.2 (1.0) –0.3 (1.0) –0.1 (1.0) –0.2 (1.0) –0.4 (1.0) –0.5 (1.0)

TS = Tanner stage; M = menarche; Pu = public school; Pr = private school; BMI-SDS = body mass index standard deviation score. † ‡ *Age v. TS p=<0001; Weight v. TS p=<0.001; Height v. TS p= <0.001; BMI v. TS p=<0.001; v. p=0.24.

Table 2. Comparison of mean age of pubertal onset (B2), menarche, weight, height and BMI of subjects in public and private school Variable Age (years) Menarche (years) Weight (kg) Height (cm) BMI (kg/m2) BMI SDS (median)

N 90 10 90 90 90 90

Private school Mean (SD) Range 9.90 (1.04) 11.75 (0.76) 32.94 (6.46) 138.39 (5.95) 17.31 (3.38) –0.20 (1.00)

N

8.00 - 13.00 11.00 - 12.92 25.10 - 55.30 125.0 - 151.5 13.80 - 29.50 –1.50 - 5.10

66 4 68 68 68 68

†

Public school Mean (SD) Range

p-value*

11.11(1.23) 15.00 (0) 29.62 (2.74) 138.23 (5.28) 15.49 (1.22) –0.20 (1.00)

0.04 0.03 0.001 0.04 -

9.00 - 13.00 23.7 - 34.9 128.5 - 149.5 12.83 - 17.94 –2.00 - 6.50

* p-values comparing public and private schools. † Excluded two girls with unknown age.

13 and 10 - 15 years, while those of their counterparts in public schools were 9 - 13 and 11 - 15 years, respectively.

24.184, p=0.001). However, multinomial regression analysis showed that ethnicity was not a predictor of the age of pubertal onset.

Influence of nutrition on the age of pubertal onset

Discussion

A total of 144 (91.1%), 2 (1.3%) and 12 (7.6%) of the subjects in Tanner B2 were normal weight, overweight and obese, respectively. None was underweight. All the subjects in Tanner B2 were between 8 and 13 years old. To determine the relationship between nutritional status and age at onset of puberty, this age range was further categorised as shown in Table 3, which indicates the influence of nutritional status on the age of pubertal onset. Eighty-six (55.1%) of girls in Tanner breast stage 2 (B2) were between 8 and 10 years old (early-normal pubertal onset), 54 (34.6%) were >10 - 12 years (middle-normal pubertal onset) and 16 (10.3%) were >12 - 13 years (late-normal pubertal onset). A little above half (51.7%) of the girls with normal weight had earlynormal pubertal onset. All the girls who were overweight or obese had early-normal pubertal onset, even though overweight and obese girls in B2 constituted the minority of the sample. A statistically significant association was observed between overnutrition (overweight and obesity) and early-normal age of pubertal onset (Fisher’s exact χ2 (FE χ2) = 12.057, p=0.006). Logistic regression analysis showed that girls who were overweight or obese were only 1.259 times more likely to have early-normal pubertal onset compared with those of normal weight (odds ratio (OR) 1.259, p=0.001).

Influence of ethnicity on the age of pubertal onset

The influence of ethnicity on the age of pubertal onset is shown in Table 4. Analysis involved all participants in B2. All of the Igbos (100%), 66.7% of the Yorubas, 50% of the Fulanis and 36.4% of the Hausa participants had early-normal pubertal onset. Half (50%) of the Fulanis and 45.5% of the Hausa participants had middle-normal pubertal onset. Only 18.2% of the Hausa participants had latenormal pubertal onset.

A statistically significant association was observed between the Igbo ethnic group and early-normal age of pubertal onset ((FE χ2) = 35

SAJCH

This study has demonstrated mean (SD) age of Tanner breast stages 1 - 5 of 9.63 (1.47), 10.50 (1.33), 11.58 (1.52), 11.58 (1.08), and 13.72 (1.33), respectively. The girls who were in a more advanced Tanner stage of breast development were older. This finding, as in most other studies, was statistically significant,[10,11,23] suggesting a linear relationship between sexual maturity and chronological age in a normal setting.

The mean (SD) age of pubertal onset of 10.50 (1.33) years found in this study is similar to the 10.55 (1.57) reported by Bazrafshan et al.[25] in Northern Iran in 2005, and slightly lower than the 10.71 years reported in Egypt by Hosney et al.[14] in 2000. However, our finding is higher than the 9.50 but similar to the 10.30 years reported by Wu et al.[9] from a study on African Americans and white Americans, respectively, in the USA in 2002. It is also greater than the 9.13 years reported in Denizli, Turkey in 2005 (Semiz et al.),[6] the 10.10 years reported in all provinces in Iran in 2006 (Rabbani et al.),[26] the 9.86 years reported in Copenhagen in 2008 (Juul et al.),[27] the 9.71 years reported in Qazvin province, Iran in 2010 (Saffari et al.),[11] the 10.10 years reported in urban South Africa (SA) in 2004 (Jones et al.)[28] and the 10.02 years reported for Nigerian Igbo girls in 2015 (Nzeako et al.).[19] These variations could be due to racial, geographical, nutritional and socioeconomic differences. The fact that the studies reporting an earlier mean age of pubertal onset were more recently conducted than those reporting later pubertal onset, particularly in the same countries,[11,25] may suggest a secular trend, a decreasing average age of puberty over time, towards the earlier onset of puberty, especially in developed countries. This is attributed to better nutrition and perhaps a greater prevalence of overweight and obese girls, and improved socioeconomic conditions. In contrast, researchers in SA[28] reported a stable onset of puberty (age of breast budding ranging

MARCH 2017 Vol. 11 No. 1

RESEARCH between 9.8 - 10.5 years) for 10 - 15 years and a significant secular trend in menarche, with an average decline of 0.5 years/decade for urban black SA girls, suggesting that the tempo of pubertal maturation is increasing in girls born in the Soweto-Johannesburg area.[28] Their finding contrasts with those of most other studies[10,11,23] that have reported a significant secular trend in breast development and stable mean age at menarche over several decades, suggesting the end of the secular trend for menarche.

Our study did not show pubertal onset below 8 years nor above 13 years of age. All girls in B2 were interestingly within the normal documented age of pubertal onset of 8 - 13 years, and therefore there was no precocious or delayed puberty in this study. This may be partly due to the small percentages of 6 - 8 year olds (4.5%) and 13 - 15 year olds (10.4%) compared with the larger percentage of the 8 - <12 year olds (74.5%) in the study population. The lower age of 6 years in this study population is the same as that in similar studies that reported precocious puberty,[6,11,26] while the upper age range of 15 years in the present study was also used in similar studies (15 20 years).[6,11,26] In this study, no girl older than 13 years was in Tanner B2.

This finding could be related to the fact that none of girls in B2 was underweight. The majority were of normal weight and the minority overweight or obese. It is documented in the literature that malnutrition (undernutrition) is probably the most important mechanism responsible for delayed puberty;[29] therefore it is not surprising that there were no girls with delayed pubertal onset in our study.

The mean (SD) menarcheal age of 12.67 (1.65) years in this study is within the range reported in other previous studies in Nigeria,[18] Africa14] and developed countries.[6,9] All the post-menarcheal girls were in B4 and B5. This buttresses the fact that menarche tended to occur toward the end of pubertal development, and therefore gives more information regarding the normal progression of puberty rather than the onset. The duration from pubertal onset to menarche of 2.17 years found in this study is similar to the 2.3 years reported by Marshall and Tanner[4] in London, UK.

Normal weight 72 (50.7)

Overweight* Obese* 2 (100) 12 (100)

Total 86 (55.1)

54 (38.0)

0 (0)

0 (0)

54 (34.6)

16 (11.3)

0 (0)

0 (0)

16 (10.3)

The finding of statistically significant increases in the mean weight, height and BMI with each progressing Tanner stage, and with menarche, is similar to the findings of Rabbani et al.[26] and Saffari et al.[11] in Iran, and those of Adesina et al.[18] in Port Harcourt, Nigeria. This suggests that the level of nutrition, commonly assessed by these anthropometric indices, plays a vital role in the onset and progression of pubertal stages. The fact that the prepubertal girls (B1) had the lowest mean weight, height and BMI and the mature girls in B5 had the highest means of these indices further supports this fact.

However, it is not possible to distinguish the effect of body weight on breast development from the effect of breast development on body weight in the present cross-sectional study, since the subjects were already in a stage of breast development before measurements of weight were taken, which is a limitation. A longitudinal study would be the best approach to distinguish between the two. Rabbani et al.[26] have shown from their study that a weight of about 30 kg is critical for the onset of puberty. This is similar to our results.

Even though our study population did not have precocious pubertal onset (i.e. onset before 8 years of age), the statistically significant association between overweight and obese girls with early-normal age of pubertal onset suggests that there is a relationship between overnutrition (overweight and obesity) and pubertal onset. This is similar to the findings in other studies.[7,10] It has been hypothesised that body fat mass is a facilitator for the timing of puberty in girls.[30] This is related to leptin levels, a hormone produced by the adipocytes, which rises in girls at the time of puberty.[31] Overweight girls, therefore, are more likely to enter puberty at an earlier age.[7,8,32] Our findings agree with this hypothesis. Semiz et al.,[6] however, found no statistically significant association with age of pubertal onset and weight, whether normal or overweight, while the menarcheal age of the overweight and obese was significantly lower than girls with normal weight. The reason for this disparity in findings is not quite clear, but may be related to leptin levels, which have been reported to be higher in black than white people.[7] The finding of a statistically significant association between the Igbo ethnic group and early-normal age of pubertal onset suggests a relationship between ethnic group and age at pubertal onset, similar to other studies in the USA.[8,9] This could be explained by a higher nutritional status of the Igbo girls, as well as other confounding factors such as socioeconomic status and probably type of diet. Furthermore, multinomial regression analysis has shown that ethnicity is not a predictor of age at onset of puberty. The finding of earlier pubertal onset in private school girls (9.90 years) compared with public school girls (11.11 years) further suggests the likely influence of socioeconomic status, which also to a large extent determines nutritional status. It is expected that girls in private schools will be of a higher socioeconomic status than those in public schools; however, the incomplete socioeconomic data obtained in the present study will not permit any conclusion on socioeconomic influence.

*142 (100)

2 (100)

12 (100)

156 (100)

Study limitations

Table 3. Influence of nutrition on the age of pubertal onset BMI-percentile category, n (%)

Age of pubertal onset Earlynormal (8 - 10) Middlenormal (>10 - 12) Late-normal (>12 - 13) Total

The cross-sectional study design limited the researchers from distinguishing the effect of weight on puberty from the effect of

2

*Fisherâ&#x20AC;&#x2122;s exact Ď&#x2021; =12.057, p=0.006.

Table 4. Influence of ethnicity on the age of pubertal onset Age of pubertal onset Early-normal

Hausa 24 (36.4)

Fulani 2 (50.0)

Yoruba 12 (66.7)

Ethnic group, n (%) Igbo* 12 (100)

Others 36 (64.2)

Total 86 (55.1)

Middle-normal Late-normal

30 (45.5) 2 (18.2)

2 (50.0) 0 (0)

6 (33.3) 0 (0)

0 (0) 0 (0)

16 (28.6) 4 (7.1)

54 (34.6) 16(10.3)

Total

66 (100)

4 (100)

18 (100)

12 (100)

56 (100)

156 (100)

2

*Fishers exact Ď&#x2021; = 24.184, p=0.001.

36

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH puberty on weight, since the girls were already in a stage of pubertal breast development before the measurements were taken.

The incomplete socioeconomic data we encountered did not permit any conclusion on the added influence of socioeconomic status on pubertal onset.

Conclusion

The mean ages of Tanner breast stages 1 - 5 and menarcheal age of girls living in Sokoto, North-West Nigeria, were within the age ranges reported worldwide. The mean age of pubertal onset demonstrated in the current study is higher than the reported age of African American girls living in the USA, but similar to white American girls living in the USA and black SA girls living in SA. Precocious and delayed puberty were not encountered. Even though the age of pubertal onset was within the normal range reported worldwide, there were variations in the age of onset within the normal range with nutrition, as the overweight or obese girls and Igbo ethnic group attained puberty in the early-normal range. It is recommended that there should be vigorous pursuit of further multicentre collaborative studies and evaluations on pubertal breast development in Nigeria and other African countries in order to validate these findings and establish normative data for the Nigerian/ African population as a whole.

Acknowledgements. The researchers are grateful to the participating schools and the girls for their co-operation, and to Dr Awosan, a consultant public health physician and statistician in UDUTH, Sokoto, for his assistance with statistics and analysis of data. We thank our research assistants, Dr Susan Adeyi and Dr Fatima Abubarkar for their help with data collection. We are grateful to Prof. Allan D Rogol, a paediatric endocrinologist, and an international tutor with the Paediatric Endocrinology Training Centre for Africa (PETCA), for critically reviewing the manuscript. Finally, we gratefully acknowledge the PETCA programme, Nairobi, Kenya, for supporting the research.

1. Dunkel L, Quinton R. Transition in endocrinology: Induction of puberty. Eur J Endocrinol 2014;170(6):229-239. http://dx.doi.org/10.1530/EJE-13-0894 2. Van den Berg SM, Setiawan A, Bartels M, et al. Individual differences in puberty onset in girls: Bayesian estimation of heritability’s and genetic correlations. Behav Genet 2006;36(2): 261-70. http://dx.doi.org/10.1007/s10519-005-9022-y 3. Burt Solorzano CM, McCartney CR. Obesity and pubertal transition in girls and boys. Reproduction 2010;140(3):399-410. http://dx.doi.org/10.1530/REP-10-0119 4. Marshall WA, Tanner JM. Variations in patterns of pubertal changes in girls.

Arch Dis Child 1969;44(235):291-303. http://dx.doi.org/10.1136/adc. 44. 235.291 5. Khadilkar VV, Stanhope RG. Secular trends in puberty. Indian Pediatr 2006;43(17):475-478.

6. Semiz S, Kurt F, Kurt DT, Zencir M, Sevinc O. Factors affecting onset of puberty in Denizli province in Turkey. Turk J Paediatr 2009;51(1):49-55.

7. Kaplowitz PB, Slora EJ, Wasserman RC, Pedlow SE, Giddens HME. Earlier onset of puberty in girls: Relation to increased body mass index and race. Pediatrics 2001;108(2): 347-351. http://dx.doi.org/10.1542/peds.108.2.347 8. Biro FM, Greenspan LC, Galvez MP, et al. Onset of breast development in a longitudinal cohort. Pediatrics 2013;132(6):1-9. http://dx.doi.org/10.1542/ peds.2012-3773

37

SAJCH

9. Wu T, Mendola P, Buck GM. Ethnic differences in the presence of secondary sexual characteristics and menarche among US girls. The Third National Health and Nutrition Examination survey 1988-1994. Pediatrics 2002;110(4):752-757. http://dx.doi.org/10.1542/peds.110.4.752 10. Atay Z, Turan S, Guran T, Furman A, Bereket A. Puberty and influencing factors in schoolgirls living in Istanbul: End of the secular trend? Pediatrics 2011;128(1):40-45. http://dx.doi.org/10.1542/peds.2010-2267 11. Saffari F, Rostamian M, Esmailzadoha N, Shariatinjad K, Karimzadeh T. Pubertal characteristics of girls in Quazvin Province Iran. Iran J Pediatr 2012;22(3):392-398.

12. Euling SY, Herman-Giddens ME, Lee PA et al. Examination of US pubertal timing data from 1940 - 1944 for secular trends: Panel findings. Pediatrics 2008;121(3):172-191. http://dx.doi.org/10.1542/peds.2007-1813D 13. Garnier D, Simondon KB, Beneface E. Longitudinal estimates of pubertal timing in Senegalese adolescent girls. Am J Hum Biol 2005;17(6):718-730. http://dx.doi. org/doi10.1002/ajhb.20435 14. Hosney LA, El-Ruby MO, Zaki ME, et al. Assessment of pubertal development in Egyptian girls. J Pediatr Endocrinol Metab 2005;18(6):577-584. https://doi. org/10.1515/JPEM.2005.18.6.577 15. Deardoff J, Molly F, Ekwaru JP, Kushi LH, Greenspan LC, Yen IH. Does neighbourhood environment influence girl’s pubertal onset? Findings from a cohort study. BMC Pediatr 2012;12(1):27. http://dx.doi.org/10.1186/1471-243112-27 16. Mutiagh M, Rabbani A, Kelishadi R et al. Timing of puberty in Iranian girls according to their living area – a national study. JRMS 2011;16(3):276-281.

17. Adesina AF, Peterside O. Age at menarche and body mass index (BMI) among adolescent secondary school girls in Port Harcourt, Nigeria. J Dent Med Sci 2013;3(5):41-46. https://doi.org/10.9790/0853-0354146 18. Onyiriuka AN, Egbagbe EE. Anthropometry and menarcheal status of adolescent Nigerian urban senior secondary school girls. Int J Endocrinol Metab 2013;11(2):71-75. http://dx.doi.org/10.5812/ijem.8052 19. Nzeako HC, Ezejindu DN, Iwunze AB, Obinna BN. Pubertal development in Nigerian Igbo girls. CIB Tech J Pharma 2015;4(2):45-51.

20. Country Quest. The people of Nigeria, Ethnicity. http://www.countriesquest. com/africa/nigeria/thepeopleofnigeria/ethnicity.htm (accessed 17 August 2016).

21. Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC growth charts for the United States: Methods and development. Vital Health Stat 2002;11(246):31-39.

22. Charan J, Biswas T. How to calculate sample size for different study designs in Medical Research. Indian J Psychol Med 2013;35(2):121-126. http://dx.doi. org/10.4103/0253-7176.116232 23. Cabrera SM, Bright GM, Frane JW, Blethren SL, Lee PA. Age at thelarche and menarche in contemporary US females: a cross-sectional analysis. J Pediatr Endocrinol Metab 2014;27(1-2):47-51. http://dx.doi.org/10.1515/jpem.2013-0286 24. Wang Y, Chen, H. Use of percentiles and z-scores in anthropometry. In: Preedy VR, ed. Handbook of Anthropometry: Physical Measures of Human Form in Health and Disease. Springer Science 2012:31-33. http://dx.doi.org/10.1007/9781-4419-1788-1-2 25. Bazrafshan H, Behnampour N, Sarabandi F, Mirpour S. Association between puberty and weight, height, and body mass index in a developing community.

JPMA 2012;62(5):454. http://dx.doi.org/10.1515/JPEM.2005.18.6.577 26. Rabbani A, Motlagh ME, Mohammed K, et al. Assessment of pubertal development in Iranian girls. Iran J Pediatr 2010;20(2):160-166.

27. Juul A, Teilmann G, Schike T, et al. Pubertal development in Danish children: Comparison of recent European and US data. Int J Androl 2006;29(1):247-55. http://dx.doi.org/10.1111/j.1365-2605.2005.00556.x 28. Jones LL, Griffiths PL, Cameron N. Is puberty starting earlier in urban South Africa? Am J Hum Biol 2009;21(3):395-397. http://dx.doi.org/10.1002/ajhb.20868 29. Pozo J, Argente J. Delayed puberty in chronic illness. Best Pract Res Clin Endocrinol Metab 2002;16(1):73-90. http://dx.doi.org/10.1053/beem.2002.0182 30. Garcia-Mayor RV, Andrade MA, Rios M, et al. Serum leptin levels in normal children: Relationship to age, gender, body mass index, pituitary-gonadal hormones, and pubertal stage. J Clin Endocrinol Metab 1997;82(9):2849-55. http://dx.doi.org/10.1210/jcem.82.9.4235 31. Ahmed ML, Ong KK, Morrell DJ, et al. Longitudinal study of leptin concentrations during puberty: Sex differences and relationship to changes in body composition. J Clin Endocrinol Metab 1999;84(3):899-905. http://dx.doi. org/10.1210/jcem.84.3.5559 32. He Q, Karlberg J. BMI in childhood and its association with height gain, timing of puberty and final height. Pediatr Res 2001;49(2):244-51. http://dx.doi. org/10.1203/00006450-200102000-0001

MARCH 2017 Vol. 11 No. 1

RESEARCH

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

Short-term and sustained effects of a health system strengthening intervention to improve mortality trends for paediatric severe malnutrition in rural South African hospitals M Muzigaba,1,2 PhD, MPhil, MPH, BSc; G Kigozi,3 PhD, MA, BA; T Puoane,2 Dr PH, MPH, BCur, BSocSci School of Clinical Medicine, College of Health Sciences, University of KwaZulu-Natal, Durban, South Africa Faculty of Community and Health Sciences, School of Public Health, University of the Western Cape, Cape Town, South Africa 3 Centre for Health Systems Research and Development, University of the Free State, Bloemfontein, South Africa 1 2

Corresponding author: M Muzigaba (mochemoseo@gmail.com) Background. Case fatality rates for childhood severe acute malnutrition (SAM) remain high in some resource-limited facilities in South Africa (SA), despite the widespread availability of the World Health Organization treatment guidelines. There is a need to develop reproducible interventions that reinforce the implementation of these guidelines and assess their effect and sustainability.

Objectives. To assess the short-term and sustained effects of a health system strengthening intervention on mortality attributable to SAM in two hospitals located in the Eastern Cape Province of SA. Methods. This was a theory-driven evaluation conducted in two rural hospitals in SA over a 69-month period (2009 - 2014). In both facilities, a health system strengthening intervention was implemented within the first 32 months, and thereafter discontinued. Sixty-nine monthly data series were collected on: (i) monthly total SAM case fatality rate (CFR); (ii) monthly SAM CFR within 24 hours of admission; and (iii) monthly SAM CFR among HIV-positive cases, to determine the intervention’s effect within the first 32 months and sustainability over the remaining 37 months. The data were analysed using Linden’s method for analysing interrupted time series data.

Results. The study revealed that the intervention was associated with a statistically significant decrease of up to 0.4% in monthly total SAM CFR, a non-statistically significant decrease of up to 0.09% in monthly SAM CFR within 24 hours of admission and a non-statistically significant decrease of up to 0.11% in monthly SAM CFR among HIV-positive cases. The decrease in mortality trends for both outcomes was only slightly reversed upon the discontinuation of the intervention. No autocorrelation was detected in the regression models generated during data analyses.

Conclusion. The study findings suggest that although the intervention was designed to be self-sustaining, this may not have been the case.

A qualitative enquiry into the moderating factors responsible for failure to sustain such an intervention, as well as the process of care, would add value to the findings presented in this study.

S Afr J Child Health 2017;11(1):38-45. DOI:10.7196/SAJCH.2017.v11i1.1201

Severe acute malnutrition (SAM) in children aged 6 - 59 months remains a public health problem worldwide.[1] Childhood SAM is defined as a weight-for-height <–3 z-score of the median of the World Health Organization (WHO) growth standard,[2] or the presence of clinical signs of bilateral pitting oedema of nutritional origin (oedematous malnutrition), despite other measures being above specified cut-off values.[3] The 2016 joint report on child malnutrition estimates, which was developed by the World Bank, WHO and the United Nations Children’s Fund (UNICEF),[4] indicates that although stunting has declined over the past 5 years, it remained as high as 23.2% in 2015. However, wasting has risen slightly to 7.4% within the same period. The joint report also indicates that in 2015, more than half (56%) of all stunted under-5 children lived in Asia and more than one third (37%) lived in Africa. Regarding wasting, 68% of all wasted under-5 children lived in Asia whereas 28% lived in Africa.

The mortality rate of undernourished children is much higher than their well-nourished counterparts.[5] Children with SAM also do not respond to medical treatment in the same way as well-nourished children.[6] Special guidelines for treating severely malnourished children are therefore required because of the peculiar pathophysiological and metabolic changes that the undernourished body undergoes. The reductive adaptation that occurs in SAM requires specialised management, and practitioners involved with the rehabilitation of SAM cases should be aware of this delicate homeostatic mechanism.[7] 38

SAJCH

The WHO 10-step guidelines for management of severe malnutrition were developed for this purpose.[8] The guidelines are currently promoted worldwide as the standard by which children with SAM should be treated.[9] Some studies have shown that with strict adherence to these guidelines, mortality can be reduced to <15%.[10] In 2004, Puoane et al.[11] conducted a study in rural hospitals in South Africa (SA) to explore why some hospitals achieved better outcomes for severely malnourished children than others. They concluded that the WHO clinical guidelines and external training are valuable, but they may be insufficient to ensure quality of care. They identified a need for a self-sustaining programme that should focus on continuous in-service training for healthcare workers, supervision and audit of the process of care, and parallel interventions for senior hospital and clinical managers to foster leadership and teamwork.

They also concluded that for the WHO guidelines to succeed, tasks have to be performed assiduously and consistently by the frontline healthcare providers.

In response to the recommendations from Puoane et al.,[11] we developed, implemented and evaluated a comprehensive health system strengthening intervention in two of the district hospitals where their study was conducted. In contrast to previous research that evaluated similar interventions,[10] we used a robust impact evaluation design – the interrupted time series (ITS) design – with a view to assessing the intervention’s short-term effects on mortality levels and trends attributable to SAM, as well as the sustainability of

MARCH 2017 Vol. 11 No. 1

RESEARCH these effects after completion of the research. More specifically, the objectives of the study were as follows: • to develop a package of interventions targeting healthcare workers, hospital managers and administrators, and other support structures within and outside two purposefully selected hospitals, with a view to improving the organisational structures, health workforce capacity and processes of care • to implement the same intervention in each of the two district hospitals over a specific period of time • to discontinue the intervention and assess, in each hospital, whether the discontinuation would result in a shift in the levels and trends of case fatality rates (CFRs) attributable to SAM compared with levels and trends in the pre-discontinuation period • to compare the temporal changes in CFRs attributable to SAM between the two hospitals before and after the intervention was discontinued.

Methods Setting

The study was conducted in two rural district hospitals located in the Eastern Cape Province (EC) that were already implementing the WHO treatment guidelines.[11] In this article, the two hospitals are referred to as hospitals A and B. The hospitals were selected based on the fact that they had participated in the initial province-wide intervention to improve the management of SAM in the EC and had been implementing the WHO 10-step guidelines more effectively than other hospitals in the region. However, hospital-level statistics at the time of the current study had revealed that the annual CFRs attributable to SAM had increased in excess of 25% in hospital A and 32% in hospital B since the last intervention by Puoane et al.,[11] which was implemented in 2004 in the same facilities.

Inputs/structures

• Amenities/facilities • Medical equipment, therapeutic resources • Hospital support services • Financial resources and incentives • Therapeutic guidelines (WHO management protocol)

• Trained healthcare personnel • More effective organisational structure • More effective communication structures • Better links to external support structures

Processes/activities

Standard of care

The standard of care consisted of routine care processes, such as: patient diagnosis on arrival at the facility; documentation of clinical/ medical history; disease classification; HIV testing and disease staging; documentation of the treatment protocol to be followed and admission of the patient to the ward as required; and treatment of the patients by the health worker using the WHO 10-step guidelines.

The inputs and infrastructure included: available amenities/facilities; medical equipment; therapeutic resources; hospital support services; financial resources and incentives; and therapeutic guidelines (WHO management protocol).

Study intervention

The study intervention was modelled on a similar clinical guideline-based intervention implemented in Kenya as part of a cluster randomised trial to improve paediatric care and outcomes in Kenyan hospitals. The intervention was designed to ameliorate treatment outcomes through improved organisational structures, health workforce capacity, and processes of care.[12] The current study intervention was designed to achieve similar goals, but was adapted to the local context and the disease of interest. In addition to the standard of care, the current study intervention included inputs and processes as described in Table 1. The components of the study intervention were introduced in each hospital from January 2009 to August 2011.

Intervention theory of change

The intervention theory of change for the current study is summarised in Fig. 1. The components of the study intervention and the standard of care are all embedded in the theory of change as shaded and

Outputs

Diagnosis, documentation of clinical/medical history, disease classification, HIV testing and disease staging (upon arrival at the hospital)

Patient diagnosis, HIV testing is done according to standard guidelines

Documentation of disease condition, treatment protocol and admission of the patient

Accountability is improved through documentation of the process of care

Patient treatment according to the WHO 10-step guidelines by doctors and nurses

Process of care is standardised

Continuous on-the-job training of healthcare workers involved in the management of SAM

Increased number of competent healthcare workers

Regular hospital visits to reinforce implementation of the guidelines

Implementation of the WHO guidelines is enforced

Regular face-to-face feedback with hospital staff to report on performance and discuss areas needing improvement

Key hospital staff receive feedback and develop plans for improvement

Externally initiated on-site didactic training of healthcare professionals and supervision of healthcare providers

Healthcare providers are more competent in the management of SAM

Induction training with new doctors and new nurses on the principles of using the WHO guidelines to treat SAM children

New healthcare providers are aware of the treatment protocol for SAM

Provision of job aids such as drug dose charts, fluid and feed charts, and basic equipment

Improved process of care

Regular advocacy meetings with hospital management and support structures

Improved organisational structure, leadership and teamwork

Short-term outcomes

Medium-term outcomes

Long-term outcomes (sustainability)

Total SAM CFR is reduced to 10% midway during the implementation of the intervention

Total SAM CFR remains at 5% after intervention discontinuation

Total SAM CFR is reduced to 5% before intervention discontinuation

CFR within 24 hrs of admission is reduced to 5% midway during the implementation of the intervention

CFR due to SAM and HIV infection remain at 5% after discontinuation of intervention

SAM CFR within 24 hrs of admission remains at 0% after intervention discontinuation

SAM CFR within 24 hrs of admission is reduced to 0% before discontinuation of intervention

CFR due to SAM and HIV infection is reduced to 5% before discontinuation of intervention

CFR due to SAM and HIV infection is reduced to 10% midway during the implementation of the intervention

Fig. 1. Intervention theory of change with activities for both the standard of care and the study intervention. Shaded boxes indicate study intervention activity components and unshaded boxes indicate components of the standard of care.

39

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH unshaded boxes, respectively. As shown in Fig. 1, each activity was designed to be executed using specific resources and to generate specific outputs which, as a collective, would result in a cascade of outcomes (short-, medium- and long-term outcomes). Each level of outcome, whether short-, medium- or long-term, had specific targets to be met during the evaluation period.

Study design

This study was approved by the University of the Western Cape Research Ethics Committee (reg. no. 12/10/37). It involved an ITS study design, which has been shown to be a rigorous quasiexperimental alternative to randomised control trials when the latter are not feasible to conduct and time series data are available.[13] As far as could be ascertained, the ITS design has not been used before to specifically assess the impact of discontinuing a facility-based nutrition rehabilitation intervention involving the WHO clinical guidelines for management of SAM. Some scientists have also argued that the ITS is the strongest quasi-experimental design that can be used to evaluate longitudinal effects of time-defined interventions.[14,15] The method usually involves the measurement of a variable (or variables) of interest before and after the introduction of specific interventions to assess whether the intervention has had an impact on that variable over time.[14,15] The impact of the intervention can then be measured by assessing the level and trend (slope) changes of that variable over time, which are presumed to be affected by the presence of the intervention alone.

During this study, we used the same ITS notion. However, we assessed whether discontinuing the study intervention, instead of introducing it, would have an effect on the level and trend of three specific mortality outcomes, i.e: (i) total monthly SAM CFR; (ii) monthly SAM CFR within 24 hours of admission; and (iii) monthly SAM CFR among cases with HIV infection. The calculation of these outcomes is explained further below. The resultant study design was therefore an ITS design with an embedded ‘removed intervention design’.

The aim of the removed intervention design was to demonstrate that mortality outcomes improved or worsened with the presence or absence of an intervention – a result that could be otherwise explained only by a threat to validity that similarly rose and fell over the same time period.[15] Fig. 2 illustrates the removed intervention design used in this study.

The same intervention was implemented in each hospital over a period of 32 months (from January 2009 to July 2011). Thereafter, the study intervention was discontinued (solely because of the end of the funding period), but mortality outcomes of interest were monitored for a further 37 months (from August 2011 to September 2014). The sustainability of the study intervention was therefore determined from the pattern of mortality outcomes in the period after July 2011, when components of the study intervention that required researchers’ input were no longer active. During this period, both hospitals were presumed to be managing SAM cases independently.

This approach was applied to time series data from both hospitals so that results could be compared. The use of a second group in

Table 1. Activity components of the study intervention that were added to the standard of care Component Three-monthly hospital visits conducted in each facility by the research team Six-monthly feedback sessions (performance review) presented by the research team to the clinical and management staff in each hospital Regular induction sessions conducted by the research team for newly appointed nurses or nurses rotating in the paediatric ward Provision of job aides. These included: • medication dosage chart to use during the administration of antibiotics, multivitamins, electrolytes and antiretrovirals as applicable;

Purpose Assess patient care and treatment outcomes using a standardised questionnaire Report on findings based on the previous visit Introduce nursing staff to the use of the guidelines and the study’s information system acilitate patient care practices and documentation of treatment F procedure and outcomes

• fluid administration chart • output chart to keep track of the patterns of diarrhoea, vomiting and urine discharge • weight monitoring chart • temperature and pulse monitoring chart • oral rehydration solution (ORSOL) chart Provision of basic equipment such as weighing scales and length/height measurement apparatus Three-monthly external supervision visits conducted by the research team over a period of 2 years during which the study intervention was active Baseline identification of a full-time senior paediatric nurse based in the paediatric ward in each hospital. The paediatric nurses were continuously mentored to sustain the intervention during and after the study period Facilitating links between the hospitals and other support structures, such as laboratory services and blood banks Six-monthly advocacy meetings with the hospital management team Linkage with a facility-based social worker, who initiated a follow-up process with community-based and government departments

40

SAJCH

Monitoring of nutritional recovery To mimic the role of a clinic-based ward supervisor – in this case a senior nurse in the paediatric ward Senior nurses acted as mentors for junior nurses and newly appointed or rotating nurses. The mentors introduced these nurses to the utilisation of the intervention components over and above their routine duties, such as solving on-site problems and managing patients Improve operational efficiency during emergency situations To alert hospital management to issues hampering optimal patient care and some of the ways to improve the status quo, as well as targets for improvement To ensure that SAM patients who are discharged receive adequate support to prevent relapses and readmissions to the hospital

MARCH 2017 Vol. 11 No. 1

RESEARCH for that month. Thus the monthly outcome statistics used in this evaluation were computed as follows:

O1 Baseline observation in Jan. 2009 before introducing the study intervention

Introduction of the study intervention in Jan. 2009

O2

On2

32 post-intervention observations measured monthly from Feb. 2009 to Aug. 2011

Discontinuation of the intervention as of Aug. 2011 (‘removed treatment’)

O3

On3

37 post-discontinuation observations measured monthly from Sep. 2011 to Sep. 2014

Fig. 2. Study design, showing ITS design with an embedded removed intervention design.

ITS analysis has been encouraged by some researchers.[13,16] However, in our study, both hospitals received the same intervention for comparison purposes, unlike in traditional multiple-group ITS, where control groups do not receive the intervention.

Study outcomes and measurement

To enhance the internal validity of the study, monthly mortality statistics for SAM were recorded objectively by a trained research nurse in each hospital, using a standardised performance monitoring tool developed by the research team. Data recorded included, among others, number of: total ward admissions; ward deaths; ward admission due to SAM; deaths due to SAM; deaths due to SAM among HIV-positive cases; deaths due to SAM within 24 hours; and children admitted with SAM receiving child support grants. In this study SAM was defined at admission as per the Wellcome classification system.[17] It was also important that the data collection process was not influenced by the intervention. Therefore, the research nurses were blinded to the aim of the study and the use of the data collected over the study period. Furthermore, the composition of the dataset at each time point measured in the study consisted of at least 95% of SAM cases that were treated in each hospital over the study period.

At the end of the study period (November 2014) the research team used a standardised registry abstraction form to gather all the data collected by research nurses over the study period. The data pertinent to this study were retrospectively verified, as far as possible, for precision, timeliness, comprehensiveness, validity and accuracy. This dataset included the total number of admissions due to SAM, number of SAM admissions with HIV infection, total number of deaths due to SAM, total number of deaths due to SAM that occurred within 24 hours of admission, and total number of deaths due to SAM among HIV-positive cases. Using these metrics, the following three study outcomes were calculated: • total monthly SAM CFR • monthly SAM CFR within 24 hours • monthly SAM CFR among HIV-positive cases.

Since the CFR for either outcome in any given month was dependent on the number of patients admitted with SAM in that month, it was important to transform each outcome so that the monthly CFRs of interest were weighted against the corresponding denominator 41

SAJCH

• Weighted monthly total SAM CFR = [Deaths due to SAM (N)/SAM admissions (N) × 100 ] × weighted score • Weighted monthly SAM CFR within 24 hours = [Deaths due to SAM (N)within 24 hours of admission/SAM admissions (N) × 100] × weighted score • Weighted monthly SAM CFR among HIV-positive cases = [SAM deaths (N)with HIV comorbidity/SAM admissions (N) with HIV comorbidity × 100] × weighted score

Data analysis

The data were transformed and analysed in Stata 13.0 (StataCorp., USA) using a widely used method developed by Linden.[17] The ITS analysis involved two groups in which data from the two hospitals were compared to determine whether there were similarities in parameter estimates of interest. The key assumption underlying the two-group ITS analysis was that the change in the level and/ or trend in the three outcome variables was the same for both the control group (in this case hospital A), and the other group (in this case hospital B). It was assumed that the confounding variables affected the two hospitals similarly.[17] The major strength of the twogroup ITS analysis is therefore its ability to test for comparability between the two hospitals, thereby supporting the predicted effects or lack thereof.

In this study, the regression model equation for the two-group analysis took the following form, which is detailed in the article by Linden[17] where:

Yt= β0 + β1Tt + β2Xt + β3XtTt + β4Z + β5ZTt + β6ZXt + β7ZXtTt + εt Yt = o utcome variable measured every month t during the study period Tt = t ime since the start of the study in January 2009 Xt =d ummy variable representing the presence or absence of the intervention (in this study, the intervention period = 0 and the period after the intervention discontinuation = 1) Z = a dummy variable to denote the group assignment (hospital A or B) XtTt, ZTt, ZXt, and ZXtTt = interaction terms β0 = the intercept (starting level of the outcome variable) β1 = t he slope (trajectory of the outcome variable until the ‘removal’ of the intervention) β2 = t he change in the level of the outcome that occurs in the period immediately after the removal of the intervention β3 = t he difference between pre- and post-intervention discontinuation slopes for the outcome β4 = t he difference in the level (intercept) of the outcome variable between the two hospitals before the discontinuation of the intervention β5 = t he difference in the slope (trend) of the outcome variable between the two hospitals prior to the discontinuation of the intervention β7 = the difference between the two hospitals in the slope (trend) of the outcome variable after the discontinuation of the intervention.

The ordinary least squares (OLS) regression model was estimated for each outcome variable, with the Newey-West standard errors to handle autocorrelation and possible heteroscedasticity. The CumbyHuizinga test was used to verify whether the model estimates accounted for the correct autocorrelation structure. Model estimates were also plotted for visual inspection of actual and predicted trends and levels in the outcome variables before and after the intervention was discontinued.

MARCH 2017 Vol. 11 No. 1

RESEARCH

Intervention discontinued: 32nd month

70 90 60 80 50 70 40 60 30 50 20 40 10 30 10 0

5

9

13 17

21 25 29

5

9

13 17

21 25 29

33 37 41

45 49

53 57

61 65

33 37 41

45 49

53 57

61 65

Study period (months)

Sep. 2014 Sep. 2014

0 20

Hospital A: Hospital B:

Actual weightedStudy CFR (%) Model-predicted weighted CFR (%) period (months) Actual weighted CFR (%) Model-predicted weighted CFR (%)

Hospital A: Hospital B:

Actual weighted CFR (%) Model-predicted weighted CFR (%) Regression with Newey-West standard errors - lag(1) Actual weighted CFR (%) Model-predicted weighted CFR (%)

Fig. 3. Level and slope changes in weighted total monthly Regression with Newey-West standard errors - lag(1) SAM CFRs: hospital-level graphical visualisation with model-predicted line plots and scatter plots of the actual values.

100 90

Intervention discontinued: 32nd month

100 80 90 70

Intervention discontinued: 32nd month

80 60 70 50 60 40 50 30 40 20 30 10

The ITS design has been widely documented as a powerful quasiexperimental design that can be used to evaluate the effects of interventions when random assignments are not feasible.[14,15,18] This

9

13 17

21

25 29

5

9

13 17

21

25 29

33 37 33 37

41

45 49

53

57 61

65

41

45 49

53

57 61

65

Study period (months)

Hospital A: Hospital B:

Actual weightedStudy CFR (%) Model-predicted weighted CFR (%) period (months) Actual weighted CFR (%) Model-predicted weighted CFR (%)

Hospital A: Hospital B:

Actual weighted CFR (%) Model-predicted weighted CFR (%) Regression with Newey-West standard errors - lag(1) Actual weighted CFR (%) Model-predicted weighted CFR (%)

Sep. 2014 Sep. 2014

5

Aug. 2011Aug. 2011

Jan. 2009 Jan. 2009

0

Regression with Newey-West standard errors - lag(1)

Fig. 4. Level and trend in weighted monthly SAM CFR within 24 hours: hospital-level visualisation of model-predicted line plots and scatter 100 plots of the actual values.

90 Intervention discontinued: 32nd month 100 80 study used the ITS design to assess the short-term and sustained effects

90 Intervention discontinued: 32nd month of70 a health system strengthening intervention on mortality attributable 80 to60 SAM in two hospitals located in the EC. Traditional experimental and 70 quasi-experimental research in public health has involved testing 50 the 60 effectiveness of an intervention following its introduction, but 40 without due regard to the sustainability once it has been discontinued.

50 30 This study therefore used a novel approach in the area of SAM, which 40 20 can be explored further in future research using other performance 30 10 indicators of choice.

20 0 This study revealed that the intervention was associated with a 10 5 trend 9 13 17 three 21 25 29 33 outcomes 37 41 45 during 49 53 the 57 intervention 61 65 downward in all mortality 0 period in both hospitals. However, this effect was not statistically 9 13for17the 21total 25 monthly 29 33 37SAM 41 45 57 61 B,65 significant,5 except CFR49in53hospital which was significantly reduced during the intervention period.

Study period (months) Hospital A:

Actual weightedStudy CFR (%) Model-predicted weighted CFR (%) period (months) Model-predicted weighted CFR (%)

Hospital A: Hospital B:

Actual weighted CFR (%) Model-predicted weighted CFR (%) Regression with Newey-West standard errors - lag(1) Actual weighted CFR (%) Model-predicted weighted CFR (%)

CFR (%) MARCH 2017 Hospital Vol. B:11 Actual No. weighted 1

Sep. 2014 Sep. 2014

Fig. 5 and Table 2 show that the baseline level of the weighted monthly SAM CFR among HIV-positive cases in both hospitals was ~12%, which declined at the rate of 0.11% every month in hospital B during the intervention period, and at a monthly rate of 0.05% in hospital A. The trend estimates post discontinuation of the intervention show that mortality trends remained fairly constant. There was a very slight and not statistically significant monthly increase in mortality of 0.01% in hospital A, and a negligible monthly decrease of 0.03% in hospital B.

10

Aug. 2011Aug. 2011

Weighted monthly SAM CFR among HIV-positive cases

SAJCH

80 100

Jan. 2009 Jan. 2009

As shown in Fig. 4, during the first month before the introduction of the intervention, the weighted monthly SAM CFR within 24 hours was again higher in hospital B (5%) compared with hospital A (3.7%), but this difference was not statistically significant (p=0.601).

Early CFR declined steadily every month in both hospitals during the intervention period, but the difference in slope between the two hospitals was not statistically significant (β=0.02, CI –0.22 - 0.26; p=0.861). The trend analysis post intervention discontinuation shows that both hospitals experienced a very negligible monthly increase in early CFR. Hospital A had a month-to-month increase of 0.01% and hospital B an increase of 0.02%. These results show that levels and trends in the weighted monthly SAM CFR within 24 hours were statistically comparable between the two hospitals during the study period.

42

Intervention discontinued: 32nd month

90

20 0

Weighted monthly SAM CFR within 24 hours

Discussion

100

Aug. 2011Aug. 2011

The regression model coefficients, 95% confidence intervals (CIs) and p-values for (i) weighted total monthly SAM CFR; (ii) weighted monthly SAM CFR within 24 hours; and (iii) weighted monthly SAM CFR among HIV-positive cases are presented in Table 2, and the corresponding graphical visualisations of level and trend changes are shown in Figs 3, 4 and 5, respectively. The Cumby-Huizinga test revealed no autocorrelation at any of the 69 lags for the regression models fitted on all three outcome variables.

The model-predicted line plot in Fig. 3 shows that the starting level of the weighted total monthly SAM CFR was about 17% in hospital B, which was 4% higher than in hospital A. However, this difference was not statistically significant (CI –3.36 - 4.62; p=0.338), as shown in Table 2. There was a downward trend in weighted total monthly SAM CFR in both hospitals during the intervention period, with small differences that were not statistically significant (β=0.09, CI –0.35 - 0.54; p=0.668). Contrary to hospital A, hospital B had a statistically significant monthly reduction of 0.4% in mortality during the period within which the intervention was active (β=–0.40; CI –0.76 - –0.04; p=0.028). The level and slope changes for both hospitals during the intervention period are shown in Fig. 3. The trajectory in mortality in both hospitals appears to have moderately fallen during this period.

The results in Table 2 and Fig. 3 also show that after the discontinuation of the intervention, there was a very small and not statistically significant increase in the weighted total monthly SAM CFR in both hospitals. The CFR increased slightly every month by 0.04% in hospital A (β=0.04; CI –0.10 - 0.18; p=0.583) and by 0.07% in hospital B (β=0.07; CI –0.08 - 0.23; p=0.358). The difference in slope between the two hospitals after discontinuing the intervention compared with the intervention period was very small (–0.03%) and not statistically significant (p=0.752). Again, these results indicate that the trends in CFR were comparable between the two hospitals throughout the study period.

Jan. 2009 Jan. 2009

Results

RESEARCH Table 2. Two-group interrupted time series regression models for the three outcome variables

Outcomes and impact parameters Weighted total monthly SAM CFR Difference in level between the two hospitals prior to the discontinuation of the intervention Trend (slope) during the intervention period for hospital B Difference in the slope between the two hospitals prior to the discontinuation of the intervention Difference in level between the two hospitals in the period immediately following the discontinuation of the intervention Difference in slope between the two hospitals after discontinuation of the intervention compared with the intervention period Trend estimates after discontinuation of the intervention Hospital A Hospital B Trend difference between hospital A and hospital B Weighted monthly SAM CFR within 24 hours Difference in level between the two hospitals prior to the discontinuation of the intervention Trend during the intervention period for hospital B Difference in the slope between the two hospitals prior to the discontinuation of the intervention Difference in level between the two hospitals in the period immediately following the discontinuation of the intervention Difference in slope between the two hospitals after discontinuation the intervention compared with the intervention period Trend estimates post-discontinuation of the intervention Hospital A Hospital B Trend difference between hospital A and hospital B Weighted monthly SAM CFR among HIV-positive cases Difference in level between the two hospitals prior to the discontinuation of the intervention Trend (slope) during the intervention period for hospital B Difference in the slope between the two hospitals prior to the discontinuation of the intervention Difference in level between the two hospitals in the period immediately following the discontinuation of the intervention Difference in slope between the two hospitals after discontinuing the intervention compared with the intervention period Trend estimates post-discontinuation of the intervention Hospital A Hospital B Trend difference between hospital A and hospital B

By design, the trend in mortality during the intervention segment, projected into the period during which the intervention was not active, served as a counterfactual.[16] Graphical inspections and regression model estimates for all three mortality outcomes showed that the effects realised during the intervention, although not always statistically significant, were slightly reversed after the intervention was discontinued, but the reversal was very minimal compared with the trends in the intervention segment and not statistically significant.

This finding can also be explained in light of the actual targets set for this study. The theory of change for the intervention posited that the study intervention would reduce the total SAM CFR to 10%, the SAM CFR in 24 hours to 5% and the SAM CFR among HIVpositive cases to 10% midway during the implementation of the project. When considering the model-predicted baseline levels for each outcome, as shown in Figs 3, 4, and 5, these short-term targets were met. The medium-term targets involved a reduction in total SAM CFR to 5%, the SAM CFR in 24 hours to 0% and the SAM CFR among HIV-positive cases to 5% just before the discontinuation 43

SAJCH

β

95% CI

p-value

–4.37 –0.40 0.09

–13.36 - 4.62 –0.76 - –0.04 –0.35 - 0.54

0.338 0.028 0.668

2.62

–6.28 - 11.52

0.561

–0.13

–0.63 - 0.36

0.601

0.04 0.07 –0.03

–0.10 - 0.18 –0.08 - 0.23 –0.25 - 0.18

0.583 0.358 0.752

–1.26 –0.09 0.02

–6.15 - 3.62 –0.31 - 0.12 –0.22 - 0.26

0.610 0.382 0.861

1.02

–3.56 - 5.60

0.660

–0.04

–0.31 - 0.23

0.787

0.01 0.02 –0.01

–0.04 - 0.06 –0.06 - 0.11 –0.12 - 0.09

0.654 0.569 0.781

–0.38 –0.11 –0.08

–14.27 - 13.50 –0.67 - 0.46 –0.79 - 0.61

0.956 0.706 0.801

1.28

–13.71 - 16.28

0.865

0.13

–0.69 - 0.96

0.742

0.01 0.03 0.05

–0.28 - 0.32 –0.33 - 0.27 –0.38 - 0.48

0.901 0.848 0.823

of the intervention. Again, based on the model-predicted mortality levels shown in Figs 3 and 4, the first two targets were met in both hospitals. However, the target of reducing SAM CFR among HIVpositive cases to 10% was not met at hospital A, unlike at hospital B, as shown in Fig 5. The theory of change also posited that the levels of mortality for all three study outcomes – which were realised before the intervention was discontinued – would be maintained during the period when the intervention was not active; however, these targets were not met, indicating that the presence of the intervention had a positive effect on the outcomes of interest.

Despite the lack of statistical significance, however, the predicted changes in mortality trends during the intervention period demonstrate the possible effect of the intervention on mortality during this period. To our knowledge, the intervention occurred independently of other changes to the healthcare milieu during the study period. Furthermore, the shifts in mortality trends and levels were similar in both hospitals. This lends support to the view that the effects reported in this study were unbiased and associated with the

MARCH 2017 Vol. 11 No. 1

Study period (months) Hospital A: Hospital B:

Actual weighted CFR (%) Actual weighted CFR (%)

Model-predicted weighted CFR (%) Model-predicted weighted CFR (%)

Regression with Newey-West standard errors - lag(1)

RESEARCH autocorrelation and involved two comparable groups, a design which is encouraged in ITS analysis. The study involved a relatively wellconceived intervention design which was informed by the literature and practical experience.

100 90

Intervention discontinued: 32nd month

80 70

Study limitations

60 50 40 30 20 10 0 13 17 21

25

29 33 37

41 45

49 53

57 61

65 Sep. 2014

Aug. 2011

9

Jan. 2009

5

Study period (months) Hospital A: Hospital B:

Actual weighted CFR (%) Actual weighted CFR (%)

Model-predicted weighted CFR (%) Model-predicted weighted CFR (%)

Regression with Newey-West standard errors - lag(1)

Fig. 5. Level and trend in the weighted monthly SAM CFR among HIVpositive cases: hospital-level visualisation of model-predicted line plots and scatter plots of the actual values. corresponding presence of the intervention.[16] The lack of statistically significant reversal in the effects realised during the intervention period for all three outcomes is unlikely related to the number of data points used to estimate the effects. There is evidence to show that the minimum number of data points required to detect the effect is 12, before and after series interruption – in this case the discontinuation of the intervention.[19] The current study involved 69 data points in total, 32 of which were used in the pre-intervention discontinuation segment and 37 in the segment following the discontinuation. It is therefore possible that the effects were reversed, but not to levels that would have shown statistical significance.

An ethnographic study would have provided more elaborate insights into the process of care and the dynamics within the healthcare environment throughout the study intervention, to help validate the estimates reported here. Taljaard et al.[18] have cautioned about using ITS design to evaluate disease management interventions in healthcare facilities, as the process of care may involve subtle changes at individual level and different intervals, which may affect outcomes of interest.

This study raises an important research topic that can be investigated in future studies. The study intervention was designed to be self-sustaining through, to mention but a few, the standardisation of the process of care, provision of job aids and quality assurance resources, establishment of a healthcare monitoring infrastructure, as well as training and mentoring of a paediatric ward champion (paediatric nurse) to enforce the implementation of the intervention components and act as a healthcare advocate and the go-to health professional for relatively junior nurses. It was unclear whether the intervention alone or the involvement of the research team, or both, had an effect on the outcomes observed during the intervention period, as the withdrawal of the research team saw a reversal in the gains made during this period.

Study strengths

One of the strengths of this study design was its potential to control for selection bias as much as a randomised control trial.[15] The study outcomes were measured objectively and constituted at least 95% of SAM patients who were admitted during the study period.

Furthermore, data analysis revealed that the model was free of 44

SAJCH

The fact that the OLS regression method used in this study assumes a linear trend over time is a possible limitation to consider in light of the reported findings. Despite the measures put in place to verify the quality of the data used in this study, it was not ascertained beyond doubt that all the patients included in the dataset died solely as a result of SAM. Both hospitals receive large number of patients with SAM with HIV comorbidities, and it is possible that some patients may have died of HIV/AIDS rather than SAM. The use of covariates in the model would have been valuable to address this potential problem. The study was not designed to contextualise the trends in mortality during the study period, in terms of the process of care.

This would have added value to findings reported here by suggesting some of the reasons why, for example, the intervention effects were not sustained after the intervention was discontinued.

Conclusion

The study showed that the presence of the intervention was associated with non-statistically significant monthly decreases in total CFR and early CFR associated with SAM, and that the discontinuation of the intervention reversed the effects slightly. These findings suggest that although the intervention was designed to be self-sustaining, this may not have been the case. A qualitative ethnographic enquiry into the process of care throughout the study period and the moderating factors responsible for the failure in sustaining such an intervention are encouraged in future research to substantiate the findings presented in this evaluation.

Acknowledgements. This research was made possible by a grant from the South African National Research Foundation (NRF) and the South African Centre for Epidemiological Modelling and Analysis (SACEMA). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the funders. Special thanks also go to all the staff from the two hospitals who contributed to the study.

Authors’ contributions. TP and MM were involved in the conception of the study and MM was responsible for the acquisition, analysis and interpretation of data, as well as the drafting of the manuscript and finalisation for submission. Both authors have also given final approval of the version to be published and agree to be accountable for all aspects of the work.

1. United Nations Children’s Emergency Fund. Global Child Malnutrition Trends (1999 - 2012). New York: UNICEF, 2013 http://www.childinfo.org/malnutrition_ dashboard.html (accessed 9 January 2016).

2. World Health Organization. Management of Severe Malnutrition: A Manual for Physicians and other Senior Health Workers. Geneva: WHO, 1999. http:// whqlibdoc.who.int/hq/1999/a57361.pdf (accessed 12 January 2016).

3. World Health Organization. WHO Child Growth Standards and the Identification of Severe Acute Malnutrition in Infants and Children. A Joint statement by the World Health Organization and the United Nations Children’s Fund. Geneva: WHO, 2009. http://apps.who.int/iris/bitstream/10665/44129/1/9789241598163_ eng.pdf (accessed 10 December 2015).

4. United Nation Children’s Fund/World Health Organization/World Bank. Levels and trends in child malnutrition. UNICEF/WHO/World Bank Group joint child malnutrition estimates. 2016. http://data.unicef.org/wp-content/uploads/2016/09/ UNICEF-Joint-Malnutrition-brochure.pdf (accessed 6 Oct-ober 2016).

5. Pelletier DL, Frongillo EA, Schroeder DG, Habicht JP. The effects of malnutrition on child mortality in developing countries. Bull World Health Organ 1995;73(4):443-448.

6. Jackson A, Ashworth A, Khanum S. Improving child survival: Malnutrition task force and the paediatrician’s responsibility. Arch Dis Child 2006;91(8):706-710. https://doi.org/10.1136/adc.2006.095596 7. Waterlow JC. Metabolic adaptation to low intakes of energy and protein. Annu Rev Nutr 1986;6:495-526. https://doi.org/10.1146/annurev.nutr.6.1.495

MARCH 2017 Vol. 11 No. 1

RESEARCH 8. Heikens GT. How can we improve the care of severely malnourished children in Africa? Plos Med 2007;4(2):e45. https://doi.org/10.1371/journal.pmed.0040045 9. Schub C. Management of severe malnutrition. S Afr J Clin Nutr 2010;23(1). https://doi.org/10.1080/16070658.2010.11734264 10. Ashworth A, Khanum S, Jackson A, Schofield C. Guidelines for the Inpatient Treatment of Severely Malnourished Children. New Delhi: World Health Organization Regional Office for South-East Asia, 2003. http://www.who.int/ nutrition/publications/guide_inpatient_text.pdf (accessed 2 June 2016). https:// doi.org/10.1177/156482650502600215 11. Puoane T, Cuming K, Sanders D, Ashworth A. Why do some hospitals achieve better care of severely malnourished children than others? Five-year follow-up of rural hospitals in Eastern Cape, South Africa. Health Policy Plan 2008;23(6):428437. https://doi.org/10.1093/heapol/czn036 12. Ayieko P, Ntoburi S, Wagai J, et al. A multifaceted intervention to implement guidelines and improve admission paediatric care in Kenyan district hospitals: A cluster randomised trial. PLoS Med 2011;8(4):e1001018. https://doi.org/10.1371/ journal.pmed.1001018 13. Biglan A, Metzler CW, Ary DV. Increasing the prevalence of successful children: The case for community intervention research. Behav Anal 1994;17(2):335-351.

45

SAJCH

14. Gillings D, Makuc D, Siegel E. Analysis of interrupted time series mortality trends: An example to evaluate regionalized perinatal care. Am J Public Health 1981;71(1):38-46. https://doi.org/10.2105/ajph.71.1.38 15. Shadish RW, Cook DT, Campbell DT. Experimental and Quasi Experimental Designs for Generalized Causal Inference. 2nd ed. Boston: Houghton Mifflin Harcourt, 2002.

16. Linden A. Conducting interrupted time-series analysis for single- and multiple-group comparisons. Stata J 2015;15(2):480-500.

17. Murgod R, Ahmed M. Instant nutrition assessment in children with protein energy undernutrition. Int J Appl Bio Pharma Tech 2015;6(1):171-177.

18. Taljaard M, McKenzie, JE, Ramsay, CR, Grimshaw, JM. The use of segmented regression in analysing interrupted time series studies: An example in prehospital ambulance care. Implement Sci 2014;19(9):77-80. https://doi. org/10.1186/1748-5908-9-77 19. Ramsay CR, Matowe L, Grilli R, Grimshaw JM, Thomas RE. Interrupted time series designs in health technology assessment: Lessons from two systematic reviews of behavior change strategies. Int J Technol Assess Health Care 2003;19(4):613-623. https://doi.org/10.1017/s0266462303000576

MARCH 2017 Vol. 11 No. 1

RESEARCH

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

Independent and interactive effects of HIV infection, clinical stage and other comorbidities on survival of children treated for severe malnutrition in rural South Africa: A retrospective multicohort study M Muzigaba,1,2 PhD, MPH, MPhil, BSc; T Puoane,2 Dr PH, MPH, BCur, BSocSci; B Sartorius,3 PhD, EPIET, MSc, BSc Hons, BSc; D Sanders,2 MB ChB, DCH, MRCP, DTPH, DSc School of Clinical Medicine, College of Health Sciences, University of KwaZulu-Natal, Durban, South Africa Faculty of Community and Health Sciences, School of Public Health, University of the Western Cape, Cape Town, South Africa 3 Discipline of Public Health Medicine, School of Nursing and Public Health, University of KwaZulu-Natal, Durban, South Africa 1 2

Corresponding author: M Muzigaba (mochemoseo@gmail.com) Background. There is still limited to no evidence on the independent and interactive effects of HIV infection, disease stage, baseline disease severity and other important comorbidities on mortality risk among young children treated for severe acute malnutrition (SAM) in South Africa (SA, using the World Health Organization (WHO) recommended treatment modality.

Objectives. To determine baseline clinical characteristics among children with SAM and assess whether HIV infection, disease stage, critical illness at baseline and other comorbidities independently and interactively contributed to excess mortality in this sample.

Methods. We followed up children aged 6 - 60 months, who were admitted with and treated for SAM at two rural hospitals in SA, and retrospectively reviewed their treatment records to abstract data on their baseline clinical characteristics and treatment outcomes. In total, 454 children were included in the study. Descriptive statistical tests were used to summarise patients’ clinical characteristics. KaplanMeier failure curves were created for key characteristics and compared statistically using log-rank tests. Univariate and multivariate Cox regression was used to estimate independent and interactive effects.

Results. The combined case fatality rate was 24.4%. HIV infection, clinical disease stage, the presence of lower respiratory tract infection, marasmus and disease severity at baseline were all independently associated with excess mortality. The critical stage for higher risk of death was when cases were admitted at WHO stage III. The interactions of two or three of these characteristics were associated with increased risk of death when compared with having none, with HIV infection and critical illness showing the greatest risk (hazard ratio 22, p<0.001).

Conclusion. The high HIV prevalence rate in the study setting and the resultant treatment outcomes support the notion that the WHO treatment guidelines should be revised to ensure that mechanisms for effective treatment of HIV comorbidity in SAM are in place.

However, a much more rigorous study is warranted to verify this conclusion.

S Afr J Child Health 2017;11(1):46-53. DOI:10.7196/SAJCH.2017v11i1.1202

Childhood undernutrition remains a major public health concern in developing countries, and has in the past been shown to contribute to ~50% of the nearly 10 million under-5 children years who die each year of preventable causes.[1] Although severe acute malnutrition (SAM) is seldom recognised, this condition remains an extremely widespread disorder associated with high rates of mortality and morbidity, and requires specialised treatment and interventions.[1] In an effort to reduce deaths from SAM and improve recovery, the World Health Organization (WHO) has developed a ‘10-step’ guideline for managing SAM.[2] These guidelines have since been promoted as the standard treatment modality for clinical care of severely malnourished children.[3] There is evidence to show that if implemented correctly, the protocol can improve case fatality rates (CFRs) from ~40% to <10%,[4] even when applied in emergency humanitarian situations.[5] These guidelines have gained recognition worldwide and are now being used in most healthcare units, including some hospitals in South Africa (SA).[6] While numerous studies have established the adverse effects of HIV infection on the survival of children treated for SAM,[7-9] few have studied the independent effects of HIV clinical stage in particular, as well as disease severity at baseline and other critical comorbidities on admission. There is also limited or no evidence on the interactive effects of these clinical characteristics on increased mortality risk.

46

SAJCH

This study was, in part, prompted by the lack of such evidence and the fact that in the study setting, the high CFRs for SAM were being attributed to HIV infection rather than to mismanagement of SAM children by healthcare workers. There was also some anecdotal evidence from clinicians in the same hospitals that, depending on the clinical stage of HIV infection, the WHO 10-step protocol may show no effect. This study therefore sought to establish whether there was (i) an independent effect of HIV infection, HIV clinical stage, baseline disease severity and other clinical factors on survival prospects of children admitted and treated for SAM using the WHO 10-step guidelines; and (ii) an added risk of death depending on the interaction of two or more of these baseline clinical characteristics.

Methods

Study setting

This study was conducted in two rural district hospitals located in the Eastern Cape Province of SA. The two hospitals were purposefully selected on the basis of being the only hospitals in the region that had shown optimal implementation of the WHO 10-step guidelines and had the basic resources to do so at the time.[10] Furthermore, the hospitals serve catchment areas where there is a high HIV prevalence. This increased the likelihood of enrolling enough HIVpositive children with SAM for the purposes of the study.

MARCH 2017 Vol. 11 No. 1

RESEARCH Study design

This study consisted of an observational study design with a prospective and retrospective component. The prospective aspect involved: (i) identifying and classifying multiple cohorts of SAM cases admitted to two different hospitals, at different study intervals, according to their nutritional and HIV status; (ii) identifying their clinical profile on admission; and (iii) initiating them on treatment and following them up to assess specific treatment outcomes. The retrospective component involved reviewing treatment records, which were compiled prospectively to document outcomes of interest, and linking this information with baseline characteristics.

Study sample

Patient treatment records were only considered for the study if they belonged to SAM patients admitted to either of the hospitals between January 2009 and May 2013 (the period during which treatment of SAM cases was enforced and monitored) and if patients were between the ages of 6 months and 5 years. The patient treatment records considered also contained clearly defined malnutrition status as per the Wellcome classification,[11] had records showing HIV test results and HIV clinical stage, and included a complete treatment record of the child while in the hospital. The Wellcome system was more practical than other classification methods, as there were some inconsistencies in the measurement of height/length. A comprehensive written medical examination by a doctor and the discharge criteria followed for patients who did not die while being treated were also used as eligibility criteria. In total, 466 treatment records meeting the inclusion criteria were identified from both hospitals during the study period. This was ~85% of the total number of treatment records that were available during the same period. The records that were selected constituted the unit of data abstraction and analysis.

Patient recruitment

This study was approved by the University of the Western Cape Research Ethics Committee (reg. no. 12/10/37). During the study period, the parents or guardians who brought their children with SAM to the hospital for treatment were approached in the ward after admission and informed about the study. They were then asked to provide informed consent for their children to be enrolled in the study. For the children whose parents provided consent, the inpatient numbers recorded on the treatment charts were noted by a research nurse for later reference during retrospective record review.

Data generation and collection

Data were extracted from patient treatment records. A structured and validated questionnaire developed by the International Malnutrition Taskforce and Muhimbili National Hospital, Dar es Salaam, Tanzania[12] was used for extraction and collection of all the data.

SAM classification was done on admission by the admitting doctor, based on the Wellcome criteria.[11] Confidential and private counselling for HIV testing was conducted by professionally trained nurses for all parents and guardians of children with SAM. Parents and guardians were also requested to give permission for an HIV polymerase chain reaction test to be done for their children. Parents/ guardians provided their own consent and were tested for HIV using an enzyme-linked immunosorbent assay test. HIV clinical staging was done by the admitting doctor as per the WHO guidelines.[13] The admitting doctor also graded oedema and dermatosis on admission if they were present, and confirmed the presence of lower respiratory tract infections (LRTIs). LRTI was an umbrella term used for cases with comorbidities such as pneumonia, bronchitis and other infections below the larynx. Tuberculosis (TB) was not considered as a comorbidity in this study as the condition was mostly underdiagnosed, which resulted in too few cases being included in the study to be studied separately.

47

SAJCH

Based on diagnostic information, SAM cases were also classified as critically ill or not. Definition of a case as critically ill was based on whether or not they were admitted with one or a combination of five clinical features, i.e.: (i) depressed conscious state (prostration or coma); (ii) bradycardia; (iii) evidence of shock with or without dehydration; (iv) hypoglycaemia; and (v) hypothermia. These clinical manifestations have been documented as the strongest predictors of early death (within 24 hours of admission).[14] Other comorbidities (excluding LRTIs, TB and HIV/AIDS described above) directly or indirectly related to SAM were also noted, e.g. lethargy, hyponatraemia and hypokalaemia, dehydration, deep acidotic breathing, anaemia and pyrexia, herbal intoxication, diarrhoea, burns and other hereditary dysfunctions commonly reported by the doctors in each hospital.[14] The outcomes in this study were death while under treatment or discharge following treatment completion, i.e. survival.

Data analysis

All the data were processed and analysed using Stata/IC 13.0 (StataCorp. LP, USA). Survival analysis was used to assess time to death for SAM cases based on baseline clinical characteristics. Firstly, Kaplan-Meier failure curves for key characteristics were created and compared statistically using log-rank tests. The log-rank test was performed among groups for each predictor variable to determine whether certain predictors needed to be in the final model. A predictor was included in the model if p<0.20. This elimination scheme, though arbitrary, allowed for the exclusion of variables that were less likely to contribute anything to the model, which included other predictors. Therefore, the unadjusted Cox proportional hazard model consisted of predictors where p<0.20. The predictors that were not statistically significant at univariate level were dropped from the final multivariate (adjusted) model. The Cox formulation was also used to quantify the impact of the interaction of factors on the survival prospects of study subjects, and to identify potential effect modifiers. Both the unadjusted, adjusted and interaction models were tabulated with hazard ratios (HRs), 95% confidence intervals (CIs) and the corresponding p-values. Throughout the analysis, p<0.05 was considered statistically significant. The final model was tested to determine whether it did not violate the proportional hazard assumption.

Results

Descriptive results

Over the study period 466 subjects met the study inclusion criteria and were considered for the study. However, 12 SAM cases defaulted or absconded treatment and were censored during data analysis, resulting in 454 cases being analysed. The majority of subjects (50%, n=225) were aged between 13 and 24 months, followed by cases aged 6 - 12 months (36%, n=165). Cases aged 3 - 5 years constituted only 14% of the study sample.

Pooled data revealed that the proportion with marasmus and kwashiorkor was similar (38% and 40%, respectively), whereas only 22% were classified as marasmus-kwashiorkor (Table 1). About a quarter (28%) of SAM cases had LRTIs on admission, and a third had other comorbidities, the most common being gastroenteritis (11%). In both hospitals, the most commonly diagnosed LRTI was pneumonia.

Pooled data also showed that 196 (43%) were HIV-positive. In total, the majority who tested positive for HIV were classified as disease stage III (37%), followed by those who were at stage II (26.5%), then 19% and 17% for stages I and IV, respectively. Approximately 26% of SAM cases were admitted in a critical condition as per study definition. The combined CFR was 24.4%.

Pooled analysis revealed that more HIV-positive SAM patients were marasmic compared with their HIV-negative counterparts (56% and 24%, respectively, p<0.001) (Table 2). The reverse was

MARCH 2017 Vol. 11 No. 1

RESEARCH Table 1. Comparison of clinical characteristics of SAM cases by hospital on admission (N=454) Variable Severe malnutrition classification Marasmus Kwashiorkor Marasmus-kwashiorkor Oedema grade None Mild Moderate Severe Dermatosis grade None Mild Moderate Severe LRTIs Yes No Other comorbidities Yes No Critically ill on admission Yes No HIV status Positive Negative HIV/AIDS disease stage 1 2 3 4 Outcome Died Discharged

Hospital A

Hospital B

Both hospitals combined

59 (40.9) 61 (42.4) 24 (16.7)

114 (36.8) 120 (38.7) 76 (24.5)

173 (38.1) 181 (39.9) 100 (22.0)

52 (36.1) 9 (6.2) 39 (27.0) 44 (30.6)

101 (32.6) 20 (6.4) 83 (26.7) 106 (34.2)

153 (33.7) 29 (6.4) 122 (26.9) 150 (33.1)

36 (25.0) 43 (29.8) 47 (32.6) 18 (12.5)

100 (32.6) 63 (20.3) 110 (35.5) 37 (11.9)

136 (29.9) 106 (23.4) 157 (34.6) 55 (12.1)

48 (33.3) 96 (66.7)

83 (26.7) 227 (73.3)

131 (28.9) 323 (71.2)

55 (38.2) 89 (61.8)

97 (31.3) 213 (68.7)

152 (33.4) 302 (66.5)

33 (22.9) 111 (77.1)

84 (27.1) 226 (72.9)

117 (25.7) 337 (74.2)

78 (54.1) 66 (45.8)

118 (38.0) 192 (61.9)

196 (43.2) 258 (56.8)

18 (23.0) 22 (28.2) 23 (29.4) 15 (19.2)

16 (13.6) 30 (25.4) 50 (42.7) 22 (18.6)

34 (17.4) 52 (26.5) 73 (37.2) 37 (18.8)

34 (24.3) 106 (75.7)

74 (24.5) 228 (75.7)

108 (24.4) 334 (75.6)

p-value* 0.171

0.851

0.122

0.156

0.140

0.344

0.001

0.191

0.960

*p-values are for differences in variable categories based on pooled data (data for both hospitals combined).

Table 2. Cross-tabulation of cases by SAM syndromic classification and HIV status: Hospital-level and pooled analyses (2009 - 2013) Marasmus Hospital A HIV-negative HIV-positive Total Hospital B HIV-negative HIV-positive Total Both hospitals combined HIV-negative HIV-positive Total

Clinical classification of severe malnutrition, n (%) Kwashiorkor Marasmus-kwashiorkor

Total, n

15 (22.7) 44 (56.4) 59 (40.9)

41 (62.1) 20 (25.6) 61 (42.3)

10 (15.2) 14 (17.9) 24 (16.7)

66 78 144

48 (25.0) 66 (55.9) 114 (36.8)

98 (51.0) 22 (18.6) 120 (38.7)

46 (23.9) 30 (25.4) 76 (25.5)

192 118 310

63 (24.4) 110 (56.1) 173 (38.1)

139 (53.8) 42 (21.4) 181 (39.9)

56 (21.7) 44 (22.5) 100 (22.3)

258 196 454

48

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH true for cases who had kwashiorkor, where 53% were HIV-negative and 21% HIVpositive (p<0.001). The same direction of differences was also observed at the level of individual hospitals, with a similar statistical significance (p<0.001). For cases with marasmus-kwashiorkor, there were no statistically significant differences across HIV status (p>0.05).

Survival analysis

Cox proportional hazard model

Univariate results The univariate results shown in Table 3 indicate that only HIV status and stage,

0.75

0.50

0.25

Log-rank test: p<0.001

0.75

0.50

0.25

0.00

0.00 0

10

20

30

40

50

0

Time since hospitalisation (days)

10

20

30

40

50

Time since hospitalisation (days) Stage I Stage II

HIV-negative HIV-positive

Stage III

Stage IV

Fig. 1. Kaplan-Meier failure curves for SAM cases stratified by HIV status and HIV clinical stages.

1.00

Log-rank test: p<0.001 Estimated survival probability

Estimated survival probability

1.00 0.75 0.50 0.25 0.00

Log-rank test: p<0.001

0.75 0.50 0.25 0.00

0

10

20

30

40

50

0

10

Marasmus Kwashiorkor Marasmus-kwashiorkor 1.00

1.00 Estimated survival probability

0.75 0.50 0.25 0.00 10

20

30

40

30

40

50

No LRTI(s) present LRTI(s) present

Log-rank test: p<0.001

0

20

Time since hospitalisation (days)

Time since hospitalisation (days)

Estimated survival probability

Univariate survival analysis Based on the log-rank test for equality of failure functions, mortality hazard did not differ significantly by hospital (p=0.341).

Hence, for subsequent analysis the hospitals were combined.

In this study, 41.15% of HIV-positive and 11.6% of HIV-negative SAM cases died while under treatment (p<0.001)(Fig. 1). HIVpositive cases had worse survival prospects than their HIV-negative counterparts in both hospitals (p<0.001). As shown in the graph, the cumulative fraction of HIVpositive SAM cases who had died by day 6 post admission was 25% compared with HIV-negative cases, which was about 5% at the same time point. These results imply that HIV-positive cases generally died sooner and in greater numbers following admission, than their HIV-negative counterparts. Stage IV cases had the worst survival prospects followed by their stage III counterparts.

The difference between these two groups was statistically significant (p<0.01) as was the difference between stage I and stage II SAM cases (p=0.043) and stage II and stage III cases (p<0.001). These results may imply that the critical stage for higher risk of death was when cases were admitted at stage III. Approximately 25% of children admitted in a critical condition had died by the second day, compared with only 1% of the non-critically ill group (p<0.001) (Fig. 2).

Mortality patterns beyond the second day showed that ~75% of critically ill children with SAM had died by the 28th day after hospitalisation, compared with about 17% who were not critically ill. Survival prospects for SAM cases who were admitted with marasmus were significantly poorer compared with cases who had either kwashiorkor or marasmic-kwashiorkor (p=0.001). No differences were found between cases that had kwashiorkor and marasmuskwashiorkor (p=0.052). Furthermore, the presence of other comorbidities and coinfection with LTRIs/TB was associated with excess mortality.

1.00

Log-rank test: p<0.001 Estimated survival probability

Estimated survival probability

1.00

50

Time since hospitalisation (days) No presence of other comorbidities Presence of other comorbidities

Log-rank test: p<0.001

0.75 0.50 0.25 0.00 0

10

20

30

40

50

Time since hospitalisation (days) Not critically ill on admission Critically ill on admission

Fig. 2. Kaplan-Meier failure curves for SAM cases stratified by SAM class, presence of LRTI(s), presence of other comorbidities and critical illness.

critical illness on admission, other baseline comorbidities, LRTIs and the three SAM syndromic classifications were significantly associated with increased hazard of death. For patients with severe oedema, the hazard of death was 46% less than the hazard of dying

49

SAJCH

MARCH 2017 Vol. 11 No. 1

among those who had no oedema (p=0.001).

Before risk factor adjustment, cases who had LRTIs had about a four times higher hazard of death compared with those who did not (HR 3.66, p<0.001). Patients admitted with other comorbidities at baseline had almost a

RESEARCH two-fold higher hazard of death than those admitted without (HR 1.94, p=0.001).

HIV status and case severity were the strongest predictors of death in the unadjusted model. Children who were critically ill on admission had a five times higher hazard of death than those who were not (HR 5.70, p<0.001). Children who were HIV-positive and at stage IV or III of infection had, respectively, eight times higher (HR 8.12, p<0.001) and five times higher (HR 5.73, p<0.001) hazard of death than their HIV-negative counterparts. Stages II and I of HIV infection were not statistically significant predictors of mortality (p>0.05).

Multivariate results

After multivariate adjustments, hospital, age, SAM syndromic classifications, dermatosis grade, other baseline comorbidities and oedema grade were not significantly associated with increased risk of death.

After multivariate adjustment, the hazard rate was reduced to 1.75 for children who had LRTIs at baseline, but remained statistically significant (p=0.001). HIV status and case severity were again the strongest predictors of death in adjusted models. Children who were critically ill on admission had a three-and-a-half times higher hazard of death than those who were not (HR 3.64), a result which was

Table 3. Univariate (unadjusted) and multivariate (adjusted) Cox proportional hazard models for factors associated with mortality among children with SAM during the intervention period Factors

n/N

UHR

Unadjusted model 95% CI p-value

AHR

Adjusted model * 95% CI

Hospital Hospital A Hospital B

34/140 74/302

Ref 1.22

0.81 - 1.84

0.346

Ref 1.55

0.91 - 1.92

0.401

Age (months) 6 - 12 13 - 24

46/164 47/225

Ref 0.73

0.49 - 1.11

0.147

0.93

0.61 - 1.43

0.771

14/64

0.74

0.41 - 1.36

0.345

0.61

0.32 - 1.15

0.129

SAM classification Marasmus Kwashiorkor

63/173 30/171

Ref 0.44

0.29 - 0.69

<0.001

Ref 1.21

0.42 - 3.45

0.806

M-kwashiorkor

16/98

0.44

0.25 - 0.77

0.004

0.53

0.20 - 1.43

0.080

54/153 6/29

Ref 0.55

0.24 - 1.29

0.171

Ref 0.58

0.22 - 1.53

0.109

Moderate

23/119

0.52

0.31 - 0.84

0.009

1.16

0.43 - 3.07

0.675

Severe

29/141

0.46

0.29 - 0.74

0.001

0.61

0.20 - 1.83

0.682

Dermatosis grade None Mild Moderate Severe

32/134 23/102 33/153 20/53

Ref 0.88 0.90 1.73

0.52 - 1.51 0.56 - 1.47 0.10 - 1.04

0.652 0.686 0.052

Ref 0.69 1.03 1.43

0.37 - 1.26 0.59 - 1.80 0.76 - 2.66

0.851 0.439 0.088

LRTIs No Yes

47/314 61/128

Ref 3.66

2.50 - 5.36

<0.001

Ref 1.74

1.12 - 2.70

0.001

Other baseline comorbidities No Yes

56/294 52/148

Ref 1.94

1.33 - 2.84

0.001

Ref 1.14

0.74 - 1.76

0.880

Critically ill at baseline No Yes

44/328 64/114

Ref 5.70

3.87 - 8.39

<0.001

Ref 3.64

2.35 - 5.64

<0.001

HIV status and stage Negative Positive/stage I Positive/stage II Positive/stage III Positive/stage IV

29/250 2/33 9/50 40/72 28/37

Ref 0.46 1.44 5.73 8.12

0.11 - 1.94 0.68 - 3.05 3.54 - 9.25 4.82 - 13.66

0.231 0.189 <0.001 <0.001

Ref 0.20 1.12 3.18 3.74

0.03 - 1.53 0.51 - 2.49 1.85 - 5.47 2.05 - 6.84

0.312 0.223 <0.001 <0.001

25 - 61

Oedema grade None Mild

p-value

â&#x20AC;

n/N = number of failure cases over total number of cases; UHR = unadjusted hazard ratio; AHR = adjusted hazard ratio; Ref = reference group; M-kwashiorkor = marasmus-kwashiorkor. *The global test for hazard proportionality revealed that the overall multivariate model did not significantly violate the proportional hazards assumption (p-value=0.184).

Overall, the variance inflation factor (VIF) for the model was 2.87 which was less than the cut-off point of VIF=10. â&#x20AC;

p=95% level of significance.

50

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH statistically significant (p<0.001). Severely malnourished children who were HIV-positive and at stage III or IV had, respectively, a five times higher (HR 8.12, p<0.001) and eight times higher (HR 5.73, p<0.001) hazard of death than their HIV-negative counterparts. The HRs dropped by about three units for stages III and IV, but remained statistically significant (p<0.001). At multivariate level, stages I and II of HIV infection remained not statistically significant predictors of mortality (p>0.05).

Interaction survival modelling

Children who had LRTIs at baseline and were critically ill at the same time had a 14 times higher hazard of death compared with those who had none (p<0.001) (Table 4). Furthermore, children who were HIV-positive and critically ill on admission had the highest hazard of death (HR 22, p<0.001) compared with those who were not exposed to any of the three risk factors. SAM cases who were HIV-positive and had LRTIs had only a nine times higher hazard of death (p<0.001). A striking result was that the hazard of death was slightly lower (HR 19, p<0.001) for cases who were HIV-positive, had LRTIs and were critically ill on admission compared with those who only had HIV infection and were critically ill on admission (HR 22, p<0.001). In comparison with cases that had none of the risk factors, those who were exposed to only one of the three risk factors had lower hazards of death than those who were exposed to more than one. The predictive power of one risk factor during interaction analysis was not statistically significant (p>0.05), except for critical illness (p=0.001).

Discussion

This study adds an arguably new and important dimension to what already exists in the literature around the management of SAM in the context of HIV infection. In addition to HIV status itself, the study revealed independent and interactive effects of HIV disease stage, baseline disease severity and other baseline comorbidities on mortality within the context of the WHO 10-step treatment modality in two resource-limited healthcare facilities in SA. With regard to the burden of HIV infection, the combined prevalence in both hospitals was slightly lower (43%) than that reported in a similar and recent study conducted in SA (51%),[9] but higher than that reported in a study conducted in Malawi (14%).[7] The current study also confirmed findings from other studies that HIV-infected SAM cases were more likely to present with marasmus than kwashiorkor on admission.[8,15] While it is usual to present with kwashiorkor in the southern African region, with the high prevalence of HIV infection, SAM patients become wasted owing to changes in physiological and metabolic functions associated with HIV infection.[16]

This study also examined the independent effect of HIV infection on survival. The CFRs in this study were higher in the HIV-infected group (41%) compared with their HIV-uninfected counterparts (11%). These estimates were slightly higher than those recently reported in Burkina Faso (39.7% v. 10.9% for HIV-infected and HIV-uninfected cases, respectively).[17] A study in Niger also reported the same direction of relationship but with much lower CFRs for HIV-infected and HIVuninfected SAM cases (20% and 14%, respectively).[18] In this study we have shown that disease severity at baseline was independently associated with excess mortality, a finding that is consistent with past evidence by Maitland et al.[14] Children with SAM who were admitted with one or a combination of clinical features such as coma, hypoglycaemia, hypothermia and bradycardia, had the worst survival probability compared with those who had other less SAM-related manifestations, such as herbal intoxication, Cushingoid facies, etc. These findings may have important practical implications.

As Maitland et al.[14] have argued, the clinical features associated with higher risk of death can be used by frontline healthcare givers to target emergency treatment and allocate resources more appropriately. This is particularly important in the context of poor human and material resources in most rural hospitals in sub-Saharan Africa, such as those in which the current study was conducted. With mortality being highest in the first 5 days of admission, there is a need to institute proper, well-supported and sustained triage and emergency management mechanisms so that the â&#x20AC;&#x2DC;at higher risk of deathâ&#x20AC;&#x2122; cases are identified and treated according to the available guidelines.

Continued attention to appropriate diagnosis and treatment of common complications such as LRTIs also needs to be prioritised, as they were shown to be strongly associated with excess mortality in this study. Such complications do not always require a specialist physician and can be managed by most professional nurses in SA, with minimal supervision from a physician. It is worth noting that TB was found to be less common in the study sample than other LRTIs. This is probably because TB was difficult to diagnose microbiologically in the study setting, despite it being an important comorbidity in the management of SAM in the context of HIV infection. The challenge of TB diagnosis has been documented before among HIV-infected children with SAM in whom the tuberculin skin test is often falsely negative.[9] Perhaps what the current study adds to the body of knowledge is the pattern of survival among HIV-infected SAM cases that were at different stages of HIV infection. Cox regression analyses revealed that the threshold for higher risk of death was when children with SAM were admitted at stage III. The reasons for excess mortality risk associated with advanced stages of HIV infections in SAM are

Table 4. Cox proportional hazard model for interaction of factors associated with mortality among children with SAM admitted to both hospitals during the intervention period Interaction terms*

HIV

LRTI/TB at baseline

Statistics

Critically ill at baseline

n/N

HR

9/165

Ref

95% CI

p-value

0

0

1

9/35

4.83

1.92 - 12.18

<0.001

0

1

0

3/35

1.55

0.42 - 5.71

0.532

0

1

1

8/15

14.64

5.64 - 38.01

<0.001

1

0

0

12/89

2.18

0.92 - 5.19

0.411

1

0

1

17/25

22.00

9.78 - 49.49

<0.001

1

1

0

20/39

9.86

4.49 - 21.68

<0.001

1

1

1

30/39

19.79

9.39 - 41.73

<0.001

*0 = no; 1 = yes.

51

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH not yet well understood, particularly in the context of antiretroviral therapy (ART).[19] However, some authors continue to attribute excess mortality to complex pathophysiological, metabolic and pharmacological changes that occur as HIV infection progresses.[17] HIV-infected SAM patients in the current study were initiated on ART using the standard guidelines. However, the timing of ART in preventing premature mortality among HIV-infected SAM cases at different disease stages remains a challenge. A randomised controlled trial has shown that half the children hospitalised for SAM developed oedema after starting ART.[20] In the same study, one in 14 children who were at an advanced stage of HIV infection became oedematous 12 weeks after ART initiation.

Another contribution of this study is the demonstration of the interactive effects of multiple clinical risk factors on mortality in the study sample. A study by Koethe and Heimburger[21] has shown that SAM, immune function and infection burden interact in children, particularly in the context of HIV infection. In this study, being critically ill and having LRTIs were, both independently and as a combination, potential effect modifiers of the higher risk of death, as they increased the hazard of death quite substantially when they were combined with HIV infection in the interaction model. The interaction model of HIV infection, LRTIs and baseline disease severity also revealed that in comparison with patients who had none of these risk factors, those who were exposed to only one of the three risk factors had lower hazards of death than those who were exposed to more than one. These findings also speak to a need to develop a triage system that helps healthcare workers to prioritise cases that are at higher risk of death.

Finally, it is also worthy to note that the combined CFR of 24.4% shown in this study is much higher than the WHO target of <5% in the context of the WHO 10-step treatment modality, implying that achieving this target may be unrealistic, particularly among HIVinfected cases in resource-limited settings. De Maayer and Saloojee[9] have also shown that, even in resource-privileged SA settings, the WHO target of <5% may be impractical.

Study limitations

This study was conducted in two rural facilities to determine whether the findings would be similar. However, although we found similarities in results, these may not be generalisable beyond the study setting. There may be a layer of both internal and external facility-specific factors that determined treatment outcomes for SAM. Furthermore, the current study did not investigate how such factors may have influenced the process of care and the treatment outcomes thereof.

Study recommendations

We recommend that a carefully controlled study, such as a randomised controlled trial, be conducted to make a better judgement about the effects of the risk factors reported here on mortality in the context of the WHO treatment guidelines.

Conclusion

Within the context of increased ART and prevention of mother-tochild transmission (PMTCT) coverage in SA, there seems to be hope that the impact of HIV infection on SAM-CFRs among children may be reduced. However, given the limited evidence on the most effective ways of managing SAM patients who are HIV infected, this group of patients will continue to pose some challenges to the healthcare workforce. The high HIV prevalence rate in the study population and the resultant treatment outcomes lend support to the notion that the WHO treatment guidelines should be revised to ensure that mechanisms for effective treatment of HIV comorbidity in SAM are in place. The revisions ought to tap into the differential energy requirements by HIV status, timing of ART initiation among HIV52

SAJCH

infected SAM cases as well as developing a broad-based triage system to more effectively identify and treat SAM cases that are at higher risk of early death. However, the target of <5% CFR may be difficult to achieve, particularly for HIV-positive SAM cases, as has been suggested in our study findings. The role of the broader healthcare system in SA in preventing incidences of SAM at community level in the study setting is discernible. The current study has shown that the admission rates attributable to SAM at both hospitals remain high, and as such require mitigation. Community-based prevention strategies developed through multisectoral engagements have a huge role to play in achieving this goal. It goes without saying that early detection of SAM would be a crucial part of preventing disease severity and the development of multiple infections, thereby reducing the likelihood of subsequent preventable death associated with SAM in the study setting. The close linkage between SAM and HIV/AIDS begs an integrated healthcare approach to ensure that hospitals and HIV clinics work in harmony to optimise outcomes for both the child and the mother. Strengthening PMTCT services in resource-limited settings would also likely reduce the burden of HIV infection among children, and mitigate the risk of developing severe malnutrition.

Acknowledgements. The authors wish to thank the staff at the two hospitals where this study was conducted, and Prof. Ann Ashworth for her immense contribution to the conceptualisation of the research.

Funding. This study was funded by the SA National Research Foundation (NRF) and partly by the South African Centre for Epidemiological Modelling and Analysis (SACEMA).

1. Collins S, Sadler K, Dent N, et al. Key issues in the success of communitybased management of severe malnutrition. Food Nutr Bull 2006;27(3) (Suppl):S49-S70. https://doi.org/10.1177/15648265060273s304 2. World Health Organization. Management of the child with a serious infection or severe malnutrition: Guidelines for care at the first-referral level in developing countries. Geneva: WHO, 2000. http://www.who.int/child-adolescent-health/ publications/referral_care/homepage.html (accessed 14 July 2015).

3. Ashworth A, Khanum S, Jackson A, Schofield C. Guidelines for the inpatient treatment of severely malnourished children. The World Health Organization Library Cataloguing in-Publication Data 1996. Geneva: WHO, 1996.

4. Cavalcante AAM, Pinheiro LMP, Monte C, Guimaraes ARP, Ashworth A. Treatment of malnutrition in Brazil: Simple solutions to common problems.

Trop Doct 1998;28(2):95-97.

5. Prudhon C, Briend A, Laurier D, Golden MH, Mary JY. Comparison of weight- and height-based indices for assessing the risk of death in severely malnourished children. Am J Epidemiol 1996;144(2):116-123. https://doi. org/10.1093/oxfordjournals.aje.a008898 6. Puoane T, Sanders D, Ashworth A, Chopra M, Strasser S, McCoy D. Improving the hospital management of malnourished children by participatory research.

Int J Qual Health Care 2004;16(1):31-40. https://doi.org/10.1093/intqhc/ mzh002 7. Chinkhumba J, Tomkins A, Banda T, et al. The impact of HIV on mortality during inpatient rehabilitation of severely malnourished children in Malawi.

Trans R Soc Trop Med Hyg 2008;102(7):639-644. https://doi.org/10.1016/j. trstmh.2008.04.028 8. Kessler L, Daley H, Malenga G, et al. The impact of the Human immunodeficiency virus type 1 on the management of severe malnutrition in Malawi. Ann Trop Paediatr 2000;20(1):50-56. https://doi.org/10.1080/02724930092075 9. De Maayer T, Saloojee H. Clinical outcomes of severe malnutrition in a high tuberculosis and HIV setting. Arch Dis Child 2011;96(6):560-564. http:// dx.doi.org/10.1136/adc.2010.205039 10. Puoane T, Sanders D, Chopra M, et al. Evaluating clinical management of severely malnourished children: A study of two rural district hospitals. S Afr Med J 2001;91:137-141.

11. Murgod R, Ahmed M. Instant nutrition assessment in children with protein energy undernutrition. Int J Appl Bio Pharma Tech 2015;6(1)171-177.

12. World Health Organization. Improving the inpatient management of severe acute malnutrition: Toolkit to monitor current management of severe acute malnutrition. Geneva: WHO, 2010. http://www.cmamforum.org/Pool/ Resources/Toolkit-to-monitor-management-SAM-2010.pdf (accessed 12 January 2015).

13. World Health Organization. Interim WHO clinical staging of HIV/AIDS and HIV/AIDS case definitions for surveillance (African region) Geneva: WHO, 2005. http://www.who.int/hiv/pub/guidelines/casedefinitions/en/index.hml (accessed 15 January 2015).

MARCH 2017 Vol. 11 No. 1

RESEARCH 14. Maitland K, Berkley JA, Shebbe M, et al. Children with severe malnutrition: Can those at highest risk of death be identified with the WHO protocol? PLoS Med 2006;3(12):e500. http://dx.doi.org/10.1371/journal.pmed.0030500 15. Prazuck T, Tall F, Nacro B, et al. HIV infection and severe malnutrition: A clinical and epidemiological study in Burkina Faso. AIDS 1993;7(1)103-108. https://doi.org/10.1097/00002030-199301000-00016 16. Mehta NM, Corkins MR, Lyman B, et al. Defining paediatric malnutrition: A paradigm shift toward aetiology-related definitions. J Parenter Enteral Nutr 2013;37(4):460-481. http://dx.doi.org/10.1177/0148607113479972 17. Savadogo GL, Donnen P, Koueta F, Kafando F, Hennart P Dramaix M. Impact of HIV/AIDS on mortality and nutritional recovery among hospitalised severely malnourished children before starting antiretroviral treatment. Open J Pediatr 2013;3(4):340-345. http://dx.doi.org/10.4236/ojped.2013.34061

53

SAJCH

18. Madec Y, Germanaud D, Moya-Alvarez V, et al. HIV prevalence and impact on renutrition in children hospitalised for severe malnutrition in Niger: An argument for more systematic screening. PloS ONE 2011;6(7):e22787. http:// dx.doi.org/10.1371/journal.pone.0022787 19. Ndekha MJ, Manary MJ, Ashorn, P. Briend A. Home based therapy with ready-to-use therapeutic food is of benefit to malnourished, HIVinfected Malawian children. Acta Paediatr 2005;94(2):222-225. https://doi. org/10.1111/j.1651-2227.2005.tb01895.x 20. Prendergast A, Dangarembizi BM, Kitaka BS, et al. Hospitalisation for severe malnutrition among HIV-infected children starting antiretroviral therapy.

AIDS 2011;25(7):951-956. http://dx.doi.org/10.1097/QAD.0b013e328345e56b 21. Koethe JR, Heimburger DC. Nutritional aspects of HIV-associated wasting in sub-Saharan Africa. Am J Clin Nutr 2010;91(4):1138S-1142S. http://dx.doi. org/10.3945/ajcn.2010.28608D

MARCH 2017 Vol. 11 No. 1

RESEARCH

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0.

Parental satisfaction in the traditional system of neonatal intensive care unit services in a public sector hospital in North India V Sankar,1 MBBS; P Batra,1 MD, FACEE; M Saroha,1 MD; J Sadiza,2 MPhil 1 2

Department of Paediatrics, University College of Medical Sciences and Guru Teg Bahadur Hospital, Delhi, India Department of Clinical Psychology, Institute of Human Behaviour and Allied Sciences, Delhi, India

Corresponding author: P Batra (drprernabatra@yahoo.com) Background. Traditional systems of neonatal intensive care unit (NICU) care predispose parents to increased levels of stress and anxiety due to parental separation from their infant. Parental satisfaction, an indicator of the quality of care, is significantly compromised during prolonged NICU stay. The research is limited in developing countries.

Objectives. To assess the parental satisfaction with traditional systems of NICU care in a public sector hospital and to identify the areas that need improvement and can be worked upon.

Methods. A semi-structured questionnaire was used to interview the parents of the neonates on the day of discharge. Fifteen questions were categorised into four domains, namely interpersonal relationships with staff, parents’ involvement, staff competence and services offered by the health system. Parental satisfaction level was marked on a three-point Likert scale, 0 corresponding to highly dissatisfied, and 2 to completely satisfied for each of the 15 questions.

Results. Out of 100 patients interviewed, communication was the chief determinant of their satisfaction. Parents expressed fair satisfaction levels with regard to the emotional support and encouragement received, but discontent at being unable to look after their own baby and breastfeed the baby. They were satisfied with the competence of the staff.

Conclusion. The traditional system of NICU care was not satisfying for the parents in many aspects and changes in the form of familycentred care should be tried for greater parental satisfaction.

S Afr J Child Health 2017;11(1):54-57 DOI:10.7196/SAJCH.2017.v11i1.1253

Neonatal intensive care unit (NICU) admission is a time of significant stress for the parents.[1] The mother is at increased risk of postpartum depression, both during the neonate’s hospitalisation and in the postdischarge period. The major concerns of NICU parents during this stressful time are their informational needs, their grief response, their parent-child role development, stress, and coping and social support. [2] In most hospitals, the traditional system of NICU care is followed, where babies are admitted to NICU under the supervision of doctors and nurses only, without much parental involvement. The parents are shown the baby once a day and updated regarding their baby’s condition during counselling sessions. The existing system forbids the presence of parents during the time of clinical rounds.

Adding to the distress is the inability of the mother to breastfeed directly, as the expressed breastmilk is fed to the baby by nursing staff as per its need. [3] The steady rise in the number of NICU admissions necessitates a quantitative as well as a qualitative improvement in the services offered by NICUs in tertiary hospitals in a developing country.

Higher parental satisfaction and lower stress levels among parents are major determinants in the prompt recovery of the neonate. [4,5] The purpose of these improvements would be for the parents to understand the situation and be able to cope with it successfully.

Studies have identified 11 dimensions of care as important to parents whose infants receive neonatal intensive care: assurance, caring, communication, consistent information, education, environment, follow-up care, pain management, participation, proximity, and support.[6] Russell et al.[7] found that the provision of information, support, and an increase in their involvement in the care of their baby were highlighted by parents as important in their experience of care, and thus contributed to greater satisfaction.

54

SAJCH

Little is known about the satisfaction level of parents of neonates requiring intensive care in developing countries.[8,9] Therefore we planned this study with the objective of assessing parental satisfaction with the traditional system of NICU care in a public sector hospital in North India, and of identifying the areas that need improvement and can be worked upon.

Methods

This cross-sectional study was carried out in the Division of Neonatology in the Department of Paediatrics of a tertiary care teaching hospital between April and July 2015, after obtaining clearance from the institutional ethical committee. Parents of 100 latepreterm and term neonates (≥34 completed weeks of gestation) admitted to NICU for more than 7 days were enrolled after written informed consent had been obtained. The parents of neonates born with major congenital malformations, with mothers who were HIVor hepatitis B-positive, or who had any other chronic illness, were excluded from the study.

Detailed methodology

A semi-structured questionnaire was used to interview the parents of the neonates, preferably the mothers, on the day of discharge. At the time of interview, the babies were either currently in the NICU, or they were with the mothers in the postnatal unit or in a transitional (lying-in) ward, where they were being taken care of by the mothers, under the constant supervision of nursing staff and doctors. The questionnaire included demographic details such as educational status, socioeconomic status and parity of the mother, along with the neonate’s gestational age, birth weight, age at the time of survey and

MARCH 2017 Vol. 11 No. 1

RESEARCH detailed diagnoses. For the assessment of satisfaction level, it included 15 questions regarding the experience of the neonatal care that was provided to their babies during admission. These questions were categorised into four domains, namely interpersonal relationships with staff, parental involvement, staff competence and services offered by the health system. Parents’ satisfaction level was marked on a threepoint Likert scale. The satisfaction level was scored as 0, 1 or 2, with 0 corresponding to highly dissatisfied, and 2 to completely satisfied for each of the 15 questions. Open suggestions for improvement in the system were also asked for and recorded if the score was either 0 or 1. The questionnaire took ~25 minutes to complete. In preparation of the questionnaire, a pilot study was conducted for 4 days followed by a role-play with the clinical psychologist, and was later modified based on the comments and suggestions.

Statistical analysis

Responses by the parents were analysed using SPSS software 20.0 (IBM Corp., USA). Descriptive statistics were applied for data analysis, with the open-ended responses being described using percentages. Factors affecting the satisfaction level were compared using a t-test.

Results

We interviewed 100 parents for the study without any parent refusing an interview. The demographic data for parents and their neonates are depicted in Table 1. The mean duration of stay of a neonate at the time of interview was 11 days. The number of parents with the satisfaction level achieved in each domain as graded on the Likert scale is depicted in Table 2.

Responses by the parents

Interpersonal relationships with staff • Communication: A large fraction of the parents mentioned communication as the chief determinant of their satisfaction. A little over half of the parents expressed their dissatisfaction with the communication with caregivers. Nearly one-third of the parents felt that doctors took very little time to answer their queries, and this increased their apprehension. Furthermore, some parents stated that doctors ought to talk to them politely, as doing so would relieve much of their anxiety. Almost half of the parents stated Table 1. Demographic profile of patients Parameter

Number (n) 68:32 67:33

Duration of stay ≤10 days

49

>10 days

51

Birth weight ≤2.5 kg

61

>2.5 kg

39

Education of parent Uneducated

15

Matric

36

Intermediate

41

Graduate

8

55

Competence of the staff Most parents expressed satisfaction with the experience and competence of the neonatal staff. They said that the counselling session held every afternoon reassured them of the skills of the doctors, and added that they were glad the doctors were doing their best. However, a few parents mentioned that they often felt overburdened by the vast amount of complex information the doctors provided them with, and suggested that it be simplified.

Parents showed dissatisfaction with this aspect of neonatal intensive care. This included blood and some medicines that had to be arranged by relatives, often straining their financial situation. The most common cause was their inability to afford the medicines required.

The majority of medications are provided by the government and the institute, and only rarely may parents be asked for something, in the case of non-availability. Some parents also commented that difficulties were faced when they were asked to arrange blood for a second time. They added that blood should be made available by the hospital if required more than once. Alternatively, some parents suggested that the hospital ask them to arrange blood beforehand, stating that it was difficult for them to arrange it at the time of need.

Parents also felt the need for a health insurance scheme for neonates, which is lacking in the present system.

Table 3 shows the mean satisfaction level of parents in each domain. The highest satisfaction levels were observed in interpersonal relationships with caregivers and the competence of staff, while parents were most dissatisfied with the services offered by the health system.

The satisfaction level of parents in relation to gender, gestational age, birth weight and the educational status of parents were comparable. With an increasing duration of hospital stay, a gradual

Gestation Term:Preterm

Parental involvement • Direct involvement in patient care: Over two-fifths of the parents expressed their dismay at being unable to look after their own baby. Parents mostly expressed concern over the long duration of the hospital stay and their separation from their baby, adding that they were worried about the baby not recognising them or responding to them.

• Expressing breastmilk: A little over one-third of the mothers expressed discontent over their inability to breastfeed the baby.

The existing system of expressing breastmilk into a container bothered many mothers. While for some the reason was uncertainty as to whether the milk was actually fed to the baby, others were concerned about the hygiene of such a practice.

Over one-third of mothers (n=36) described how they felt under considerable pressure to produce breastmilk because of frequent reproach by the nurses, and suggested that they should instead be reassuring in such circumstances.

Services offered by the health system

Sex of the neonate M:F

that they felt the information provided to them by the caregivers was inadequate. They felt, however, that nurses were more easily accessible than doctors for obtaining information and added that they were more considerate towards their circumstances.

A little over a quarter (n=27) of the parents suggested that the hospital should provide them with a leaflet at the time of discharge graphically representing the danger signs for them to look out for in their babies.

• Emotional support: Parents expressed fair satisfaction levels with regard to the emotional support and encouragement received. They expressed contentment at having received encouragement and praise from the nurses. Concurrently, having identified doctors as a superior authority, many voiced their desire for doctors to be more supportive. This, they felt, would conceivably improve their confidence.

SAJCH

MARCH 2017 Vol. 11 No. 1

RESEARCH Table 2. Responses of parents for each question Number of parents (N)

Domains

0 (Not at all satisfied)

1 (Moderately satisfied)

2 (Completely satisfied)

1. W ritten informed consent before admission in NICU

9

27

64

2. Updating infant’s status by doctors/staff in NICU

2

31

67

3. Complete information given by doctors/staff about infants

0

44

56

4. Empathetic attitude of doctors/staff in NICU

8

50

42

5. P rovision of phone call during emergency conditions

2

34

64

6. Comprehensive care by doctors/staff in NICU

0

29

71

7. Comprehensive care by doctors/staff in lying-in ward

1

28

71

8. Information given during the time of discharge

0

30

70

9. Process of feeding in NICU

7

46

47

10. Process of feeding in lying-in ward

0

29

71

11. Process of identification of infant

8

45

47

12. Complete care of infants in NICU

0

18

82

13. C are provided by doctors/staff in emergency conditions

2

29

69

14. D emands for blood/medicines by doctors/staff in 21 emergency conditions

44

35

15. Sanitation of the hospital

33

67

Domain 1 (Interpersonal relationship with caregivers, including communication and emotional involvement)

Domain 2 (Parental involvement in feeding)

Domain 3 (Competence of the staff)

Domain 4 (Services offered by the system)

0

increase in satisfaction level with respect to parental involvement was observed, with a slight decline in interpersonal relationship satisfaction, although it did not reach statistically significant levels.

Discussion

Parents face a number of challenges when their newborn infant is admitted for NICU care. These include the loss of work hours, separation from the baby, the inability to feed the baby and the financial burden, to name a few. An overburdened public sector hospital in a developing country lacks the support systems required to take care of these concerns, resulting in dissatisfied parents and poor neonatal care. An assessment of the satisfaction level of parents is an initial step in identifying the areas that require intervention so that parents are more satisfied with the healthcare system and can cope more successfully. Our study concluded that parents showed most dissatisfaction with the services offered by the healthcare system, and with parental involvement in care of the baby. Both these aspects need to be addressed for better outcomes for sick and preterm newborns.

Establishing effective communication with the parents and providing them with adequate information about their infants and the required care can result in increased satisfaction (p<0.01), as is evident from the study conducted by Weiss et al.[10] A study by Ranchod et al. [6] concluded that increased perinatal counseling allowed parents to take a more active decision-making role and invariably led to higher rates of parental satisfaction with neonatal 56

SAJCH

care in intensive care units. Constraints on staff limit the time available for extensive parent counselling by physicians, leaving parents to depend on nurses to help explain their infant’s status. [8] Reis et al.[11] report that interaction with nurses and verbal and written information regarding the condition of infants were essential, and of course the method of communication was also of significance.

It has also been observed that excessive information can lead to parental confusion, which therefore can decrease confidence in healthcare systems, increase anxiety and eventually decrease parental satisfaction.[12,13] Family-centred care (FCC) is an alternative system of NICU care that can provide an answer to these problems faced by parents, as under this system they are actively involved in the management of the baby, under the strict supervision of nursing staff and doctors.

Encouraging parents to spend time with their infants and actively participate in the care process can facilitate the development of parental roles and increase the satisfaction rate.[14] Bakewell-Sachs and Gennaro[15] have indicated that active maternal involvement in neonatal care and mother-infant contact (e.g. touching the baby) increases maternal confidence in taking care of the infant after discharge, and consequently lead to higher maternal satisfaction.

Integrated notes can be kept by the side of the baby’s cot so that information is accessible to parents in clear language. Bastani et al.[16] compared FCC with controls in a randomised trial and found that mothers in the FCC group were more satisfied with the aspects of information and participation in care. The number of neonatal

MARCH 2017 Vol. 11 No. 1

RESEARCH Table 3. Satisfaction levels of each domain Satisfaction Domain

Satisfaction, %

Score, mean (SD)

Range

Interpersonal relationship with caregiver

80.19

12.8 (0.12)

6 - 16

Parental involvement

77.75

3.11 (0.22)

1-4

Competence of the staff

81.33

4.88 (0.22)

2-6

Services offered by the health system

70.25

2.81 (0.37)

1-4

readmissions was lower in the FCC group compared with the control group, and the mean duration of hospitalisation was shorter compared with the control group (6.96 v.

12.96 days; p<0.001). Ortenstrand et al.[17] found a 5-day reduction in the duration of hospitalisation for preterm infants, and reported that parental involvement can directly affect the stability and morbidity of the infants. Parents who spend more time with their infants have more opportunities for perceiving the signs of infants’ discomfort and their other needs; consequently, they function better in comparison with nurses who are responsible for the care of multiple infants.[18] Parental satisfaction in a single-family room NICU was higher in comparison with the traditional open-bay NICU system. The single-family room environment seemed more conducive to the provision of FCC.[19,20] In our study, most of the patients were of low and lower-middle socioeconomic status, and were semi-skilled or unskilled workers with minimal family support. A study conducted in private sector hospitals may result in different levels of satisfaction and different responses. As expressed by a couple of parents, feeling intimidated by the doctors and the sophistication of the NICU may have resulted in lower levels of satisfaction.

There is a paucity of data from India, and ours is an initial attempt to rectify this. Open-ended questions were used and suggestions were also invited for improvement. A possible limitation of this study is ‘gratitude bias’, which was partly solved as the interviewer was not directly involved in the care of the baby. Also, the interviews were conducted on the day of discharge so that parents would not feel reluctant to give critical comments. On the other hand, this could have led to higher

satisfaction levels since their baby’s health had improved. If the interviews had been conducted when the baby was sick, this might have led to lower satisfaction levels.

Conclusion

This study provides valuable information regarding the various factors considered most important by parents in determining their overall satisfaction with their experiences of neonatal intensive care. Although a small fraction of parents reported satisfaction with the care being provided, most parents felt that there was significant scope for improvement in various aspects. Many parents were forthcoming with suggestions which, if implemented, could perhaps result in significantly higher levels of satisfaction.

Improved parental satisfaction with care, and the potential for enhanced FCC, need to be considered in decisions made regarding the configuration of NICU facilities in the future.

1. Institute of Medicine (US) Committee on Understanding Premature Birth and Assuring Healthy Outcomes. Behrman RE, Butler AS, eds. Preterm birth: Cuses, consequences, and prevention. Washington: National Academies Press (US), 2007. http://dx.doi.org/10.17226/11622 2. Fishering R, Broeder JL, Donze A. A qualitative study: NICU nurses as NICU parents. Adv Neonatal Care 2016;16(1):74-86. http://dx.doi. org/110.1097/anc.0000000000000221 3. Chourasia N, Surianarayanan P, Adhisivam B, Vishnu Bhat B. NICU admissions and maternal stress levels. Indian J Pediatr 2013;80(5):380-384. http://dx.doi.org/10.1007/s12098-012-0921-7 4. Bastani F, Abadi TA, Haghani H. Effect of family-centred care on improving parental satisfaction and reducing readmission among premature infants: A randomized controlled trial.

J Clin Diagn Res 2015;9(1):SC04-8. http://dx.doi. org/10.7860/jcdr/2015/10356.5444 5. Blackington SM, Mclauchlan T. Continuous quality improvement in the neonatal intensive care unit: Evaluating parent satisfaction. J Nurs Care Qual 1995;9(4):78-85. http://dx.doi. org/10.1097/00001786-199507000-00011

57

SAJCH

MARCH 2017 Vol. 11 No. 1

6. Ranchod T, Ballot DE, Martinez AM, Cory BJ, Davies VA, Partridge JC. Parental perception of neonatal intensive care in public sector hospitals in South Africa. S Afr Med J 2004;94(11):913-916.

7. Wainer S, Khuzwayo H. Attitudes of mothers, doctors, and nurses toward neonatal intensive care in a developing society. Pediatrics 1993;91(6):1171-1175.

8. Russell G, Sawyer A, Rabe H, et al. Parents’ views on care of their very premature babies in neonatal intensive care units: A qualitative study. BMC Pediatr 2014;14,230. http://dx.doi. org/10.1186/1471-2431-14-230 9. Franck LS, Oulton K, Nderitu S et al. Parent involvement in pain management for NICU infants: A randomized controlled trial. Pediatrics 2011;128(3):510-518. http://dx.doi.org/110.1542/ peds.2011-0272 10. Weiss S, Goldlust E, Vaucher YE. Improving parent satisfaction: An intervention to increase neonatal parent-provider communication. J Perinatol 2010;30(6):425-430. http://dx.doi.org/10.1038/ jp.2009.163 11. Reis MD, Scott SD, Rempel GR. Including parents in the evaluation of clinical microsystems in the neonatal intensive care unit. Adv Neonatal Care 2009;9:174-179. http://dx.doi.org/110.1097/ anc.0b013e3181afab3c 12. Lupton D, Fenwick J. ‘They’ve forgotten that I’m the mum’: Constructing and practicing motherhood in special care nurseries. Soc Sci Med 2001;53(8):1011-1021. http://dx.doi. org/110.1016/s0277-9536(00)00396-8 13. Conner JM, Nelson EC. Neonatal intensive care: Satisfaction measured from a parent’s perspective.

Pediatrics 1999;103(Suppl 1):S336-S349.

14. Wielenga JM, Smit BJ, Unk LK. How satisfied are parents supported by nurses with the NIDCAP model of care for their preterm infant?

J Nurs Care Qual 2006;21(1):41-48. http://dx.doi. org/110.1097/00001786-200601000-00010 15. Bakewell-Sachs S, Gennaro S. Parenting the post-NICU premature infant. Am J Matern Child Nurse 2004;29(6):398-403. http://dx.doi. org/10.1097/00005721-200411000-00011 16. Zelkowitz P, Feeley N, Shrier I, et al. The cues and care trial: A randomized controlled trial of an intervention to reduce maternal anxiety and improve developmental outcomes in very low birth weight infants. BMC Pediatr 2008;26:8-38. http://dx.doi.org/110.1186/1471-2431-8-38 17. Ortenstrand A, Westrup B, Broström EB, et al. The Stockholm neonatal family-centered care study: Effects on length of stay and infant morbidity.

Paediatrics 2010;125(2):278-285. http://dx.doi. org/110.1542/peds.2009-1511 18. Voos KC, Ross G, Ward MJ, Yohay AL, Osorio SN, Perlman JM. Effects of implementing familycentred rounds (FCRs) in a neonatal intensive care unit (NICU). J Matern Fetal Neonatal Med 2011;24(11):1403-1406. http://dx.doi.org/110.310 9/14767058.2011.596960 19. Stevens DC, Helseth CC, Khan MA, Munson DP, Reid EJ. A comparison of parent satisfaction in an open-bay and single-family room neonatal intensive care unit. Health Env Res Design 2011;4(3):110-123. http://dx.doi.org/110.1177/193758671100400309 20. Abdel-Latif ME, Boswell D, Broom M, Smith J, Davis D. Parental presence on neonatal intensive care unit clinical bedside rounds: Randomized trial and focus group discussion. Arch Dis Child Fetal Neonatal Ed 2015;100(3):F203-209. http:// dx.doi.org/110.1136/archdischild-2014-306724

CPD March 2017 The CPD programme for SAJCH is administered by Medical Practice Consulting.

CPD questionnaires must be completed online at www.mpconsulting.co.za

True (T) or false (F): Regarding medication errors in paediatric wards 1. Most medicines used in paediatrics are used off-label.

2. Medication errors have a lower incidence than the correct administration of medicines in paediatric in-patients.

3. Medication errors were more common in surgical disciplines than in paediatric medical disciplines.

Regarding breast development and menarche in Nigerian schoolgirls 12. Menarche usually occurs in early puberty.

13. The mean age of thelarche was 10.5 years.

14. The age of pubertal onset in similar in Nigerian girls to that of white American girls.

Regarding the nutritional status of children in schools in Limpopo 4. Less than 50% of children aged 9 - 13 years ate breakfast five or more times a week.

5. Overall less than 10% if children aged 9 - 13 years were stunted (<â&#x20AC;&#x201C;2SD).

Regarding strengthening of interventions to improve survival of severely malnourished children in SA hospitals 15. Severe acute malnutrition is defined as a weight for age z-score of <â&#x20AC;&#x201C;3.

16. Using the WHO 10-step treatment guidelines, in-hospital mortality associated with severe acute malnutrition should be <15%.

17. The positive effect of interventions persisted after withdrawal of the interventions.

Regarding oral health promotion in KwaZulu-Natal 6. Cochrane reviews are inconclusive about the effectiveness of school-based oral health programmes.

7. Less than 40% of 6-year-old children have evidence of dental caries of their primary teeth.

Regarding selenium status in HIV-positive children 8. Twice as many children with HIV infection had low selenium levels when compared with controls.

9. Selenium deficiency was associated with antiretroviral therapy in HIV-positive children.

Regarding the effect of HIV on mortality rates in severe acute malnutrition 18. Children with severe acute malnutrition are more likely to suffer from marasmus than kwashiorkor if they are HIV-infected.

19. Children with kwashiorkor were more likely to die than those with marasmus.

20.

Lower respiratory tract infection increases the mortality of children with severe acute malnutrition.

Regarding under-5 mortality and decision making in communities 10. Women empowerment plays no role in the uptake of antenatal care services by mothers in developing countries.

11. Women empowerment has been found to be an independent predictor of child mortality.

A maximum of 3 CEUs will be awarded per correctly completed test.

CPD questionnaires must be completed online via www.mpconsulting.co.za.

After submission you can check the answers and print your certificate.

Accreditation number: MDB015/172/02/2017 (Clinical)

58

SAJCH

MARCH 2017 Vol. 11 No. 1