SAMJ Vol 108, No 3 (2018) - Supplement by HMPG

MARCH 2018

Vol. 108 No. 3 (Suppl 1)

Maternal and Child Health Supplement

The production of this supplement was made possible with funding from UNICEF and the South African Medical Research Council.

March 2018 | VOL. 108 | NO. 3 (Suppl 1)

Maternal and Child Health Supplement

EDITOR Bridget Farham, BSc (Hons), PhD, MB ChB

Contents

EDITORS EMERITUS Daniel J Ncayiyana, MD (Groningen), FACOG, MD (Hon), FCM (Hon) JP de V van Niekerk, MD, FRCR

On the path to reach the SDG targets: Decreasing maternal and child mortality in South Africa Y Pillay, P Barron

Improvements in maternal mortality in South Africa J Moodley, S Fawcus, R Pattinson

HMPG

Reducing neonatal deaths in South Africa: Progress and challenges N Rhoda, S Velaphi, G S Gebhardt, S Kauchali, P Barron

S17

Closing the gaps to eliminate mother-to-child transmission (MTCT) of HIV in South Africa: Understanding MTCT case rates, factors that hinder the monitoring and attainment of targets, and potential game changers A Goga, W Chirinda, N K Ngandu, K Ngoma, S Bhardwaj, U Feucht, N Davies, M Ntloana, O Mhlongo, T Silere-Maqetseba, F Moyo, G Sherman

S25

Child mortality in South Africa: Fewer deaths but better data are needed L Bamford, N H McKerrow, P Barron, Y Aung

S33

Inpatient case fatality rates improvements in children under five: Diarrhoeal disease, pneumonia and severe acute malnutrition L Bamford, P Barron, S Kauchali, N Dlamini

on) S38

S44

S50

ASSOCIATE EDITORS Q Abdool Karim, A Dhai, R C Pattinson, A Rothberg, A A Stulting, J Surka, B Taylor, M Blockman, J M Pettifor, W Edridge, R P Abratt, D L Clarke

Innovations in the clinical care of mothers and children in South Africa: The contributions of district clinical specialist teams U Feucht, C Marshall, S Kauchali, P Barron, L Slavin, S Bhardwaj, Y Pillay Implementation of strategies to improve programme effectiveness lead to an improvement in maternal and child health outcomes in South Africa S Bhardwaj, R Pattinson, S Kauchali, N Dlamini, C Marshall, M van der Merwe, P Barron Notes

CEO AND PUBLISHER Hannah Kikaya Email: hannahk@hmpg.co.za MANAGING EDITORS Claudia Naidu Naadia van der Bergh TECHNICAL EDITORS Emma Buchanan Kirsten Morreira Paula van der Bijl PRODUCTION MANAGER Emma Jane Couzens DTP AND DESIGN Clinton Griffin CHIEF OPERATING OFFICER Diane Smith | Tel. 012 481 2069 Email: dianes@hmpg.co.za SALES MANAGER (CAPE TOWN) Azad Yusuf JOURNAL ADVERTISING Reneé Hinze Ladine van Heerden Makhadzi Mulaudzi Charmalin Comalie ONLINE SUPPORT Gertrude Fani FINANCE Tshepiso Mokoena HMPG BOARD OF DIRECTORS Prof. M Lukhele (Chair), Dr M R Abbas, Mrs H Kikaya, Prof. E L Mazwai, Dr M Mbokota, Dr G Wolvaardt ISSN 0256-9574 SAMA website: www.samedical.org Journal website: www.samj.org.za © Copyright: Health and Medical Publishing Group (Pty) Ltd, a subsidiary of the South African Medical Association Use of editorial material is subject to the Creative Commons Attribution – Noncommercial Works Licence. https:// creativecommons.org/licenses/by-nc/4.0 Printed by TANDYM PRINT

March 2018, Vol. 108, (3 Suppl 1)

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

EDITORIAL

On the path to reach the SDG targets: Decreasing maternal and child mortality in South Africa Despite some progress made between 2008 and 2015, South Africa (SA) did not reach the Millennium Development Goals targets for maternal and child health by the end of 2015. We cannot fail to reach the Sustainable Development Goals (SDGs) and targets. This means that we need to provide robust benchmarks and specific plans for the 2017 - 2030 period and implement the plans with a great sense of urgency. This series of papers on maternal, neonatal and child health provides an update on the progress to date and highlights critical issues that we need to attend to immediately. The Global Strategy for Women’s, Children’s and Adolescents’ Health 2016 - 2030[1] provides guidance on the key issues that countries should focus on, with an emphasis on interventions to ensure greater survival rates, the importance of thriving, as well as the need for transformation of the enabling environment. SA has signed up to the Global Strategy as well as the SDGs. The Global Strategy’s three objectives of survive, thrive and transform are underpinned by a range of action areas: country leadership; financing for health; health-system resilience; individual potential; community engagement; multisectoral action; innovation and research; and accountability for resources, rights and results. We should therefore read the series of papers in this supplement with these objectives and action areas in mind as well as the targets set in the SDGs. There are 7 papers in this series. The first[2] deals with the progress made in reducing institutional maternal mortality, i.e. mortality after admission to a health facility. While we have considerable data and knowledge of institutional maternal mortality through the work of the National Committee for Confidential Enquiry into Maternal Deaths (NCCEMD), we have a paucity of information on maternal deaths that occur outside health facilities, which is estimated to add about 20% additional deaths to those that are recorded in health facilities. This is an area that we need to address with a greater sense of urgency. The second paper[3] addresses neonatal mortality after admission to a public health facility using data collated by the National Perinatal Morbidity and Mortality Committee (NaPeMMCo). Similar to the data on maternal mortality, we have little information on neonatal deaths that occur outside health facilities. We need greater community involvement as well as to strengthen our community-based health services to generate these data and intervene to reduce the number of these deaths. This is true for all maternal, neonatal and child deaths that take place outside health facilities. The third paper[5] focuses on under-5 mortality using data collated by the Committee on Morbidity and Mortality in Children under 5 years (CoMMiC). The three ministerial committees, which prepare annual and triennial reports, have made a significant contribution to understanding the causes of maternal, neonatal and under-5 mortality and in making recommendations to the Minister of Health. By requesting their participation in this series, we hoped that their insights would reach a wider audience so that we can be well on the path to meeting the targets of the SDGs. In their paper on maternal mortality, Moodley et al.[2] document the impressive annual decline in the institutional maternal mortality

rate from a peak of 189 per 100 000 in 2009 to 135 per 100 000 in 2016. This means that in 2016 there were around 400 fewer maternal deaths than in 2009. The main reason for this was the large-scale reduction in deaths owing to HIV infection as a result of the success of the HIV antiretroviral treatment programme in pregnancy and beyond.[2] After additional training in the essential steps in the management of obstetric emergencies was provided, there was also a decline in deaths due to obstetric haemorrhage between 2015 and 2016. However, more needs to be done to reduce deaths owing to pregnancy-related hypertension and it is hoped that the introduction of additional antenatal visits (increased from 5 to 8 visits following new WHO guidelines) will assist in this regard. However, there is still much to do to achieve the SDG target of 70 per 100 000 by 2030.[4] Every manager in the healthcare system working with communities must be supported to redouble efforts to reduce maternal mortality. This includes the provision of good quality sexual and reproductive health services, ensuring early antenatal care attendance and safe delivery. Neonatal mortality contributes about one-third of under-5 mortality in SA. Rhoda et al.[3] show the slow decline in the neonatal mortality rate over the past 15 years. Even though SA is very close to the SDG target of 12 neonatal deaths per 1 000 live births, we should be doing better as a middle-income country, and experts are of the opinion that roughly 50% of all neonatal deaths in SA are avoidable. According to NaPeMMCo, the key causes of neonatal mortality continue to be asphyxia, prematurity and infections. A focus on training in the Helping Babies Breathe programme will assist in reducing deaths from asphyxia, and reducing teen pregnancies will reduce deaths related to prematurity. An area for additional focus is the high rate of stillbirths. Some of these issues relate to intrapartum care, which emphasises the need for midwives, obstetricians and gynaecologists to work very closely with neonatologists and paediatricians. In their paper, Bamford et al.[5] demonstrate the remarkable fall of inpatient mortality rates (around 40 - 50%) for three of the most important causes of mortality in children under 5 years of age, i.e. pneumonia, diarrhoea and malnutrition. Similar to its impact on maternal mortality, the contribution of HIV/AIDS to under-5 mortality has declined significantly in the past few years but it remains high. Although we are unable to precisely quantify the numbers of reduced deaths due to HIV, we can quantify the success of the prevention of mother-to-child transmission (PMTCT) programme, which is shown in the paper on PMTCT. Notwithstanding these dramatic improvements, the authors make the point that SA has a much higher child mortality rate than comparable middle-income countries such as Brazil and Mexico.[6] Bamford et al.[5] also suggest a number of interventions (e.g. increased exclusive breastfeeding, and greater access to water and sanitation) that would further reduce fatality rates from these diseases. In the article on improvements in child mortality, Bamford et al.[7] highlight the need for improved data quality with which to quantify and measure child deaths. Part of the data problem is that as many as half of all deaths during the postneonatal period occur out of hospital and accurate population metrics are dependent

March 2018, Vol. 108, (3 Suppl 1)

EDITORIAL

on all child deaths being accurately reported to the Department of Home Affairs. Bamford et al.[7] highlight the importance of focusing on trends at the provincial level – one might add district and facility levels. There is significant variation in mortality rates across provinces, with the Western Cape usually having the lowest mortality rates. It is unacceptable, in our quest for equity in outcomes, that the Free State has double the under-5 mortality rate (49.6/1 000 live births) of the Western Cape’s 24/1 000 live births. Many of the improvements in maternal, neonatal, child and inpatient mortality have resulted from improvements in the quality of care. Previous improvements in the PMTCT have been well documented.[8-10] In their paper, Goga et al.[11] show how there have been further decreases in the rate of early (in utero) transmission of HIV from mother to child. Improvements have occurred as a possible result of the use of ward-based outreach teams and community caregivers to trace HIV-exposed infants and connect mothers to care. Continuous quality improvement processes within facilities and districts have also helped to decrease transmission rates. The introduction of birth testing and early initiation on antiretroviral treatment of infants who test HIV-positive, as well as ensuring that HIV-positive mothers are virally suppressed throughout the breastfeeding period, will also go a long way to reduce mother-to-child transmission of HIV. To qualify for elimination, we need to reach the WHO/UNICEF goal of <50 paediatric HIV infections per 100 000 live births. Other examples in quality improvement are shown by Bhardwaj et al.,[12] who use three case studies, i.e. training in essential steps in managing obstetric emergencies (ESMOE), management of undernutrition in young children and strengthening of exclusive breastfeeding, to illustrate how improved quality of care and coverage achieved better health outcomes for maternal and child mortality. Lastly, but not least, Feucht et al.,[13] demonstrate through a number of examples that the district clinical specialist teams deployed in districts around the country to improve clinical governance have impacted on a range of practices that are linked to improved care for maternal and child patients.[13] Clearly, SA has done much to improve maternal and child health over the last decade and is well on track to achieve the SDGs for neonatal mortality. However, as a middle-income country SA should be doing better. The HIV epidemic has taken its toll but increasingly is being brought under control, with consequent declines in both maternal and child mortality. To go the next mile will require much greater consistency in improvement of quality throughout the health system. These improvements must be accompanied by strengthened community systems and greater involvement of men as partners, both in the context of maternal health as well as in the health of their children. In this context it is worrying that the recent STATSSA report[14] shows that there were 1 444 births to children 10 - 14 years old, with a further 107 730 births to adolescents aged 15 - 19 years old! The quest for universal health coverage as reflected in the national health insurance policy must be matched by quality improvements in every home, community and health facility, with each individual playing his/her part to achieve this goal.

Y Pillay National Department of Health, Pretoria, South Africa

P Barron School of Public Health, University of the Witwatersrand, Johannesburg, South Africa pbarron@iafrica.com

1. World Health Organization. The Global Strategy for Women’s, Children’s and Adolescents’ Health (2016 - 2030). Geneva: WHO, 2015. http://www.who.int/life-course/partners/global-strategy/ewecglobalstrategyreport-200915.pdf?ua=1 (accessed 1 October 2017). 2. Moodley J, Fawcus S, Pattinson R. Improvements in maternal mortality in South Africa. S Afr Med J 2018;108(3 Suppl 1):S4-S8. https://doi.org/10.7196/SAMJ.2018.v108i3.12770 3. Rhoda N, Velaphi S, Gebhardt G, Kauchali S, Barron P. Reducing Neonatal Deaths in South Africa: Progress and challenges. S Afr Med J 2018;108(3 Suppl 1):S9-S16. https://doi.org/10.7196/SAMJ.2018. v108i3.12804 4. United Nations. Sustainable Development Goals. United Nations 2015. https://www.un.org/pga/wpcontent/uploads/sites/3/2015/08/120815_outcome-document-of-Summit-for-adoption-of-the-post2015-development-agenda.pdf (accessed 24 September 2017). 5. Bamford L, Barron P, Kauchali S, Dlamini N. In-patient case fatality rates improvements in children under five: Diarrhoeal disease, pneumonia and severe acute malnutrition. S Afr Med J 2018; 108(3 Suppl 1):S33-S37. https://doi.org/10.7196/SAMJ.2018.v108i3.12772 6. UNICEF. Countdown to 2015 maternal newborn and child survival: A Decade of Tracking Progress for Maternal, Newborn and Child Survival: The 2015 Report. New York: UNICEF, 2015. http:// countdown2030.org/documents/2015Report/Countdown_to_2015_final_report.pdf (accessed 24 September 2017). 7. Bamford L, McKerrow NH, Barron P, Aung Y. Child mortality in South Africa: Fewer deaths but better data are needed. S Afr Med J 2018;108(3 Suppl 1):S25-S32. https://doi.org/10.7196/SAMJ.2018. v108i3.12779 8. Barron P, Pillay Y, Doherty T, et al. Eliminating mother-to-child HIV transmission in South Africa. Bull World Health Organ 2013;91:70-74. https://doi.org/10.2471/BLT.12.106807 9. Bhardwaj S, Barron P, Pillay Y, et al. Elimination of mother-to-child transmission of HIV in South Africa: Rapid scale-up using quality improvement. S Afr Med J 2014;104(3 Suppl 1):239-243. https:// doi.org/10.7196%2FSAMJ.7605 10. Sherman G, Lillian R, Bhardwaj S, Candy S, Barron P. Laboratory information system data demonstrate successful implementation to the prevention of the mother-to-child transmission programme in South Africa. S Afr Med J 2014;104(3 Suppl 1):235-238. https://doi.org/10.7196%2FSAMJ.7598 11. Goga A, Chirinda W, Ngandu NK, et al. Closing the gaps to eliminate mother-to-child transmission (MTCT) of HIV in South Africa: Understanding MTCT case rates, factors that hinder the monitoring and attainment of targets, and potential game changers. S Afr Med J 2018;108(3 Suppl 1):S17-S24. https://doi.org/10.7196/SAMJ.2018.v108i3.12817 12. Bhardwaj S, Pattinson R, Kauchali S, et al. Implementation of strategies to improve programme effectiveness lead to an improvement in maternal and child health outcomes in South Africa. S Afr Med J 2018;108(3 Suppl 1):S44-S49. https://doi.org/10.7196/SAMJ.2018.v108i3.12812 13. Feucht U, Marshall C, Kauchali S, et al. Innovations in the clinical care of mothers and children in South Africa: The contribution of district clinical specialist teams. S Afr Med J 2018;108(3 Suppl 1): S38-S43. https://doi.org/10.7196/SAMJ.2018.v108i3.12808 14. Statistics South Africa (StatsSA). Recorded Live Births 2016. Pretoria: StatsSA, 2017. http://www. statssa.gov.za/publications/P0305/P03052016.pdf (accessed 1 October 2016).

S Afr Med J 2018;108(3 Suppl 1):S2-S3. DOI:10.7196/SAMJ.2018.v108i3.12901

March 2018, Vol. 108, (3 Suppl 1)

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

RESEARCH

Improvements in maternal mortality in South Africa J Moodley,1 MB ChB, FCOG, FRCOG; S Fawcus,2 MA(Oxon), MB BCh, FRCOG; R Pattinson,3 BSc, MB BCh, FRCOG, MD, FCOG (SA), MMed (O&G) Department of Obstetrics and Gynaecology, School of Clinical Medicine, College of Health Sciences, Nelson R Mandela School of Medicine, University of KwaZulu-Natal, Durban, South Africa; and Chairperson, National Committee for Confidential Enquiries into Maternal Deaths, South Africa 2 Department of Obstetrics and Gynaecology, Faculty of Health Sciences, University of Cape Town, South Africa; and Deputy Chairperson, National Committee for Confidential Enquiries into Maternal Deaths, South Africa 3 South African Medical Research Council Maternal and Infant Health Care Strategies Unit; Department of Obstetrics and Gynaecology, Faculty of Health Sciences, University of Pretoria, South Africa; and Editor, National Committee for Confidential Enquiries into Maternal Deaths report, South Africa 1

Corresponding author: R Pattinson (robert.pattinson@up.ac.za)

South Africa introduced confidential enquiries into maternal deaths in 1997, and the first report was published in 1998. Subsequently, six triennial reports have been published, and the seventh was being prepared at the time of submission of this manuscript. This paper is focused on data from the last 5 years and documents the reduction in maternal deaths since 2009. The main reason for this decline in maternal deaths has been the success of the antiretroviral treatment programme for HIV-positive women. There has also been a decline in deaths owing to obstetric haemorrhage, but deaths due to complications of hypertension remain stubbornly high. There has been a continuous reduction in potentially preventable maternal deaths since 2008. However, there is still much to do, and efforts need to concentrate on improving the health system to reduce the deaths due to haemorrhage and hypertension. S Afr Med J 2018;108(3 Suppl 1):S4-S8. DOI:10.7196/SAMJ.2018.v108i3.12770

Deaths of women during pregnancy, childbirth and the puerperium remain a major public health issue, particularly in low- and middle-income countries, such as South Africa (SA). Clinical audits have contributed considerably to the reduction of maternal deaths in high-income countries such as the United Kingdom.[1] As far as the authors are aware, SA is the only country in Africa to have a sustained and functional National Confidential Enquiry into Maternal Deaths Committee (NCCEMD). The NCCEMD is a ministerial committee of experts representing obstetrics, midwifery, anaesthesia and representatives from each of the nine provinces with the Department of Health’s support. It was mandated to analyse reported maternal deaths and to make recommendations that would result in a reduction in the Maternal Mortality Ratio (MMR).[2] The NCCEMD process was modelled on the UK CEMD process, which has been operational since 1952, and advisors from the UK committee assisted SA to develop its enquiry process. This process has evolved over the years and the current enquiry process is illustrated in Fig. 1, which shows that there are essentially two recommended audit loops in the process. Firstly, there is discussion at the facility where the death occurred, so that avoidable factors can be identified immediately and remedied locally. The

to the co-ordinator together with a photocopy of all clinical records, after which it is assessed by independent provincial assessors that are teams that include obstetricians, medical officers and midwives, and anaesthetists. Assessors work as a pair (usually a midwife and an obstetrician or an experienced

second audit loop is the CEMD process, which is as follows: The maternal death is reported to the provincial maternal, child and women’s health (MCWH) co-ordinator, who allocates a unique number to the case. A Maternal Death Notification Form (MDNF) that was designed to capture information is completed and is sent

Maternal death MNDF and reports completed within 3 weeks

Facility

Death notified and a unique number given

Sent Provincial assessor

Returned within 30 days

Provincial MCHW

Local remedial action Anaesthetic assessments Quality assurance

Provincial assessor's report sent

MMaMMAS entry (provincial assessor)

NCCEMD Secretariat

Provincial data set

NCCEMD

Distributed to provinces, which distribute information to districts

Report

All basic data destroyed

Fig. 1. The process of CEMD in South Africa. (MDNF = maternal death notification form; MCWH = maternal, child and women’s health; MaMMAS = maternal morbidity and mortality audit system; NCCEMD = National Committee for the Confidential Enquiries into Maternal Deaths.)

March 2018, Vol. 108, (3 Suppl 1)

Fig. 4 illustrates the change in the iMMR of the underlying conditions causing maternal deaths. Since 2011 there has been a major reduction in deaths due to non-pregnancy-related infections (NPRI) and this has been the main reason for the overall reduction in the iMMR (risk ratio (RR) 0.52; 95% confidence interval (CI) 0.41 - 0.61; p<0.00001). The NPRI category includes deaths due to tuberculosis, pneumonia, meningitis and malaria. Approximately 95% of these women were HIV-infected. The reduction reflects the major impact the PMTCT and ARV programmes have had on maternal deaths. There has also been a small drop in the number of deaths due to obstetric haemorrhage (OH) (RR 0.85; 95% CI 0.71 - 1.04; p=0.12). Deaths due to the complications of hypertension in pregnancy have shown a slight increase (RR 1.12; 95% CI 0.93 1.36; p=0.21). Data from the recent triennial and interim SMRs demonstrate a concerning increase in the numbers of and case fatality rates (CFRs) for excessive bleeding associated with caesarean delivery (BLDACD),

19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 19

98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16

Deaths, N

Fig. 2. Maternal deaths reported to the NCCEMD between 1998 and 2016.

190 180 200 170 190 160 180 150 170 140 160

130

150

08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16

130

140

05 20 06

120

000 live births iMMR/100iMMR/100 000 live births

200

120 12 20 13 20 14 20 15 20 16

05 20 06

Fig. 3. iMMR per year for South Africa: 2005 - 2016. (iMMR = institutional maternal mortality rate.) 70 60

2011

2012

2013

2014

2015

2016

0 HT OH

Ec car RS AR mb P E is M

. nk

&S AC P N M U

Fig. 4. Trend in iMMR per underlying cause: 2011 - 2016. (iMMR = institutional maternal mortality rate; HT = complication of hypertension in pregnancy; OH = obstetric haemorrhage; EC = ectopic pregnancies; Miscar = miscarriage; PRS = pregnancy-related sepsis; AR = anaesthetic-related deaths; Emb = embolism; AC = acute collapse, cause unknown; NPRI = non-pregnancy related infections; M&S = preexisting medical and surgical conditions, Unk. = unknown.)

which 100 increased from 78 cases in 2002 - 2004 to 221 during 2011 - 2013. This number reduced slightly to 217 in 2014 - 2016. The 80 CFR from BLDACD has dropped from 33.1 deaths per 100 000 CDs 60 (2011 - 2013) to 31 (2014 - 2016). Annual reviews indicate that the 40 BLDACD CFR was highest in 2012 at 40 cases and has since declined [4] to 30 20cases in 2016. The majority of these deaths (98%) occurred at district hospitals (DHs) in the public sector, where the majority of 0 the CDs were performed. A significant number of deaths occurred at provincial tertiary hospitals, probably due to the fact that women were referred to tertiary centres from DHs after CD with unresolved bleeding. There were also large inequities between provinces in BLDACD CFRs, ranging from 76.05 in Mpumalanga to 15.6/100 0000 CD in the Western Cape during the 2014 - 2016 period.

Potentially preventable deaths, %

Causes

Year

Trends over time

Fig. 2 illustrates the number of maternal deaths reported to the NCCEMD since its inception. Since 2009 there has been a fall in the number of maternal deaths compared with the 2003/4 era. The slight declines at the end of 2001, 2004, 2007, 2010 and 2013 were due to the closure of the database at the end of each triennium for data analysis. Fig. 3 illustrates that the institutional MMR (iMMR) declined substantially from 2009 and appeared to level off in 2016, when 64 fewer deaths were reported and there was a decline in the number of live births. This decrease in the iMMR has also been demonstrated in all ways of measuring maternal deaths by different organisations.[3]

1000 800 600 400 200 0

iMMR/100 000 live births

medical officer) in order to overcome inherent biases. All deaths where the woman received an anaesthetic are also assessed by an anaesthetist. Assessors are appointed by the provincial departments based on their knowledge of the subject and respect within the maternal, newborn and child healthcare fraternity. Assessors identify causes and avoidable factors using a structured form, and the data are entered into an electronic data-collection system, the Maternal Morbidity and Mortality Audit System (MMaMMAS), which then collates information from all the provinces. There is a quality-control component at the provincial level to ensure that assessments are as accurate as possible. At the national level, the MMaMMAS data are used to generate tables and the information is collated into reports. These reports, called Saving Mothers reports (SMRs), are produced as annual interim reports and more comprehensive triennial reports, which include chapters on each of the major conditions that cause death. Prior to publication of the reports, there are national and provincial stakeholder discussion meetings to identify the key recommendations that arise from the data. These recommendations must be implemented to reduce the maternal mortality ratio (MMR) in SA. The key recommendations to the provinces, districts and academic institutions are discussed at educational sessions, where the SMRs are distributed. The CEMD process, which is detailed in the Saving Mothers comprehensive reports, is confidential, and all copies of case notes are destroyed after publication of the reports. The data collected by the enquiry and the specific notification forms are for use by the CEMD process only, and cannot be used for medicolegal or disciplinary processes that, when they occur, are completely separate and parallel. This has been ratified by relevant judicial bodies. The NCCEMD also produces guidelines for completion of the MDNF.

Deaths, N

RESEARCH 2000 1800 1600 1400 1200 1000 800 2000 600 1800 400 1600 200 1400 0 1200

March 2018, Vol. 108, (3 Suppl 1)

M&S NPRI Ec Miscar. PRS OH HT AR Ad. drug Emb. AC Miscell. Unk. Total MD

iMMR/100 000 live births

RESEARCH

2011

2012

2013

2014

2015

2016

0 HT

Ec car RS AR mb P E is M

I k. AC PR &S Un N M

M&S NPRI Ec Miscar. PRS OH HT AR Ad. drug Emb. AC Miscell. Unk. Total MD

100 80 60 40 20 0

iMMR/100 000 iMMR/100 000live livebirths births

Fig. 5. Potentially preventable deaths per underlying cause: 2014 - 2016. (M&S = pre-existing medical and surgical conditions; NPRI = nonpregnancy-related infections; EC = ectopic pregnancies; Miscar. = miscarriage; PRS = pregnancy-related sepsis; OH = obstetric haemorrhage; HT = complication of hypertension in pregnancy; AR = anaesthetic-related deaths; Ad. drug = adverse drug effects; Emb. = embolism; AC = acute collapse, cause unknown; Miscell. = miscellaneous; Unk. = unknown; MDs = maternal deaths.)

200

200 180 180 160 160 140 140 120 120 100 100 80 80 60 60 40 40 20 20 0

76.22

61.4661.46

51.03 51.03

NP iMMR NP iMMR PP iMMR PP iMMR

100 100

92.6 92.6

83.3

0 2008 - 2010 2011 - 2013 2014 - 2016 2008 - 2010 2011 - 2013 2014 - 2016 Triennium

Triennium

Fig. 6. Potentially preventable and non-preventable maternal deaths per triennium. (iMMR = institutional mortality rate; PP= potentially preventable; NP = non-preventable.) 35

2008 - 2010

2011 - 2008 2013 - 2010

2011 - 2013

2014 - 2016

25 20

2014 - 2016

20 10 15 5 5

E M C isc ar. PR S AR Em b. AC M& S

RI OH HT

0 10

Preventable Preventable deaths,deaths, % %

AR Em b. AC M& S

E M C isc ar. PR S

At the end of the assessment of every maternal death, the assessors (a doctor and professional nurse) are asked to classify the deaths as follows: unavoidable; unavoidable but there was some substandard care; possibly avoidable, and there was some substandard care that may have impacted on the death of the woman; and probably avoidable, where there was some substandard care that probably contributed significantly to the death of the woman. The cases were assessed for factors within the control of the healthcare system, and factors related to the behaviour of the woman and/or her family were excluded. Potentially preventable maternal deaths are defined as possibly and probably avoidable deaths. Although subjective, there are over 100 assessors in the country and this assessment is probably fairly robust. Fig. 5 illustrates the potentially preventable deaths in the last triennium for the underlying causes of maternal deaths. Overall, ~60% of all maternal deaths were thought to be potentially preventable by the assessors, and this has been the case since the inception of this method of assessment in 2008, which indicates poor quality of care. It is possible to calculate the iMMR of the potentially preventable maternal deaths, and Fig. 6 illustrates the trend for the three triennia. There has been a consistent decline (17%) in the iMMR of preventable deaths since 2008, which suggests continued improvement in the quality of care. There has been a noteworthy decline (33%) in non-preventable maternal deaths, which likely reflects a decline in patient/communityrelated modifiable factors. Owing to its predominantly facility-based nature, the NCCEMD can assess health system modifiable factors related to the healthcare provider’s management and administrative factors. Thus, potentially preventable deaths refer to preventable factors within the healthcare system. Less well explored by the NCCEMD process are, however, patient/community-related factors that are influenced by a multitude of factors, such as access to social grants, improved knowledge of health needs and rights and the influence of the MomConnect initiative, which are less well explored by the NCCEMD. These factors may have changed over the years but have not been adequately explored owing to limitations in the NCCEMD methodology. According to NCCEMD assessors, ‘delay in accessing medical help’ by women accounted for about one-third of assessable deaths in previous reports, but this figure is not included in the sum of the total preventable deaths illustrated in Figs 6 - 8. The distribution of the major causes of potentially preventable maternal deaths over the three triennia indicated that the contribution of NPRI to the iMMR was declining, with a concomitant increase in the importance of HT and OH (Fig. 7). Originally, NPRI deaths were the most common causes of potentially preventable maternal deaths and are currently the third most common cause. An analysis of the iMMR of potentially preventable deaths for the underlying causes of maternal deaths during the three triennia (Fig. 8) revealed a marked reduction in iMMR due to NPRI, and a small reduction in OH, PRS and anaesthetic-related iMMR, with no changes for hypertension-related deaths. The reduction in potentially preventable deaths is mainly attributed to the introduction of an improved HIV testing and ARV treatment strategy in 2010. The less-marked reduction in the direct cause of maternal deaths reflects some improvement in the quality of care, but this is inadequate and it remains disappointing that there have not been greater reductions in deaths due to haemorrhage and hypertension in pregnancy. The management of these two conditions requires adequate functioning of the maternity healthcare system in terms of treatment, referral systems and personnel. In particular, deaths due to haemorrhage occur very quickly after the onset of

Potentially preventable deaths, %

Trends in preventable maternal deaths

Fig. 7. Distribution of potentially preventable deaths: 2008 - 2016. (NPRI = non-pregnancy-related infections; OH = obstetric haemorrhage; HT = complication of hypertension in pregnancy; EC = ectopic pregnancies; Miscar. = miscarriage; PRS = pregnancy-related sepsis; AR = anaestheticrelated deaths; Emb. = embolism; AC = acute collapse, cause unknown; M&S = pre-existing medical and surgical conditions.)

bleeding, and the patients do not survive referral. This necessitates definitive care at the place where the bleeding occurs, which is frequently level one. All of the above-mentioned aspects of care have not improved sufficiently. Despite the overall reduction in maternal deaths the iMMR is still too high, and there are a number of unaddressed challenges.

March 2018, Vol. 108, (3 Suppl 1)

iMMR/100 000 live births

RESEARCH

2008 - 2010

35 30 25 20 15 10 5 0

2011 - 2013 2014 - 2016

programme is currently being implemented and requires scale-up to all appropriate facilities. Greater reductions in the iMMR are anticipated with the implementation of the Safe CD package of care

Improve antenatal care

RI NP

E M C isc ar PR S AR Em b AC M& S

Continuously improve the knowledge and skills of all healthcare professionals

Challenges for further reduction in MMR in South Africa

Fig. 8. iMMR of potentially preventable deaths over the three triennia: 2008 - 2016. (iMMR = institutional maternal mortality rate; NPRI = non-pregnancy related infections; OH = obstetric haemorrhage; HT = complication of hypertension in pregnancy; EC = ectopic pregnancies; Miscar. = miscarriage; PRS = pregnancy-related sepsis; AR = anaestheticrelated deaths; Emb. = embolism; AC = acute collapse, cause unknown; M&S = pre-existing medical and surgical conditions.)

An assessment of maternal and perinatal deaths due to complications of hypertension involve two major factors, i.e. the failure to refer women with hypertension, mostly from the primary healthcare clinics, and infrequent antenatal visits. The first factor was addressed by concentrating training on hypertension and creating mechanisms for women to be easily referred.[5] The second factor was not directly related to the women; rather, it was a function of the attendance schedule for women who were classified as having low-risk pregnancies, which initially stipulated 6-weekly visits from 20 weeks’ gestation. This was too infrequent to detect sudden changes in a woman’s condition. Pre-eclampsia can occur rapidly and, therefore, the National Department of Health (NDoH), through the National Health Council, has decided to increase the antenatal visits in the third trimester to fortnightly. This should have a major impact on efforts to identify women with hypertension in pregnancy and ensure that they are referred to the appropriate place.

Safe CD

Bleeding associated with CD has been identified as a problem that still requires attention, despite the small observed decrease in CFRs during the 2014 - 2016 period. In addition, CD is an underlying factor in other causes of maternal deaths such as puerperal sepsis, thromboembolism and eclampsia. The NCCEMD has addressed the problem of safety of CD by implementing the following measures: • Clinical protocols: NCCEMD algorithms in SMRs; postpartum haemorrhage (PPH) and caesarean-section monographs; and a PPH training programme, including a lecture, accompanied dissemination of the PPH monograph throughout the country in 2011. • Practical training: Since 2008, the Essential Steps in Managing Obstetric Emergencies (ESMOE) training programme has taught practical modules on managing bleeding at CD. The programme was recently updated to include training videos on the surgical management of PPH and a DVD training lecture on PPH. The videos and lecture were developed by the Health System Trust in 2014 and the material was circulated to district clinical specialist teams (DCSTs) to use for teaching in the districts. • The requirements for a facility to be able to provide safe CD were stipulated in a consensus document by the NCCEMD. The document is titled ‘Setting minimum standards for safe CD’ and is directed at facility and healthcare managers. • An action plan has been detailed for the implementation of safe CD. The plan has clearly allocated responsibilities for healthcare personnel. • The developments of accreditation tools to assess: (a) the competency of the surgeon performing the CD; (b) the competency of the doctor providing the anaesthetic; and (c) the compliance of the facility with the minimum standards. This may involve consolidation of services performing CD together with improved availability of onsite interfacility transport for referrals. The recent declines in the number of deaths associated with obstetric haemorrhage and BLDACD have followed on the implementation of some of the abovementioned initiatives, for which all the material has been expanded and compiled into a Safe CD package of care – the

The NDoH has conducted an intensive outreach in training healthcare professionals in managing obstetric emergencies. This was done using the ESMOE and ensuring that emergency obstetric simulation training exercises were performed. In areas where these programmes have been implemented, the maternal deaths have decreased by 29.3% overall and the direct causes of maternal death decreased by 17.5%.[6] The expansion of these programmes to all districts is an ongoing process.

Development of strategies to improve the identification of patient/community-related factors in maternal deaths Here, there are two major components: (a) gaining a greater understanding of patient/community-level preventable factors; and (b) identifying non-facility deaths.

Patient/community-level preventable factors These factors are poorly evaluated in the NCCEMD process, but delays in seeking medical care remains the most common community-related problem. Strategies to circumvent this problem include improving knowledge of and access to care, with initiatives such as expanding the role of community health workers (CHWs) in ward-based outreach teams (WBOTS) to identify and monitor pregnant women, use of the MomConnect antenatal messaging service and providing maternity waiting areas for women who live far from delivery facilities. Maternal deaths outside of institutions The discrepancy between NCCEMD maternal mortality rates and World Health Organization global estimates is probably due to the occurrence of maternal deaths at home/outside facilities, which are not counted or analysed in the NCCEMD process. These deaths may be classified as any of the following: women who never accessed maternity care; women who were booked but were unable to reach a facility at the time of an emergency; or women who died postpartum at home after delivery at a facility. At the time of submission of this manuscript, the NCCEMD was exploring three mechanisms for improving identification and analysis of these deaths:

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

• Working with groups conducting ongoing demographic health surveillance (DHS) such as those in Agincourt, Mpumalanga and the Africa Centre for Health and Population Studies in KwaZuluNatal – both provide population estimates of maternal mortality. • Working with the Burden of Disease Unit at the Medical Research Council of SA, which receives Statistics SA data derived from death certificates (DHA-1663 forms) to enable identification of maternal deaths from the recent pregnancy question on the death certificate or the codes used to classify the cause of death. In addition, deaths are categorised by location (i.e. facility or outside). • Working with traditional leaders/chiefs who are informed of deaths in their communities, and who are able to notify authorities of deaths using the DHA-1680 forms, has been investigated in KwaZulu-Natal and Limpopo as a means to improve the reporting of home deaths. A preliminary review of these three mechanisms indicates that at least a quarter of maternal deaths occur outside facilities, but the pattern of causes is similar to that revealed by the NCCEMD data. The NCCEMD needs to develop a system based on the WHO verbal autopsy technique for exploring both the circumstances around deaths outside of facilities and the community/patient-level avoidable factors in deaths outside institutions. The verbal autopsy approach and narrative summaries for each death in the DHS sites can provide insight into these issues. This needs to be replicated across the country, possibly by CHWs and WBOTS. This remains a challenge for the NCCEMD but would provide very useful information on the many reasons that delay or obstruct women in accessing care timeously.

Conclusion

The SA NCCEMD recognises that confidential enquiries offer a good opportunity to link clinical practice with health policy to impact maternal mortality. This has been achieved by overcoming resistance to confidential enquiries, such as a lack of time, poor documentation, lack of interest or accountability, poor organisation, fear of litigation and deaths outside of the healthcare facilities. The last is still a problem in SA, but the NCCEMD intends to work with non-governmental organisations to conduct interviews with key informants. This has been shown to produce useful information in other countries.[7] Such enquiries may be a catalyst to improve documentation and practice in information gathering about events that occur at the community level in rural health districts. The NCCEMD has shown that setting targets for the reduction of deaths due to specific causes, and focusing on the provision of information and training in these areas, results in significant

reductions in the number of maternal deaths. The strategies to reduce deaths related to BLDACD have had some positive effects and it is anticipated that implementing the Safe CD package of care will enable further improvements in the quality of care of women having CDs, thereby reducing maternal mortality. Future interventions are more difficult to implement as they involve improving the healthcare system. Managing complex conditions such as obstetric haemorrhage and hypertension, which often involve many role-players in different healthcare facilities and emergency care services, can only be effective if each aspect of the system works efficiently and in harmony. Improving knowledge, skills and resources is not sufficient; the system must function efficiently. Acknowledgements. We would like to thank the members of the NCCEMD and all the provincial assessors for their continued contributions to reducing maternal deaths – their unstinting work is paramount to the success of the work of the NCCEMD. We would also like the thank the Minister of Health, Dr Aaron Motsoaledi, for permission to publish the article and Dr Peter Barron for reviewing it. Author contributions. All authors contributed equally to the conceptualisation and writing of the manuscript. RCP extracted the relevant data from the NCCEMD reports. Funding. None. Conflicts of interest. None.

1. United Nations Development Programme. Human Development Report 2011. Sustainability and Equity: A Better Future for All. New York: Palgrave Macmillan, 2011. 2. National Department of Health (NDoH). Saving Mothers: First Comprehensive Report on Confidential Enquiries into Maternal Deaths in South Africa. Pretoria: NDoH, 1998. http://www.kznhealth.gov.za/ mcwh/Maternal/Saving-Mothers-2011-2013-short-report.pdf (accessed 20 August 2017). 3. Dorrington RE, Bradshaw D, Laubscher R, Nannan N. Rapid Mortality Surveillance Report 2015. Cape Town: South African Medical Research Council, 2016. http://www.mrc.ac.za/bod/ RapidMortalitySurveillanceReport2015.pdf (accessed 20 August 2017). 4. National Committee for Confidential Enquiries into Maternal Deaths. Saving Mothers 2015: Annual Report and Detailed Analysis of Maternal Deaths Due to Bleeding During or After Caesarean Delivery. Pretoria: NDoH, 2016. 5. GS Gebhardt, R Pattinson, J Hofmeyr, et al. Basic Antenatal Care (BANC) Package in South Africa. Motivation to Increase the Routine Number of Antenatal Visits. Pretoria: NDoH, 2016. 6. Pattinson RC, Moodley J. Essential Steps in Managing Obstetric Emergencies (ESMOE) – Emergency Obstetric Simulation Training (EOST) Scale-up Programme: Effects on knowledge and skills of healthcare professionals, functionality of healthcare facilities and maternal and perinatal mortality. Report to National Department of Health. Pretoria: South African Medical Research Council, 2016. 7. Mgawadere F, Unkels R, Kazembe A, van den Broeke N. Factors associated with maternal mortality in Malawi: Application of the 3 delays model. BMC Pregnancy Childbirth 2017;17(1):219. https://doi. org/10.1186/s12884-017-1406-5

Accepted 26 September 2017.

March 2018, Vol. 108, (3 Suppl 1)

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

RESEARCH

Reducing neonatal deaths in South Africa: Progress and challenges N R Rhoda,1 MB ChB, FCPaed, Certificate in Neonatology; S Velaphi,2 MB ChB, FCPaed; G S Gebhardt,3 MB ChB, MMed, MSc (MedSci), FCOG, PhD; S Kauchali,4 FCPaed, MS, MPhil; P Barron,5 FFCH Department of Neonatology and Paediatrics, Faculty of Health Sciences, University of Cape Town, South Africa Department of Paediatrics, School of Clinical Medicine, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa 3 Department of Obstetrics and Gynaecology, Faculty of Health Sciences, Stellenbosch University, Stellenbosch, and Tygerberg Hospital, South Africa 4 National Department of Health, South Africa, and Department of Paediatrics and Child Health, University of KwaZulu-Natal, Durban, South Africa 5 School of Public Health, University of the Witwatersrand, Johannesburg, South Africa 1 2

Corresponding author: N R Rhoda (natasha.rhoda@uct.ac.za)

Although current levels of the neonatal mortality rate (NMR) are within reach of the Sustainable Development Goal (SDG) target of 12 per 1 000 live births, the absolute number of deaths is unacceptably high for a lower-middle-income country such as South Africa (SA). Neonatal mortality over the last decade has declined very slowly, and is not commensurate with the level of government investment in healthcare. The recent neonatal mortality rate of 21 per 1 000 live births reported by the SA Demographic Health Survey is of major concern. This paper reviews recent efforts to reduce the neonatal mortality rate, including support for the implementation of neonatal policies and plans, and strengthening programmes to deliver low-cost, high-impact interventions. We review recent estimates of the NMR and causes of neonatal deaths, and discuss how the mortality from preventable causes of death could be reduced. If SA is to meet the SDG target, special attention should be given to the availability of high-impact interventions, providing an adequate number of appropriately trained healthcare providers and a more active role played by ward-based community health workers and district clinical specialist teams. S Afr Med J 2018;108(3 Suppl 1):S9-S16. DOI:10.7196/SAMJ.2018.v108i3.12804

In 2005, about 4 million newborns died worldwide,[1] with more than 80% of the deaths occurring in lower-middle-income countries in sub-Saharan Africa and South Asia.[2] Since the Millennium Development Goals project ended in 2015, the annual reduction rate in mortality was noted to be slower during the neonatal period than in the post-neonatal period (1 - 59 months): 3.1% v. 4.7%, with a subsequent increase in the proportion of neonatal deaths accounting for 45% of the global under-5 mortality rate.[3] The United Nations Inter-agency group for child mortality reported that the global neonatal mortality rate (NMR) fell from 36 per 1 000 live births to 19 per 1 000 live births. This translates to a 47% reduction in neonatal deaths, to 2.7 million deaths per year, for the 2005 - 2015 period.[4] Many lower-middle-income countries have unreliable measurements and estimates of neonatal mortality. In-country vital statistics data that counts every birth and death, coupled with an accurate cause of death, is necessary for decision-making, planning and resource allocation. In most African countries, fewer than 25% of deaths are registered, with only 2% of African and Asian countries having complete data.[5] South Africa (SA) is an exception, and it is one of the few sub-Saharan African countries that has varied and rich sources of mortality data for children. These are shown in Table 1. This article reviews recent estimates of the NMR and causes of neonatal deaths in SA. It also reviews the strategies that are being used to reduce the NMR. In order to reach and surpass the sustainable development target for the NMR, we propose approaches to accelerate the rate of reduction of neonatal mortality resulting from preventable causes of death.

Methods

The NMRs for the period 2012 - 2015 were analysed using a number of data sources. Routine data were extracted from the District

Health Information System (DHIS), the routine information system used by the public health sector in SA. The DHIS does not report on neonatal deaths occurring in private sector healthcare facilities or those that occur outside of the healthcare system. These data were compared with the Rapid Mortality Surveillance (RMS) reports[6-9] produced by the Medical Research Council (MRC) annually. The RMS reports are based on a detailed analysis of the relevant death reports from the vital statistics of Statistics SA (StatsSA).[10,11] The DHIS data were also compared with two facility audit tools completed by hospital clinicians following mortality review meetings (the Perinatal Problem Identification Programme and the Child Problem Identification Programme (PPIP and Child PIP)). Further data on population dynamics were obtained from the SA Demographic Health Survey[12-14] (SADHS) carried out by the MRC and the National Department of Health (NDoH). The rates are reported as number of deaths per 1 000 live births. The two audit programmes generate the Saving Babies[15-23] and Saving Children[24-30] reports, respectively.

Results

Neonatal mortality rates and numbers

In combining the data from the DHIS and RMS reports, the resulting mortality rates in SA from 2002 to 2015 are shown in Fig. 1. This shows that the under-5 mortality rate rapidly declined from a peak of 80 per 1 000 at the height of the AIDS epidemic in 2003 - 2005, to 41 per 1 000 in 2012, with a slow decrease thereafter to 37 per 1 000 live births in 2015. The NMR has remained the same, at 11 - 12 per 1 000 live births between 2012 and 2015. In 2015, the NMR in SA accounted for 44% of the infant mortality rate (27 per 1 000 live births), and 32% of the under-5 mortality rate (37 per 1 000 live births), respectively.[6]

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Table 1. National mortality data sources and reports for neonates in SA Data source National Population Register. Department of Home Affairs, SA. Vital Registration. Statistics South Africa. District Health Information System. NDoH.

Advantages Allows for trends to be monitored.

Demographic Health Survey.

Provides population-based demographic information that contributes to the understanding of population dynamics. Records primary obstetric and final neonatal cause of death. Documents the avoidable causes of death. Records all births in weight categories. Validation checks inbuilt. Records all late neonatal deaths occurring in paediatric and emergency wards.

Perinatal Problem Identification Programme.

Child Problem Identification Programme.

Records all births and deaths in public and private. Records all births and deaths occurring in the public and (where recorded) private health sectors.

Limitations Only includes deaths of individuals with an SA birth certificate or identification document. Data has a 2-year lag phase. Collects data on both early and late deaths; although late deaths are not consistently reported. Only carried out every 5 years.

Report generated from data Rapid Mortality Surveillance.[6-9]

Perinatal Health Report by Statistics South Africa.[10,11] Routine DHIS data reports. Annual Report of NDoH. NDoH Annual Performance Plan (not available to public). South African Demographic Health Survey (SA-DHS).[12-14]

Collected on voluntary basis by enthusiast Saving Babies Report.[15-23] doctors. Represents only 77% of the DHIS data.

40% of hospitals do not submit data.

Saving Children Report.[24-30]

SA = South Africa; NDoH = National Department of Health; DHIS = District Health Information System.

IMR (RMS)

U5MR (RMS)

2 500

NMR (VR)

IMR (VR)

U5MR (VR)

2 000

IMR (direct)

NMR (direct)

NMR (direct DHIS)

1 500

2012 2013 2014 2015 2016

0 Ca p

Fr ee

te r

Ga ut en Kw g aZ ul uNa ta l Lim po po M pu m ala ng a No rth W es No t rth er n Ca pe W es te rn Ca pe

500

St at e

1 000

Ea s

Deaths per 1 000 live births

Deaths, n

NMR (DHIS)

Fig. 2. Provincial early neonatal deaths, 2012 - 2016.

30 20 10

00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15

Year

Fig. 1. SA mortality rates 2002 - 2015 (NMR = neonatal mortality rate; IMR = infant mortality rate; U5MR = under-5 mortality rate; RMS = Rapid Mortality Surveillance; VR = vital registration; DHIS = District Health Information System).

Data from the DHIS for 2016 show that the NMR was 12.6 per 1 000 live births, and the early and late neonatal mortality rates were 10.2 and 2.4 per 1 000 live births, respectively. The majority (81%) of neonatal deaths in 2016 occurred in the first week of life. The individual provincial DHIS data for 2012 - 2016 are shown in Fig. 2. There was an increase in the total number of early neonatal deaths in the Gauteng, KwaZulu-Natal (KZN) and Western Cape provinces when comparing deaths in 2013 with the subsequent years. Late neonatal deaths increased by 17%, from

S10

1 793 deaths (1.9 per 1 000) in 2012 to 2101 (2.4 per 1 000) in 2016. Recently, the SADHS reported a neonatal mortality rate of 21 per 1 000 live births for 2016, which is higher than the rate reported by the DHIS.[12]

Causes of death

The 2016 PPIP data show that the main causes of all neonatal deaths (birth weight ≥500 grams) are complications of prematurity (47.9%); intrapartum-related events, mainly intrauterine hypoxia (24.3%); and infections (including pneumonia, at 11.6%) (Fig. 3). A total of 60% of the premature deaths was accounted for by extreme low birth weight (ELBW) babies (birth weight <1 000 grams), who die primarily of extreme organ immaturity. If one excludes ELBW infants from the deaths when assessing the causes of death for the period under review, deaths due to intrapartum-related events supersede the complications of preterm births in babies weighing ≥1 000 grams. A comparison of cause-specific mortality over three 5-year cohorts, excluding ELBW infants, reflects this picture (Fig. 4). Following the Every Newborn Counts global campaign,[31] which

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

advocates the counting of every newborn death, Fig. 5 shows the calculated number of early neonatal deaths per cause of death per level of care from PPIP, with adjustment using the DHIS numbers (R Pattison, unpublished research, 2017). At district hospital level, where most babies are born, there is also the highest burden of premature deaths and intrapartum-related injuries. Infection and congenital abnormalities are the 3rd and 4th most common causes of early neonatal deaths, respectively. The overall picture of late neonatal deaths is uncertain, as very few of these deaths are captured by the current databases.

Deaths, %

100 80 60 40 20 0

6.7

12.3 13.8

8.9 11.6

29.1

47.9

34.9

24.3

≥1 000 g

≥500 g

87.8 2

<500 g

Weight categories Intrapartum-related events Infection Miscellaneous

Prematurity Congenital abnormalities

Deaths, %

Fig. 3. Cause-specific percentage of neonatal deaths in babies weighing <500 g, ≥1 000 g and ≥500 g (missing data = unknown cause). 100 90 80 70 60 50 40 30 20 10 0

9 10.5 12.2

8.9 9.6 12.4

9.9 11.8

38.9

35.9

32.3

30.2

29.3

2002 - 2006

2007 - 2011

2012 - 2016

13.1

Years Complications of preterm births Congenital abnormalities

Intrapartum-related events Miscellaneous

Neonatal infection

Fig. 4. Cause-specific percentage of neonatal deaths in babies >1 kg for three 5-year cohorts. Intrapartum-related injuries CHC 0.8% DH 14.2% RH 6.3% Tertiary 6.7%

Intrapartum-related events 28%

Prematurity CHC DH RH Tertiary

Congenital abnormalities CHC 0.2% DH 3.0% RH 2.0% Tertiary 3.6%

Congenital abnormality 8.8% Unknown cause of death 1.5%

1.2% 20.0% 13.0% 15.0%

Trauma 0.3% Prematurity 49.2%

Other 4.5% Infection 7.7%

Infection CHC DH RH Tertiary

0.2% 1.6% 2.3% 3.6%

Fig. 5. Causes of neonatal deaths per level of care (CHC = community health centre; DH = district hospital; RH = regional hospital).

S11

Avoidable factors identified in neonatal deaths

The top three causes of neonatal deaths have remained the same since 2012.[32] Table 2 shows the top 10 probably avoidable healthsystem-related factors for neonatal deaths, and includes those related to medical personnel, and administrator-related factors. A total of 3 105 neonatal deaths could probably have been avoided, of a total of 12 192 deaths.

Strategies and programmes to improve neonatal outcomes

In 2010, 5 years from the MDG deadline, SA was not on track to achieve MDG 4, the reduction of child mortality, of which neonatal mortality is a key component and the most difficult area to influence. The NDoH, with high ministerial political commitment, embarked on a range of initiatives to reduce neonatal mortality. These included the appointment of a neonatal-care improvement advisor, and the subsequent establishment of the National Neonatal Co-ordinating Committee (NNCC) in 2013, which provided a forum within the NDoH to co-ordinate and give oversight to improvements in newborn care in SA. Fig. 6 highlights the multitude of initiatives embarked upon since 2008 to improve care across the continuum. The NNCC’s main function was to ensure linkage and synergy in terms of the targets, goals, objectives and tasks of these national strategies, plans and campaigns. The National Perinatal Morbidity and Mortality Committee (NaPeMMCo) made its recommendations for the 2008 - 2010 triennium[33] on saving babies. They are summarised as HHAPINeSS, where HHAPI is an acronym used to summarise the key recommendations needed to improve newborn care and neonatal survival. The acronym is derived from the following: • Improve the Health system for mothers and babies • Improve the skills of Healthcare providers in maternal and neonatal care • Reduce deaths due to Asphyxia • Reduce deaths due to Prematurity • Reduce deaths due to Infection. The implementation of interventions for each of the recommendations formed the basis of the newborn survival strategy (NeSS). In 2014, NaPeMMCo and the Limpopo Initiative for Newborn Care finalised priority actions for newborn care for the country. Table 2. Top 10 probably avoidable healthcare-system-related factors and number of deaths, 2014 - 2015 (PPIP) (N=3 105) (unpublished ministerial report, NDoH) Modifiable factor Inadequate facilities/equipment in neonatal unit/ nursery Nosocomial infection Fetal distress not detected intrapartum; fetus monitored Delay in referring patient for secondary/tertiary treatment No accessible neonatal ICU bed with ventilator Neonatal care: management plan inadequate Neonatal care: inadequate monitoring Insufficient nurses on duty to manage the patient adequately Lack of transport – home to institution Lack of transport – institution to institution NDoH = National Department of Health; ICU = intensive-care unit.

March 2018, Vol. 108, (3 Suppl 1)

n 617 423 417 337 306 288 254 167 148 148

RESEARCH

Discussion

Neonatal numbers, mortality rates and causes

SA has several databases or sources that collect information on neonatal deaths, but most of them focus on deaths occurring within healthcare facilities. Outside of such facilities, the number of deaths is largely unknown, and could explain the discrepancy between the NMR of 12 per 1 000 reported from the DHIS data, (a facility-based database) and that reported by SADHS (21 per 1 000), a population-based survey. The latest Committee On Morbidity And Mortality In Children (CoMMiC) report[34] estimates that 45% of the under-5 deaths occur outside healthcare facilities. The child death review (CDR) process modelled and rolled out in high income-countries[35] addresses this gap, and SA’s 2013 CDR pilot study, at two mortuary sites in KZN and the Western Cape, have shown that 44.3% of the infant (<1 year old) pneumonia deaths were associated with prematurity, with a number of these deaths occurring within 72 hours after discharge from hospital.[36] Therefore quantifying, capturing, analysing

Every Newborn Counts launched National Perinatal Mortality and Morbidity Committee reviews neonatal deaths Implementation of the three streams of the Primary Healthcare Re-engineering plan MCHW Strategic Plan 2012-2016 President and Minister of Health sign NSDA and set MTSF targets for 2019 Every Newborn Action Plan launched and adopted by SA DCST posts established RMCH programme

Initiatives

The neonatal implementation plan aligned the eight high-impact Lives Saved Tool factors, the three targeted interventions to reduce neonatal mortality and the HHAPINeSS Strategy recommendations (internal document; NDoH, personal communication, 2014) (Fig. 7). All provinces aligned their provincial implementation plans for newborns with these priority actions. In efforts to achieve the above strategies, a number of training programmes have been implemented. These include Helping Babies Breathe (HBB), Management of Sick and Small Newborns (MSSN) and Essential Steps in Managing Obstetric Emergencies (ESMOE). HBB and MSSN are both training programmes to address the high number of deaths due to prematurity and intrapartumrelated events. Between August 2013 and December 2014, the NDoH capacitated provinces with master trainers for both programmes. The ESMOE Department for International Development-funded training, which has been scaled up in the last 2 years, also now includes HBB training. The PPIP data systems provided disaggregated data in weight categories, and one can calculate indicators to monitor the quality of care (see footnote in Table 3). The 9th Saving Babies report[23] showed, for the first time, a positive shift in the quality of care provided at district level (Tables 3 and 4). This trend was maintained for 2014 and 2015.

ESMOE pilot started

ESMOE scaled up

CARMMA signed HHAPI-NeSS Strategy developed by NaPeMMCo NCIA post established

NCIA reappointed

NNCC established HBB rollout National Neonatal Implementation Plan Deputy Director for Neonatal Care post established MSSN rollout CPAP phase 1

2008

2009

2010

2011

2012

2013

2014

2015

2016

CPAP phase 2

2017

2018

2019

2020

Year

Fig. 6. Initiatives to improve neonatal care. (MCHW = Maternal, Newborn, Child and Women’s Health and Nutrition in South Africa 2012 - 2016; NDSA = negotiated service delivery agreement; MTSF = medium-term strategic framework; SA = South Africa; DCST = district clinical specialist team; RMCH = Reducing Maternal and Child Mortality; ESMOE = Essential Steps in the Management of Obstetric Emergencies; CARMMA = Campaign for the Accelerated Reduction of Maternal Mortality in Africa; NCIA - neonatal care improvement advisor; NNCC = National Neonatal Co-ordinating Committee; HBB = Helping Babies Breathe; MSSN = managing small and sick newborns; CPAP = continuous positive airway pressure.) Recommendation

Interventions

Improve the Health system for mothers and babies

• Ensure 24-hour access to functioning emergency obstetric and neonatal care (both basic and comprehensive). Dedicated ambulances, maternal waiting homes, KMC sites in all hospitals, etc. • Ensure accessible and appropriate contraceptive services for all women, which are integrated into all levels of healthcare and are available on site for post-miscarriage and postpartum women.

Improve the skills of Healthcare providers in maternal and neonatal care

• Train all healthcare workers involved in maternity and neonatal care on the ESMOE-EOST programme (including HBB) and in MSSN infants. • Train all healthcare workers who deal with pregnant women in HIV advice, counselling, testing and support; and initiation and monitoring of HAART.

Reduce deaths due to Asphyxia

• Ensure that labour is monitored appropriately by a skilled birth attendant. • Ensure that all birth attendants are skilled at a minimum level in neonatal bag and mask preparation. • Ensure that the partogram is used to monitor labour and that the fetus and mother are monitored according to the prescribed norms, ensuring proper data interpretation.

Reduce deaths due to Prematurity

• Ensure that corticosteroids are given to every woman in preterm labour. • Ensure that antibiotics are given with preterm rupture of membranes. • Ensure that the appropriate hospitals are skilled in the use of nasal CPAP. • Ensure that all mothers of immature infants have easy access to KMC.

Reduce deaths due to Infection

• Promote breastfeeding (especially exclusive breastfeeding). • Ensure clean cord care. • Ensure strict adherence to basic hygiene in labour wards and nurseries. • Ensure that presumptive antibiotic therapy for at-risk newborns is available. • Ensure case management of neonatal sepsis, meningitis and pneumonia.

Fig. 7. The HHAPI-NeSS strategy. (KMC = kangaroo mother care; ESMOE-EOST = essential steps in managing obstetric emergencies and emergency obstetric simulation training; HBB = Helping Babies Breathe; MSSN = managing small and sick newborns; HAART = highly active antiretroviral treatment; CPAP = continuous positive airway pressure.)

S12

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Table 3. Trends in the quality-of-care indicators among newborns weighing >2.5 kg per level of care in SA, 2012 - 2015 (PPIP)* Year

2011 2012 2013 2014 2015

Community health centres PNMR ENNDR >2.5 kg >2.5 kg 2.5 0.5 3 0.8 2.7 0.5 2.1 0.5 2.3 0.6

District hospitals PNMR ENNDR >2.5 kg >2.5 kg 11.2 4.1 11.1 4.1 10.7 4.0 10.1 3.8 9.3 3.4

Regional hospitals PNMR ENNDR >2.5 kg >2.5 kg 9.0 2.6 9.9 2.9 10.4 3.2 8.8 2.8 9.8 3.4

Tertiary hospitals PNMR ENNDR >2.5 kg >2.5 kg 16.6 5.6 17.6 6.2 15 5.0 14.8 5.3 16.2 6.1

National central hospitals PNMR ENNDR >2.5 kg >2.5 kg 15.8 5.5 18.7 7.8 11.6 3.3 11.1 3.3 12.9 2.8

SA = South Africa; PPIP = Perinatal Problem Identification Programme; PNMR = perinatal mortality rate; PNMR >2.5 kg = total number of perinatal deaths weighing >2.5 kg (stillbirths and early neonatal deaths)/total number of births ×1 000; ENNDR = early neonatal death rate; ENNDR >2.5 kg = total number of neonatal deaths weighing >2.5 kg from day 0 - 6 completed days/ total number of live births × 1 000. *Quality of intrapartum care and early neonatal resuscitation calculated by the PNMR >2.5 kg and the ENNDR >2.5 kg.

Table 4. Trends in the early neonatal death rates among newborns weighing 1 - 1.49 kg per level of care in SA public health facilities, 2012 - 2015 (PPIP)* Year 2011 2012 2013 2014 2015

Community health centres 49.9 32.4 35.0 58.9 47.1

District hospitals 220.3 228.0 211.5 221.2 213.0

Regional hospitals 91.1 113.3 120.8 109.0 124.2

Tertiary hospitals 138.9 149.0 112.0 135.3 180.4

National central hospitals 122.4 100.4 85.4 67.5 40.9

SA = South Africa; PPIP = Perinatal Problem Identification Programme. Quality of neonatal care calculated by the early neonatal death rate of newborns in the 1 - 1.49 kg weight category

Table 5. Lives saved using the Lives Saved Tool in SA The LiST newborn interventions Antenatal corticosteroids for preterm labour Labour and delivery management Prevention of mother-to-child transmission of HIV Oral rehydrate solution Handwashing with soap Case management of severe neonatal infection Water connection in the home Antiretroviral treatment Pneumococcal vaccine Therapeutic feeding for severe wasting Treatment of injuries

Lives saved, % 12 10 9 9 7 7 5 4 4 4 4

SA = South Africa; LiST = Lives Saved Tool.

and reviewing these out-of-facility deaths is vital for an overall reduction in the under-5 mortality rate, and especially in the NMR. Differences in the NMR between the two sources can also be attributed to the definition of the neonatal period: the SADHS uses a postnatal age of 31 days, while the DHIS uses 28 days. Secondly, SADHS data come from questionnaires administered to mothers in the community; therefore, it is possible that the recall of these mothers includes stillbirths. This explanation is informed by a recent Malawian study[37] that found that in contrast, 20% of neonatal deaths were wrongfully classified as stillbirths after a full birth history by the mother. StatsSA and the MRC will review the SADHS data to confirm whether the SA NMR is double the current DHIS estimates. The NMR in SA is much higher than that reported in 2013 from developed regions (3.38 per 1 000), countries from Eastern Asia (7.69

S13

per 1 000), and Latin America (9.21 per 1 000), but lower than the average of countries (including SA itself) from sub-Saharan Africa (31.1 per 1000), and similar to that reported from North Africa. Within the sub-Saharan countries, only 2 island countries, the Seychelles and Mauritius, have lower NMRs than SA, at 8.82 and 8.90 per 1 000 live births, respectively.[38] Unless efforts to reduce NMRs are intensified, there is a high risk of not meeting the global Sustainable Development Goal (SDG) target for the NMR, and the issue of newborn survival remains an unfinished item on the agenda. The database that reports on the causes of deaths is PPIP, but of major concern is the fact that only four provinces (Mpumalanga, the Western Cape, Limpopo and Free State) have complete PPIP data in relation to the DHIS. Unfortunately, the most densely populated provinces, KZN, Gauteng and the Eastern Cape, have gaps of >25% between PPIP and DHIS data, which makes planning and analysis based on causes of death problematic. The Eastern Cape data is of concern, as it shows a 40% reduction in neonatal deaths, which is unlikely, and preliminary investigations have shown that this reflects poor data collection and not a true reduction in neonatal deaths. After the provincial closure of neonatal units in the early 2000s, the province has now embarked upon the process of reopening them, since 2013. Prematurity is the driver of neonatal deaths in the country, with those born weighing <1 000 grams (ELBW) contributing significantly to the mortality figures. Therefore, at facility level, the cause-specific mortality rate-per-weight categories offer facilities an opportunity to decide on where best to focus their interventions, and also provide hospitals with the ability to benchmark within a level of care (e.g. district hospitals). The reduction in ELBW deaths lies primarily in the upstream factors of provision of antenatal steroids, and regular antenatal visits, and not necessarily in the building of more neonatal intensive-care units, which require extra qualified staff and

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

expensive equipment that requires regular maintenance. Studies in lower-middle-income countries have shown that providing these surviving infants with breastfeeding, kangaroo mother care and basic continuous positive airway pressure (CPAP) are cost-effective ways of reducing mortality. SA is one of the countries with the lowest exclusive breastfeeding rates in the world, and national campaigns are now underway to improve upon this rate. Early initiation of breastfeeding has been shown in random-effects analyses to lower the risks of all-cause neonatal mortality among all live births (relative risk (RR) 0.56; 95% confidence interval (CI) 0.40 - 0.79) and among low birth weight babies (RR 0.58; 95% CI 0.43 - 0.78), and of infectionrelated neonatal mortality (RR 0.55; 95% CI 0.36 - 0.84).[39) The reported increase in numbers of late neonatal deaths could be the direct result of better reporting in paediatric and emergency wards, since more facilities are now introducing Child PIP at facility level. The NDoH has recently embarked on an initiative to align the death reporting numbers between the DHIS (where the numbers of deaths are reported), PPIP (where the causes of death, and other pertinent variables such as weight and gestational age, are recorded) and Child PIP. This will help to ensure that in-facility neonatal mortality rates are aligned and as accurate as possible.

Strategies to reduce neonatal deaths

At the heart of all the policies, plans and programmes developed by the NDoH and partners has been a commitment to improve the quality of care for mothers and babies, and thereby reduce preventable neonatal deaths. NaPeMMCo highlighted the need to look at and track quality-of-care indications, as well as the impact indicators of neonatal mortality rates. The improvement in quality of care seen at district level is therefore encouraging, as most of the efforts towards improved neonatal care have targeted district hospitals, community health centres and clinics. These efforts may also reflect the early result of input from the district clinical specialist team (DCSTs) within the districts, and the rollout of training programmes such as HBB across the country. Three interventions, HBB, MSSN and CPAP, were included in the 15 interventions developed by PRICELESS SA (Priority Cost Effective Lessons for Systems Strengthening South Africa) that, when taken to scale, would reduce neonatal mortality sufficiently to achieve SDG 3 by 2030. However, the costs estimated by PRICELESS SA excluded infrastructure development. For example, CPAP as a non-invasive mode of ventilation has been successful worldwide in the management of respiratory distress syndrome in preterm infants. The early neonatal mortality rate is highest in the <1.5 kg weight category of preterm babies. However, the provision of the CPAP model critically requires the availability of medical air. The PPIP data have shown a 20% overall reduction in absolute deaths in babies >500 g coded as hyaline membrane disease over the last 5 years. While difficult to ascribe this reduction only to CPAP, a similar trend was seen when CPAP was provided to rural district hospitals in the Western Cape and Limpopo. HBB and MSSN have been rolled out in all provinces, but the monitoring and evaluation of these training programmes has not been robust, and thus the national coverage is unknown. In contrast, ESMOE training was fully funded, and the pilot sites have shown a 35% reduction in their NMRs. As ESMOE goes to scale countrywide, a decline in intrapartum-related deaths is expected. This highlights the need to ensure adequate and ongoing funding not only for the implementation of programmes, but importantly, also for monitoring, evaluation and scale-up. The latest PPIP data show a marked reduction in the number of deaths due to a lack of syphilis screening, inadequate

S14

resuscitation and insufficiently trained personnel, which might be ascribed to the rollout of the targeted interventions. Leadership on the issue of neonatal mortality has proved to be lacking at all levels,[40] but is particularly essential at provincial level to effect change. The three provinces with the best quality of provincially co-ordinated newborn care planning and training programmes, Limpopo, KZN and the Western Cape, have created and funded posts to which provincial paediatricians have been appointed. Strong provincial leadership ensures that accountability mechanisms are in place, and restrains unnecessary expenditure. This attracts funding for neonatal care, as both KZN and Limpopo have secured grants for their rollout of provincial neonatal training programmes. Provincial coverage of targeted interventions such as CPAP has been successful in the Western Cape and Limpopo. Care at regional and tertiary levels has shown no improvement. At regional hospitals, this is probably multifactorial, and may reflect the low health levels of the patients received, and not necessarily the quality of care provided. A key aspect of the neonatal implementation plan was to support the development of tailored, evidence-based plans at district level. DCSTs were the drivers of implementation and overall clinical governance, and their leadership role in neonatal care is discussed in the DCST article in this supplement.[41]

Addressing the challenges

The suggestions below are intended to address some of the challenges in addressing neonatal mortality in SA. Scale up evidence-based medical interventions to 80% coverage Chola et al.â&#x20AC;&#x2122;s[42] modelling has shown that the additional cost of scaling up key maternal, neonatal and child interventions will amount to less than 1% of the annual national health budget, which seems affordable in light of a per capita health expenditure of about USD645.00. If these interventions (e.g. improved case management of severe neonatal infection) achieve 95% coverage with high quality of care, then the modelling suggests that the NMR could be reduced to 6 per 1 000 live births.[42] Increase usage of antenatal steroid in preterm labour In SA, 22.5% of perinatal deaths are due to spontaneous preterm labour. Despite a 96% facility birth rate, there is only 25% antenatal steroid (ANS) coverage. This is similar to that of Brazil prior to 2014, when it embarked on a national ANS scale-up campaign. A similar effort will enable SA to fully implement a newly developed national guideline that enables midwives to prescribe the first dose of ANS on confirmation of preterm labour. Provision of ANS will make the greatest difference in saving the number of newborn lives (12%) if coverage increases from the current estimated baseline of 20 - 40%.[43) Provide a postnatal care package (including the supportive role of community health workers) in neonatal care The provision of postnatal care is essential for neonate survival. However, this has not been implemented at scale, owing to capacity constraints. Most of the investments in neonatal health have been at facility level, and the link between hospital and home has not yet successfully been bridged by the primary healthcare wardbased outreach teams (WBOTS). Central to the WBOTS are the community health workers (CHWs), and to date only 45% of these teams are completed. There is growing evidence that sub-Saharan African countries that have done well at reducing child mortality have invested in CHWs and empowered them with therapeutic roles within their health systems.[44] This curative role is supported by

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

World Health Organization guidelines, a Cochrane review[45] and the integrated community case management strategy.[46] The call is out for SA to increase the number of CHWs, and to include treatment of the common causes of child deaths (pneumonia and diarrhoea) within their scope of practice.

Conclusions

Reducing neonatal mortality in lower-middle-income countries is complex, and especially difficult if the preventable causes of death have decreased. A Brazilian cohort study[47] showed that the Brazil NMR did not change for two decades despite improvements in maternal and neonatal care and maternal health-seeking behaviour. They attributed the static NMR to an epidemiological and demographic transition the country was going through, with medicalisation of pregnancy and delivery, rapidly increasing the preterm births and deaths, which then offset the gains made in the better survival rate of term infants. SA has a double burden of disease driving neonatal mortality; term babies are dying owing to intrapartum-related events, and preterm deaths are linked to related complications. Consistently, however, mortality reviews conducted at facility level have identified that >50% of these neonatal deaths are probably avoidable. As most births occur in facilities, the focus should remain on the quality of perinatal and intrapartum care provided by healthcare workers, and the ability of facilities to provide essential emergency and basic neonatal care. SA cannot afford to rapidly increase the capacity of its healthcare workers, but can redistribute and train the existing workforce to deliver better-quality care. This will require that accountability mechanisms are enforced by all managers at all levels to achieve a faster annual rate of reduction in NMR. We need to continue to improve upon existing infrastructure, ensure the sustainability of the health system gains and intensify efforts to achieve 95% coverage of the 15 LiST interventions currently being costed by NDoH (Table 5). This should help to ensure that the appropriate resources and evidence-based interventions in improving neonatal care are made available in all public healthcare facilities. Underpinning all these efforts is the improvement in the quality of the data available, with attention to modifiable factors in out-of-facility deaths. Plans are afoot by the NDoH to synchronise and align PPIP and Child PIP with the DHIS. This will reduce data duplication, and hopefully allow for easier institutionalisation of these programmes. The investments made by the NDoH in system strengthening, by providing leadership in policies and placing an emphasis on doing the basics right, should enable SA to achieve the SDG for the NMR by 2030. Acknowledgements. We would like to acknowledge Dr Debbie Bradshaw (SA MRC Burden of Disease Unit) and Prof. Bob Pattinson (SA MRC Maternal and Infant Health Care Strategies Research Unit). Author contributions. NRR, SV, GSG and SK conceptualised the paper and co-ordinated the writing process. NRR, GSG and SV were responsible for the analysis and figures. SK and PB reviewed the manuscript. All authors contributed to article drafts, and approved the final manuscript. Funding. None Conflicts of interest. None 1. Lawn JE, Cousens S, Zupan J. 4 million neonatal deaths: When? Where? Why? Lancet 2005;365(9462):891900. https://doi.org/10.1016/S0140-6736(05)71048-5 2. Knippenberg R, Lawn JE, Darmstadt GL, et al. Systematic scaling up of neonatal care in countries. Lancet 2005;365(9464):1087-1098. https://www.ncbi.nlm.nih.gov/pubmed/15781104 (accessed 9 February 2018). 3. Liu L, Oza S, Hogan D, et al. Global, regional, and national causes of under-5 mortality in 2000 -15: An updated systematic analysis with implications for the Sustainable Development Goals. Lancet 2016;388(10063):30273035. https://doi.org/10.1016/S0140-6736(16)31593-8

S15

4. Lawn J, Blencoew H, Kinney MV, Bianchi F, Graham WJ. Evidence to inform the future for maternal and newborn health. Best Pract Res Clin Obstet Gynaecol 2016;36:169-183. https://doi.org/10.1016/j. bpobgyn.2016.07.004 5. Joubert J, Rao C, Bradshaw D, Dorrington RE, Vos T, Lopez A. Characteristics, availability and uses of vital registration and other mortality data sources in post-democracy South Africa. Glob Health Action 2012;5(1):19263. https://doi.org/10.3402/gha.v5i0.19263 6. Dorrington RE, Bradshaw D, Laubscher R, Nannan N. Rapid Mortality Surveillance Report 2015. Cape Town: South African Medical Research Council, 2016. http://www.mrc.ac.za/bod/ RapidMortalitySurveillanceReport2015.pdf (accessed 9 February 2018). 7. Dorrington RE, Bradshaw D, Laubscher R, Nannan N. Rapid Mortality Surveillance Report 2014. Cape Town: South African Medical Research Council, 2015. http://www.mrc.ac.za/bod/ RapidMortalitySurveillanceReport2014.pdf (accessed 9 February 2018). 8. Dorrington RE, Bradshaw D, Laubscher R, Nannan N. Rapid Mortality Surveillance Report 2013. Cape Town: South African Medical Research Council, 2014. http://www.mrc.ac.za/bod/ RapidMortalitySurveillanceReport2013.pdf (accessed 9 February 2018). 9. Dorrington RE, Bradshaw D, Laubscher R. Rapid Mortality Surveillance Report 2012. Cape Town: South African Medical Research Council, 2014. http://www.mrc.ac.za/bod/ RapidMortalitySurveillanceReport2012.pdf (accessed 9 February 2018). 10. Statistics South Africa. Perinatal deaths in South Africa, 2014. Pretoria: StatsSA, 2014. http://www. statssa.gov.za/publications/P03094/P030942014.pdf (accessed 27 October 2016). 11. Statistics South Africa Perinatal deaths in South Africa, 2011 - 2013. Pretoria: StatsSA, 2015. http:// www.statssa.gov.za/publications/P03094/P030942013.pdf (accessed 9 February 2018). 12. National Department of Health, Statistics South Africa, South African Medical Research Council, ICF [International Classification of Functioning, Disability and Health]. South African Demographic and Health Survey 2016: Key Indicators. Pretoria, South Africa and Rockville, Maryland, USA: NDoH, StatsSA, SAMRC, ICF, 2017. http://www.statssa.gov.za/publications/Report%2003-00-09/Report%20 03-00-092016.pdf (accessed 9 February 2018). 13. Department of Health, Medical Research Council, ORC Macro, South Africa. Demographic and Health Survey 2003. Pretoria: Department of Health, 2007. https://dhsprogram.com/publications/ publication-fr206-dhs-final-reports.cfm (accessed 9 February 2018). 14. Department of Health, Medical Research Council. South Africa Demographic and Health Survey 1998. Pretoria: DoH, 2001. http://www.mrc.ac.za/bod/SADHS1998FullReport.pdf (accessed 9 February 2018). 15. Pattinson R, PPIP Group. Saving babies 2000: First report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-babies-2000.pdf (accessed 9 February 2018). 16. Pattinson R, PPIP Group. Saving babies 2001: Second report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-babies-2001.pdf (accessed 9 February 2018). 17. Pattinson R, PPIP Group. Saving babies 2002: Third report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-babies-2002.pdf (accessed 9 February 2018). 18. Pattinson R, PPIP Group. Saving babies 2003: Fourth report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-babies-2003.pdf (accessed 9 February 2018). 19. Pattinson R, PPIP Group. Saving babies 2003 - 2005: Fifth report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-Babies-2003-5.pdf (accessed 9 February 2018). 20. Pattinson R, PPIP Group. Saving babies 2006 - 2007: Sixth report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-babies-2006-7.pdf (accessed 9 February 2018). 21. Pattinson R, PPIP Group. Saving babies 2008 - 2009: Seventh report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-Babies-2008-9.pdf (accessed 9 February 2018). 22. Pattinson R, PPIP Group. Saving babies 2010 - 2011: Eighth report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/Saving-Babies-2010-2011. pdf (accessed 9 February 2018). 23. Pattinson P, Rhoda N, PPIP Group. Saving babies 2012-2013: Ninth report on perinatal care in South Africa. Pretoria: Tshepesa Press, 2014. https://www.ppip.co.za/wp-content/uploads/SavingBabies-2012-2013.pdf (accessed 9 February 2018). 24. A Krug, RC Pattinson, editors. Saving Children 2004: A survey of child healthcare in South Africa. Pretoria: MRC Research Unit for Maternal and Infant Health Care Strategies, 2006. http://www.childpip. org.za/images/stories/documents/report_saving_children_2004.pdf (accessed 9 February 2018). 25. ME Patrick, CR Stephen, editors. Saving Children 2005: A survey of child healthcare in South Africa, Pretoria, 2007. http://www.childpip.org.za/documents/report_saving_children_2005.pdf (accessed 9 February 2018). 26. CR Stephen, ME Patrick, editors. Saving Children 2006: A survey of child healthcare in South Africa. Pretoria: University of Pretoria, MRC, CDC, 2008 http://www.childpip.org.za/documents/report_ saving_children_2006.pdf (accessed 9 February 2018). 27. CR Stephen, MC Mulaudzi, S Kauchali, ME Patrick eds. Saving Children 2005 - 2007: A fourth survey of child healthcare in South Africa. Pretoria: University of Pretoria, MRC, CDC, 2009 http://www. childpip.org.za/images/stories/documents/Saving_Children_2005_2007.pdf (accessed 9 February 2018). 28. Stephen CR, Bamford LJ, Patrick ME, Wittenberg DF, editors. Saving Children 2009: Five Years of Data. A sixth survey of child healthcare in South Africa. Pretoria: Tshepesa Press, MRC, CDC, 2011. http://www.childpip.org.za/documents/saving_children_2009.pdf (accessed 9 February 2018). 29. Stephen CR, Bamford LJ, editors. Saving Children 2010 - 2011: A seventh survey of child health care in South Africa. Pretoria: Tshepesa Press, MRC, CDC, 2013 http://www.childpip.org.za/images/stories/ saving_children_2010-2011.pdf (accessed 9 February 2018). 30. CR Stephen, editor. Saving Children 2012-2013: An eighth survey of child healthcare in South Africa. Pretoria: Tshepesa Press, 2016 http://www.childpip.org.za/images/stories/documents/saving_ children_2012-2013.pdf (accessed 9 February 2018). 31. Moxon SG, Ruysen H, Kerber KJ, et al. Count every newborn; a measurement improvement roadmap for coverage data. BMC Pregnancy Childbirth 2015;15(Suppl 2):S8 https://doi.org/10.1186/1471-239315-S2-S8 32. Velaphi S, Rhoda N. Reducing neonatal deaths in South Africa â&#x20AC;&#x201C; are we there yet, and what can be done? S Afr J Child Health 2012;6(3):67-71. https://doi.org/10.7196/sajch.493 33. National Perinatal Morbidity and Mortality Committee. National Perinatal Mortality and Morbidity Committe (NaPeMMCo) Triennial Report (2008 - 2010). Pretoria: National Department of Health, 2011. 34. National Department of Health. 2nd Triennial Report of the Committee on Morbidity and Mortality in Children under 5 Years (CoMMiC): 2014 (Abridged version). Pretoria: NDoH, 2014. http://www. kznhealth.gov.za/mcwh/2nd-CoMMiC-Triennial-Report-Abridged.pdf (accessed 9 February 2018). 35. Fraser J, Sidebotham P, Frederick J, Covington T, Mitchell EA. Learning from child death review in the USA, England, Australia, and New Zealand. Lancet 2014;384(9946):894-903. https://doi.org/10.1016/S01406736(13)61089-2 36. Mathews S, Martin LJ, Coetzee D, et al . The South African child death review pilot: A multiagency approach to strengthen healthcare and protection for children. S Afr Med J 2016;106(9):895-899. https://doi.org/10.7196/ SAMJ.2016.v106i9.11234

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

37. Liu L, Kalter D, Chu Y, et al. Understanding misclassification between neonatal deaths and stillbirths: Empirical evidence from Malawi. PLOS ONE 2016;11(12):e0168743. https://doi.org/10.1371/journal.pone.0168743 38. Liu L, Oza S, Hogan D, et al. Global, regional, and national causes of child mortality in 2000 - 13, with projections to inform post-2015 priorities: An updated systematic analysis. Lancet 2015;385(9966):430-440. https://doi.org/10.1016/S0140-6736(14)61698-6 39. Debes AK, Kohli A, Walker N, et al. Time to initiation of breastfeeding and neonatal mortality and morbidity. Bio Med Central Public Health 2013;13(Suppl 3):S19. https://doi.org/10.1186/1471-2458-13-S3-S19 40. National Department of Health, South Africa. National Report for the Mid-Term Review of the Strategic Plan for Maternal, Newborn, Child and Womenâ&#x20AC;&#x2122;s Health (MNCWH) and Nutrition in South Africa November 2015. Pretoria: NDoH, 2015. http://www.emtct-thelastmile.co.za/wp-content/uploads/2016/07/Mid-Termreview-report-MNCWH.pdf 41. U Feucht, C Marshall, S Kauchali, et al. Innovations in the clinical care of mothers and children in South Africa: The contribution of district clinical specialist teams. S Afr Med J 2018;108(Suppl 1):S38-S43. https:// doi.org/10.7196/SAMJ.2018.v108i3.12808 42. Chola L, Pillay Y, Barron P, Tugendhaft A, Kerber K, Hofman K. Cost and impact of scaling up interventions to save lives of mothers and children: Taking South Africa closer to MDGs 4 and 5. Glob Health Action 2015;8(1):27265. https://doi.org/10.3402/gha.v8.27265 43. Lloyd LG, de Witt TW. Neonatal mortality in South Africa: How are we doing and can we do better? S Afr Med J 2013;103(8):518-519. https://doi.org/10.7196/SAMJ.7200

S16

44. Doherty T, Kroon M, Rhoda N, Sanders D. Ending preventable child deaths in South Africa: What role can ward-based outreach teams play? S Afr Med J 2016;106(7):672-674. https://doi.org/10.7196/ SAMJ.2016.v106i7.10790 45. Lassi ZS, Haider BA, Bhutta ZA. Community-based intervention packages for reducing maternal and neonatal morbidity and mortality and improving neonatal outcomes. Cochrane Database Syst Rev 2010;10(11):CH007754. https://doi.org/10.1002/14651858.CD007754.pub3 46. Rasanathan K, MuĂąiz M, Bakshi S, et al. Community case management of childhood illness in subSaharan Africa - findings from a cross-sectional survey on policy and implementation. J Glob Health 2014;4(2):020401. https://doi.org/10.7189/jogh.04.020401 47. Barros FC, Victora CG, Barros AJD, et al. The challenge of reducing neonatal mortality in middleincome countries: Findings from three Brazilian birth cohorts in 1982, 1993, and 2004. Lancet 2005;365(9462):847-854. https://doi.org/10.1016/S0140-6736(05)71042-4

Accepted 12 September 2017.

March 2018, Vol. 108, (3 Suppl 1)

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

RESEARCH

Closing the gaps to eliminate mother-to-child transmission of HIV (MTCT) in South Africa: Understanding MTCT case rates, factors that hinder the monitoring and attainment of targets, and potential game changers A Goga,1, 2 PhD; W Chirinda,1 PhD; N K Ngandu,1 PhD; K Ngoma,3 MSc; S Bhardwaj,3 MD; U Feucht,2,4 PhD; N Davies,5 MPH; M Ntloana,6 Dip Nursing; O Mhlongo,7 BCur; T Silere-Maqetseba,6 BCur; F Moyo,8,9 MSc; G Sherman,8, 9 PhD Health Systems Research Unit, South African Medical Research Council, Cape Town, South Africa Department of Paediatrics and Child Health, Faculy of Health Sciences, University of Pretoria, Pretoria, South Africa 3 Health section, UNICEF South Africa, Pretoria, South Africa 4 Tshwane District Health Services, Gauteng Department of Health, Pretoria, South Africa 5 Wits Reproductive Health and HIV Institute, University of Witwatersrand, Johannesburg, South Africa 6 National Department of Health, Pretoria, South Africa 7 KwaZulu-Natal Department of Heath, Pietermaritzburg, South Africa 8 Department of Paediatrics and Child Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa 9 Centre for HIV and STI, National Institute for Communicable Diseases, Johannesburg, South Africa 1 2

Corresponding author: A Goga (ameena.goga@mrc.ac.za) Background. Ninety percent of the world’s HIV-positive pregnant women live in 22 countries. These 22 countries, including South Africa (SA) have prioritised the elimination of mother-to-child transmission of HIV (EMTCT). Since 2016 all 22 countries recommend lifelong antiretroviral treatment for all HIV-positive pregnant and lactating women. To measure SA national, provincial and district-level progress towards attaining EMTCT, we analysed the number of in utero (IU) paedatric HIV infections per 100 000 live births (IU case rate), and synthesised factors hindering the monitoring of EMTCT progress and attainment from the viewpoint of provincial and district-level healthcare managers and implementers. We highlight potential innovations to strengthen health systems and improve EMTCT programme delivery. Methods. We reviewed national-, provincial- and district-level birth HIV testing data from routine National Health Laboratory Services (NHLS) records between April 2016 and March 2017. To obtain a qualitative perspective from healthcare managers and implementers, we synthesised information from the nine 2016 provincial-level EMTCT stock-taking workshops. These workshops involve key provincial and district-level staff, mentors and supporting partners. Lastly, we highlight potential innovations presented at these workshops to overcome operational challenges. Results. The national IU mother-to-child transmission (MTCT) rate was 0.9%, which translated to an IU case rate of 245 HIV-positive neonates per 100 000 live births. Provincial IU percent MTCT risk ranged from 0.6% to 1.3%, with IU case rates ranging between 168 and 325 cases per 100 000 live births. District-level IU percent MTCT risk ranged from 0.4% to 1.9%. Potential game changers include: pre-conception counselling to optimise maternal-partner health, weekly dissemination of HIV polymerase chain reaction (PCR) and viral load reports from the NHLS to specific individuals who trace mothers and infants needing care, use of ward-based outreach teams and community caregivers to trace HIV-infected mothers and their infants to link them into care, inclusion of a unique identifier in patient-held infant Road to Health booklets to facilitate infant tracing and continuous quality improvement (CQI) processes within facilities and districts and implementation of an HIV-positive-baby tool to understand the characteristics and risks of HIV-positive infants. On an ecological level, provinces and districts using community-based approaches and CQI methodology seemed to have lower MTCT and IU case rates. Conclusions. More quantitative analyses are needed to understand what proportion of the success can be attributed to community-based and CQI approaches and the impact of the potential game changers on progress towards EMTCT. S Afr Med J 2018;108(3 Suppl 1):S17-S24. DOI:10.7196/SAMJ.2018.v108i3.12817

Ninety percent of the world’s HIV-positive pregnant women live in 22 countries. These countries, including South Africa (SA), have prioritised the elimination of mother-to-child transmission of HIV (EMTCT), measured using three coverage and two impact indicators (Fig. 1).[1] The two impact indicators are the percentage mother-to-child HIV transmission (MTCT) among HIV-positive mothers (MTCT risk) and the number of new paediatric HIV infections per 100 000 live births (MTCT case rate). The global target is <5% final MTCT in breastfeeding countries and ≤50 new

S17

paediatric HIV infections per 100 000 live births. Consequently, to achieve EMTCT, all 22 countries currently recommend lifelong antiretroviral treatment for all HIV-positive pregnant and lactating women. This is known as PMTCT Option B+. Between the 1980s and 2015 the discovery of more effective biomedical interventions to prevent mother-to-child transmission of HIV (PMTCT) accelerated the improvement of global and national PMTCT policies; such rapid advancements in policy required flexible, responsive health systems, services and staff to assure implementation.[1-6] For example, in

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

changes yielded a concomitant decrease in the national risk of early (6 weeks postpartum) MTCT, from 3.5% in 2010 to 1.1% in 2015-16 (Fig. 2).[13-16] Additionally, cross-sectional surveillance data suggest that ART initiation prior to conception or during the first trimester of pregnancy could reduce early MTCT to <1.2%.[14] Despite these national-level successes, interprovincial and inter-district differences in MTCT and maternal ART uptake exist: the provincial in utero percent MTCT risk ranged from 1.4% to 5.9% in 2010 and 0.7 to 1.7% in 2015/16; maternal ART uptake ranged from 78% to 98% in 2015/16.[14,15] The district-level early MTCT rate ranged from 2.2% to 8.1% in 2010 and 0% to 3.4% in 2015/16 (Fig. 2),[15,17,18] while maternal ART uptake varied from 46% to 100% in 2015/16.[15] We analysed the number of IU paedatric HIV infections per 100 000 live births (IU case rate) as an initial step to monitor

MTCT rate, as defined by WHO: % HIV-exposed infants who acquire HIV infection from their mothers Case rate: number of new HIV infections in children per 100 000 live births In utero case rate: number of new HIV infections in children measured at <7 days of age, per 100 000 live births

Fig. 1. Key concepts in the elimination of mother-to-child transmission of HIV.

2001 the SA national PMTCT programme recommended single dose nevirapine (NVP) for HIV-positive pregnant women at the onset of labour and for HIV-exposed infants within 72 hours of delivery. These recommendations were made alongside modified obstetric practices and avoidance or reduced duration of breastfeeding.[7] Between 2008 to 2013, national PMTCT improved dramatically to include the initiation of maternal antiretroviral drugs (ARVs) earlier in gestation or at higher CD4 cell counts.[8-10] By January 2015, the national PMTCT policy recommended lifelong triple antiretroviral therapy (ART) for all pregnant and lactating HIV-positive women (PMTCT Option B+ policy), with infant HIV testing at birth to identify in utero infection early, expedite ART initiation and improve infant outcomes, which are poor if ART initiation is delayed.[11,12] In 2016, SA also launched the Last Mile Plan for EMTCT. This multipronged plan acknowledges the importance of delivery systems for EMTCT and prioritises five pillars, namely: (i) leadership, governance and coordination; (ii) scaling up PMTCT coverage; (iii) integrating PMTCT interventions into routine maternal and child healthcare (MCH) and primary healthcare (PHC); (iv) monitoring and evaluation; and (v) increased community awareness and involvement.[13] The PMTCT policy and

Last Mile Plan require implementation at all levels of the healthcare system, during the pre-conception, antenatal and postnatal care periods, and involve mother, father, baby and in some instances the extended family, illustrating the complexities of PMTCT programme delivery. In a national-level review published in 2017, we demonstrate that these policy

Fig. 2. Early mother-to-child transmission measured at national, provincial and district levels in South Africa.

Table 1. Definitions applied to routine NHLS data Definition Birth testing coverage: Total number of HIV PCR tests performed in HIV-exposed infants within the first 7 days

Numerator Total number of HIV PCR tests performed in neonates aged <7 days old

IU % MTCT risk: % MTCT within the first 7 days of life IU case rates: Number of new HIV infections among- infants <7 days old expressed as a standardised rate per 100 000 live births

Neonates within the first 7 days of life with a positive HIV PCR test Number of HIV PCR-positive neonates per 100 000 live births

Denominator This is an estimate calculated from the DHIS total live births multiplied by the antenatal HIV seroprevalence rates obtained from the 2013 national antenatal HIV seroprevalence survey[30] Total number of neonates tested during the first 7 days of life Total live births obtained from the DHIS

PCR = polymerase chain reaction; DHIS = District Health Information System; IU = in utero; MTCT = mother-to-child-transmission.

S18

March 2018, Vol. 108, (3 Suppl 1)

% Testing coverage 93.5 109.7 87.9 96.8 91.0 94.2 110.7 84.8 94.6 96.4

Lowest IU case rate/antenatal prevalence category WC FS KZN

district- and provincial-level progress with EMTCT. Additionally, we synthesised factors that hinder the monitoring of EMTCT progress and attainment, from the viewpoints of provincial- and districtlevel healthcare managers and implementers. We have highlighted potential innovations to strengthen healthcare systems and improve EMTCT programme delivery. We include a focus on district-level, as opposed to a sole focus on national-level data; inclusion of districtlevel data increases the granularity of our analysis.

Methods

There are sparse published qualitative national-, provincial- or district-level data on factors affecting the monitoring of EMTCT progress and attainment, and innovative solutions to accelerate progress towards EMTCT. Consequently, to understand district-level perspectives, we reviewed reports from the provincial-level EMTCT stock-taking workshops conducted between April 2016 and March 2017. These workshops are conducted annually, are co-ordinated by the NDoH across all 52 districts, and are part of the institutionalised monitoring and evaluation activities outlined in the National Action

S19

Total live births in facility DHIS Apr 2016 - Mar 2017 878 675 19 847 91 798 117 491 203 885 55 422 39 703 101 468 70 995 178 066

PCR-positive Total PCR tests infant <7days infant <7 days (deduplicated) NHLS Apr 2016 - Mar 2017 243 925 2 153 3 809 42 15 092 154 23 088 301 53 076 488 14 721 146 13 101 86 27 032 259 25 173 231 68 833 446 Antenatal prevalence (%) 2013* 29.7 17.5 18.7 20.3 28.6 28.2 29.8 31.4 37.5 40.1 Province South Africa Northern Cape Western Cape Limpopo Gauteng North West Free State Eastern Cape Mpumalanga KwaZulu-Natal

Qualitative synthesis

Antenatal prevalence (%) Category 25 - <35 <25 25 - <35 ≥35

Only two district-level sources of data are available in SA, namely the District Health Information System (DHIS) and the National Health Laboratory System (NHLS).[19] Data from the DHIS were muddied by the fact that changes to guidelines were not accompanied by changes to DHIS indicators. There have also been concerns about the accuracy and completeness of routine DHIS data.[20] Consequently, we analysed birth HIV testing data from the routine NHLS Corporate Data Warehouse (CDW), between April 2016 and March 2017. We focused on NHLS data because the NHLS has protocols in place to clean and monitor data quality. We focused on MTCT at birth for four reasons. Firstly, the SA National Department of Health (NDoH) recommends HIV testing at birth for all HIV-exposed infants. Secondly, no other data are available to determine the precise number of HIV-infected neonates. Thirdly, birth polymerase chain reaction (PCR) results are a marker of pre-conception and antenatal care for women. Fourthly, birth PCR results serve as an early warning sign of gaps in early infant diagnosis and PMTCT. HIV PCR test data for HIV-exposed neonates tested before 7 days of age were extracted from the NHLS CDW. A patient-linking algorithm using deterministic matching of patient demographics (using name, surname and date of birth) and probabilistic matching was used to identify duplicates among birth HIV tests and to deal with minor mismatches created by spelling errors. Additional manual matching was performed to ensure that HIV PCR-positive neonates were not doubly counted. Programmatic outcomes, namely, birth testing coverage, IU percent MTCT risk, and IU case rates were calculated at national, provincial and district level from the de-duplicated data (Table 1). As maternal antenatal HIV seroprevalence and number of live births influence the absolute number of IU-infected neonates, and the case rate, we separated provinces and districts into three antenatal HIV seroprevalence categories of <25%, 25 - 34% and ≥35%. Within these categories, provinces with ≥90% and districts with ≥85% birth testing coverage, were ranked by case rates.

Table 2. Percentage positivity and case rates at the provincial level for in utero HIV infections: April 2016 - March 2017, NHLS

Quantitative data review

IU percent MTCT risk 0.9 1.1 1.0 1.3 0.9 1.0 0.7 1.0 0.9 0.6

IU case rate/100 000 live births 245 212 168 256 239 263 217 255 325 250

We aimed to deepen our understanding of district- and provinciallevel progress with eliminating MTCT using mixed methods (quantitative and qualitative). Thus, we sought data that represent an entire district or province, rather than data from one single clinic or hospital.

March 2018, Vol. 108, (3 Suppl 1)

NHLS = National Health Laboratory Services; IU = in utero; PCR = polymerase chain reaction; IU Case Rates are measured per 100 000 live births in that province according to DHIS (April 2016 - March 2017); FS = Free State; KZN = KwaZulu Natal; WC = Western Cape *2013 Antenatal HIV Prevalence Survey South Africa.

RESEARCH

Framework, 2011 - 2015, as well as the Last Mile Plan, 2016 - 2021.[13] The workshops provide an opportunity for districts to review district implementation plans (DIPs) and to track the achievement towards current targets.[21] All workshop reports were systematically reviewed by one author (WC) using the following framework questions: (i) what challenges are reported at provincial and district levels?; (ii) what game changers, i.e. radical innovations that fundamentally change PMTCT implementation, have been identified?;[23] and (iii) have the effects of the game changers been measured? This approach was consistent with our aims of moving away from a ‘broad-brushed’ national approach to understanding EMTCT progress, as well as increasing the granularity of our understanding.

programme came to the fore (Fig. 4). Despite the fact that KwaZuluNatal (KZN) was the province with the second highest number of live births and had the highest antenatal HIV prevalence, its IU MTCT risk was the lowest of all provinces (0.6%), while its IU case rate was the fifth highest (250 per 100 000 live births). Ecological observations could attribute a portion of this success to the continuous quality improvement undertaken by the province: monthly teleconferences are held with district co-ordinators to review key indicators; PCR, antenatal care and postnatal care linkage forms are used to link mothers and children into care; and weekly teleconferences are held with facilities and districts to facilitate PCR monitoring.[23]

Results

Our quantitative analysis of birth MTCT data demonstrates varying MTCT risk and case rates by province and district, with no specific relationship between antenatal HIV prevalence, IU percent MTCT risk and case rates. The IU percent MTCT risk at birth was much

Quantitative data review

Birth testing coverage was 93.5% (range 85% - 111% at provincial level); some percentages are >100% as denominators are estimated, leading to smaller denominators than the actual. The national percentage IU percent MTCT risk was 0.9%, which translated to a national IU case rate of 245 HIV-positive neonates per 100 000 live births. Provincial IU percent MTCT risk ranged from 0.6% to 1.3%, with IU case rates ranging from 168 to 325 cases per 100 000 live births (Table 2). The Western Cape, Free State and KwaZulu-Natal had the lowest case rates within each HIV prevalence category (Table 2). District-level IU percent MTCT risk and IU case rates range from 0.4% to 1.9% and 72 to 360, respectively (Fig. 3 and Table 3).

Qualitative synthesis

The 2016/17 EMTCT stock-taking workshops identified several factors that hinder EMTCT attainment, including poor internet connectivity, poor alignment between routine indicators and programmatic interventions, lack of longitudinal monitoring, poor maternal viral load monitoring and inconsistent feeding advice (Fig. 4). Despite these challenges, the workshops highlighted the following points: (i) an increasingly strong focus on strengthening pre-conception, antenatal and postnatal maternal and child health services; (ii) alignment between PMTCT interventions, the last mile EMTCT plan and DIPs, to improve programme outcomes; (iii) an increased level of accountability at district and facility levels; (iv) that discussions enabled prioritisation of high MTCT districts; and (v) the importance of involving different district management team members (District Clinical Specialist Teams (DCSTs), PHCand programme managers) in ongoing monitoring visits to ensure PMTCT integration into routine services. Ten potential game changers that are currently implemented to achieve fundamental healthcare system strengthening and durable improvement in the

Discussion

1. Poor internet connectivity and access to the weekly emails containing the NHLS Results for Action reports. 2. Poor alignment between routine indicators and PMTCT programmatic interventions. Additionally, non-reporting and inconsistent reporting yield poor quality data. 3. Lack of longitudinal cohort monitoring. 4. Poor maternal viral load monitoring. 5. Inconsistent infant feeding advice or practices.

Fig. 4. Challenges identified during the elimination of mother-to-child transmission stock-taking workshops.

Pillar 1: Leadership, governance and co-ordination • The shift in priority to district, subdistrict and facility levels, with support from DCSTs and other managers. Pillar 2: Scaling up coverage of PMTCT interventions • Quality improvement interventions linking mothers to community-based healthcare personnel • The Mothers2Mothers ‘Mother Mentor’ programme, which facilitates coverage and retention in care. Pillar 3: Integrating PMTCT into MCH • Providing safe pre-conception services to couples: Two Gauteng-based services have supported >800 couples with preconception care. There have been no recorded vertical or horizontal HIV transmissions and early antenatal care booking rates have improved for women achieving pregnancy. Additionally, no HIV-negative women have tested HIV- positive at the time of pregnancy confirmation. • A pre-printed unique identifier (ID) in each Road-to-Health Booklet (image) is being tested in the Tshwane District. This number can be used as an additional identifier on the NHLS laboratory form to facilitate continuity of care.

8.1

Pre-printed unique ID •

MTCT, %

5.9 4.7 3.5 2.2 1.4

2.5 2 0.3

2010

National average

2012/13

3.5 3 2.2 2

4.1

0.7

2.1 1.5 1.5 0.8

2013/14

2014/15

Provincial range

Pillar 4: Best practices relating to monitoring and evaluation • Dashboards and cascades at facility and district levels that examine numbers as well as percentages facilitate understanding of progress and risk • The NHLS HIV PCR and Viral Load Results for Action reports provide weekly reports to health personnel within districts and identify children that require follow-up

3.4 1.7 1.1 0.7 0 2015/16 District range

Fig. 3. Spatial distribution of IU case rates, by provincial antenatal HIV prevalence category.

S20

The Positive Baby Tool, which retrospectively assesses where breakdowns in the PMTCT cascade occurred, culminating in a vertical HIV transmission. This helps districts obtain a better understanding of their risk profiles.

Fig. 5. Potential game changers to strengthen health systems and programme delivery for the elimination of mother-to-child transmission of HIV. (DCSTs = district clinical specialist teams; PMTCT = prevention of mother-to-child transmission of HIV; MCH = maternal and child health; NHLS = National Health Laboratory Service; PCR = polymerase chain reaction.)

March 2018, Vol. 108, (3 Suppl 1)

Province District <25% Antenatal HIV prevalence NC Namakwa WC Central Karoo WC West Coast WC Overberg WC Cape Winelands LP Vhembe NC Pixley Ka Seme WC Eden LP Greater Sekhukhune NC Frances Baard NC Siyanda/ ZF Mgcawu LP Capricorn City of Cape Town WC Metro NW Ngaka Modiri Molema NC John Taolo Gaetsewe GP City of Tshwane Metro Dr Ruth Segomotsi NW Mompati LP Mopani 25 - <35% Antenatal HIV prevalence EC Alfred Nzo FS Xhariep FS Fezile Dabi LP Waterberg City of Johannesburg GP Metro Cacadu/ Sarah EC Baartman GP Sedibeng EC Buffalo City Metro FS Thabo Mofutsanyana FS Mangaung Metro

S21

86 103 537 549 1 442 4 286 442 1 087 4 425 1 656 594 5 265

2.3 6.9 9.6 13.9 15.0 15.0 15.1 15.6 18.1 18.2 20.1 21.1 21.7 22.3 23.2 23.4 23.4 24.6 25.3 25.8 25.9 27.3 27.3 27.5 29.2 29.5 30.1 30.4

1 396 931 4 828 3 193 13 579 28 874 2 759 8 964 24 558 7 280 3 924 25 863

60 303 14 593 4 488 49 375

8 921 23 904

11 590 764 6 489 14 292

March 2018, Vol. 108, (3 Suppl 1)

66 984

5 621 13 627 14 423 11 507 12 434

1 433 3 530 3 952 3 587 4 280

17 006

3 366 298 2 205 3 881

2 263 5 231

11 374 3 604 1 031 11 982

Total PCR tests infant<7 days‡

Total live births in Antenatal facility* prevalence †

17 20 33 14 39

167

34 2 15 42

25 68

103 39 5 126

1 2 8 9 15 54 7 17 58 19 10 79

PCR positive infant <7 days intrauterine (IU) infections (deduplicated)‡

1.2 0.6 0.8 0.4 0.9

1.0

1.0 0.7 0.7 1.1

1.1 1.3

0.9 1.1 0.5 1.1

1.2 1.9 1.5 1.6 1.0 1.3 1.6 1.6 1.3 1.1 1.7 1.5

IU percent MTCT risk

Table 3. Percentage positivity and case rates at district level for in-utero HIV infections: April 2016 - March 2017, NHLS

302 147 229 122 314

249

293 262 231 294

280 284

171 267 111 255

72 215 166 282 110 187 254 190 236 261 255 305

IU case rate/ 100 000

92.7 88.7 92.9 103.6 113.2

93.0

114.8 151.2 131.2 99.5

108.4 89.0

86.9 110.7 99.0 103.7

267.8 160.3 115.9 123.7 70.8 99.0 106.1 77.7 99.6 125.0 75.3 96.5

Coverage(%)§

Sedibeng (2) Thabo Mofutsanyana (1) continued...

City of Cape Town Metro (5) John Taolo Gaetsewe (3)

Namakwa (1) West Coast (4) Cape Winelands (2)

Five lowest IU case rate/ antenatal prevalence category (rank)

RESEARCH

S22

March 2018, Vol. 108, (3 Suppl 1) 35.3 35.7 36.6 37.6 38.1 38.9 39.9 40.0 40.8 41.1 44.1 45.9 42.5

8 103 10 604 7 593 36 424 16 327

18 064 12 062 11 592 15 893 54 150 13 834 9 683 15 717

6 611 5 027 4 381 6 983 21 774 5 505 3 998 6 742

2788 3 235 2 913 12 846 5 794

3 995 5409 3 445 2 731 6 905 2 853 4 190 16 368 5 344 3119

31.4 31.5 31.8 32.3 32.6 32.8 33.0 33.5 34.4 34.5

18 491 20 001 11 907 8 509 27 939 8 440 14 739 59 160 18 678 10 657

36 28 21 48 172 22 32 35

28 16 20 131 49

50 55 27 16 52 15 33 142 52 33

0.5 0.6 0.5 0.7 0.8 0.4 0.8 0.5

1.0 0.5 0.7 1.0 0.8

1.3 1.0 0.8 0.6 0.8 0.5 0.8 0.9 1.0 1.1

IU percent MTCT risk 0.8

199 232 181 302 318 159 330 223

346 151 263 360 300

270 275 227 188 186 178 224 240 278 310

IU case rate/ 100 000 258

94.1 104.5 94.5 107.7 97.8 90.2 90.0 100.9

97.5 85.5 104.8 93.8 93.1

68.8 85.9 91.0 99.4 75.8 103.1 86.1 82.6 83.2 84.8

Coverage(%)§ 103.4

IU = in utero; PCR = polymerase chain reaction; EC = Eastern Cape; FS = Free State; GP = Gauteng; KZN = KwaZulu-Natal; LP = Limpopo; MP = Mpumalanga; NW = North West; NC = Northern Cape; WC = Western Cape. *IU Case Rates are measured per 100 000 live births in that district according to DHIS (April 2016 - March 2017). † 2013 Antenatal HIV Prevalence Survey South Africa. ‡ NHLS April 2016 - March 2017. § some coverage is >100% because of the difficulty with de-duplicating data and challenges with denominators.

District Joe Gqabi Nelson Mandela Bay EC Metro NW Bojanala Platinum NW Dr Kenneth Kaunda FS Lejweleputswa EC O R Tambo KZN Amajuba GP West Rand GP Ekurhuleni Metro MP Nkangala EC Chris Hani ≥35% Antenatal HIV prevalence EC Amathole KZN Umzinyathi KZN Harry Gwala MP Ehlanzeni KZN Zululand uThungulu (King KZN Cetshwayo-DHIS) KZN Ugu KZN uThukela MP Gert Sibande KZN eThekwini Metro KZN uMkhanyakude KZN iLembe KZN uMgungundlovu

Province EC

Total PCR tests infant<7 days‡ 1 474

Total live births in Antenatal facility* prevalence† 4 644 30.7

PCR positive infant <7 days intrauterine (IU) infections (deduplicated)‡ 12

Table 3. (continued) Percentage positivity and case rates at district level for in utero HIV infections: April 2016 - March 2017, NHLS

uThungulu (4) uThukela (3) uMkhanyakude (2) uMgungundlovu (5)

Umzinyathi (1)

Lejweleputsa (5) O R Tambo (4) Amajuba (3)

Five lowest IU case rate/ antenatal prevalence category (rank)

RESEARCH

lower than the 5% MTCT target, which demonstrates success (Fig. 1 and Tables 2 and 3), but IU case rates at the district level are up to seven-fold higher than the global target. IU case rates depend on the number of HIV-exposed infants, and thus maternal HIV prevalence. From the synthesis of district-level discussions, several game changers are currently being implemented in a non-systematic way to strengthen healthcare systems and close the gaps in EMTCT programme delivery in SA. The implementation of any game changer requires close collaboration between health facilities, laboratories, communities, healthcare staff and supporting partners as data from Uganda demonstrate that the failure to strengthen healthcare systems could be a rate-limiting step to sustainable PMTCT success.[24] Evidence demonstrates that PMTCT programme outcomes improve after integration into the mainstream routine healthcare system for maternal and child health.[25] Game changers should not add to the complexity of PMTCT implementation, but should strengthen routine systems to improve service delivery. One key stumbling block to the attainment of EMTCT is poor monitoring of maternal viral suppression. Data from more experienced Option B+ countries corroborate this key challenge.[26,27] In fact, a systematic review and meta-analysis in 51 countries demonstrated 73.5% overall ART adherence (defined as >80% intake of recommended pill doses), which disaggregates into 75.7% during pregnancy and 53.3% postnatally.[26] Being on triple antiretroviral therapy rather than prophylaxis reduced the odds of adherence.[26] Individual-level data from 19 facilities in Malawi demonstrated that ART initiation among pregnant women with high CD4 cell counts were five times more likely to miss followup visits and women who started ART while breastfeeding were twice as likely to miss follow-up visits, compared with pregnant women needing ART for their own health.[27] This begs the question: Why do women not adhere to their ART and follow-up visits? SA data demonstrate that conflict with work commitments, negative treatment from healthcare workers and the lack of disclosure reduce retention in care, and by deduction, adherence.[28] HIV viral load monitoring is now even more critical following the 2016 WHO infant feeding update, which recommends exclusive breastfeeding for 6 months and continued breastfeeding for 24 months or longer for HIV-negative and HIV-positive women.[29] Thus, strengthening healthcare systems and implementing game changers that promote retention in care and infant follow-up are urgently needed. Ongoing monitoring and in-depth qualitative and quantitative studies are needed to identify high-burden districts and to test the impact of potential game changers on maternal and child health outcomes.

Conclusion

IU percent MTCT risk is significantly less than the 5% final MTCT global target however, final MTCT rates at the district, provincial and national levels need to be closely monitored, as we anticipate a national increase in breastfeeding prevalence after the 2016 infant feeding update. IU case rates are considerably higher than the global target, because of high maternal HIV prevalence. We hypothesise a possible link between using community-based approaches and CQI and lower IU percent MTCT risk/IU case rates in one province (KZN) with the highest antenatal HIV prevalence. However, more in-depth studies and data are needed to understand what quality improvement methods have the greatest impact and what proportion of the success can be attributed to community-based approaches. More district-level systematic studies are needed, to quantify bottlenecks to EMTCT and to test the impact of potential game changers on the health, development, growth and HIV-free

S23

survival of HIV-exposed infants until breastfeeding cessation, and the health of their mothers. Acknowledgements. The development and publication of this paper was supported by the South African Medical Research Council. The NHLS provided data for the analyses. C Kajese developed the map. Author contributions. All authors contributed equally to the conceptualisation, reviewing of drafts and approval of the final version of the manuscript. NN wrote the first draft of the introduction. WC wrote the first draft of the qualitative methods and results and undertook the analysis of EMTCT stock-taking reports. GS wrote the first draft of the quantitative methods and results, and provided data from the NHLS. FM extracted and analysed data on IU case rates and assisted with the write up. ND contributed information from the preconception strengthening intervention. UF contributed information on the unique identifiers used in the Tshwane district. OM contributed information about the interventions in KwaZulu-Natal. SB and KN provided general strategic direction for the paper. MN and TT liaised with provinces to obtain best practices and game changers. AG coordinated the writing process, combined all the contributions, synthesised the paper, and circulated drafts for comments. AG wrote the first draft of the discussion. Funding. The South African Medical Research Council paid for the time of AG, WC and NN, and, in partnership with UNICEF, covered the cost of this publication. Conflicts of interest. None

1. World Health Organization (WHO), Alliance for Health Policy and Systems Research. World Report on Health Systems and Health Policy Research. Geneva: WHO, 2017. http://www.who.int/alliancehpsr/news/2017/worldreport-hpsr/en/ (accessed 11 August 2017). 2. WHO. Prevention of HIV in Infants and Young Children: Review of Evidence and WHOâ&#x20AC;&#x2122;s Activities. Geneva: WHO, 2002. http://www.popline.org/node/236104 (accessed 17 August 2017). 3. WHO. Stategic Approaches to the Prevention of HIV Infection in Children: Report of a WHO Meeting, Morges Switzerland 20-22 March 2002. http://www.who.int/hiv/mtct/StrategicApproaches. pdf (accessed 17 August 2017). 4. WHO. Antiretroviral Drugs for Treating Pregnant Women and Preventing HIV Infection in Infants: Towards Universal Access. Recommendations for a Public Health Approach. Geneva: WHO, 2006. http://www.who.int/hiv/pub/mtct/arv_guidelines_mtct.pdf (accessed 17 August 2017) 5. WHO. Antiretroviral Drugs for Treating Pregnant Women and Preventing HIV Infection in Infants: Towards Universal Access. Recommendations for a Public Health Approach. Geneva: WHO, 2010. http://www.who.int/hiv/pub/mtct/arv_guidelines_mtct.pdf (accessed 17 August 2017). 6. WHO. Use of Antiretroviral Drugs for Treating Pregnant Women and Preventing HIV Infection in Infants. Geneva: WHO, 2012. http://www.who.int/hiv/pub/mtct/programmatic_update2012/en/ (accessed 17 August 2017). 7. Barron P, Pillay Y, Doherty T, et al. Eliminating mother-to-child HIV transmission in South Africa. Bull World Health Organ 2013;91:70-74. https://doi.org/10.2471/blt.12.106807 8. National Department of Health (NDoH). Policy and Guidelines for the Implementation of the PMTCT programme. Pretoria: NDoH, 2008. http://www.ilo.org/wcmsp5/groups/public/---ed_protect/-protrav/---ilo_aids/documents/legaldocument/wcms_125633.pdf (accessed 17 August 2017). 9. NDoH. The South African Antiretroviral Treatment Guidelines 2013. Pretoria: NDoH, 2013. http:// www.kznhealth.gov.za/medicine/2013_art_guidelines.pdf (accessed 17 August 2017). 10. NDoH, South African National AIDS Council. Clinical Guidelines: PMTCT (Prevention of Motherto-Child Transmission). Pretoria: NDoH, 2010. http://www.fidssa.co.za/Content/Documents/ PMTCT_Guidelines.pdf (accessed 17 August 2017). 11. NDoH. National Conslidated Guidelines for the Prevention of Mother to Child Transmission of HIV (PMTCT) and the Management of HIV in Children, Adolescents and Adults. Pretoria: NDoH, 2015. https://www.health-e.org.za/2015/07/02/guidelines-national-consolidated-guidelines-for-pmtct-andthe-management-of-hiv-in-children-adolescents-and-adults/ (accessed 17 August 2017). 12. Violari A, Cotton M, Gibb D, et al. Early antiretroviral therapy and mortality among HIV-infected infants. N Engl J Med 2008;359:2233-2244. https://doi.org/10.1056/nejmoa0800971 13. NDoH, South Africa. The Last Mile Plan for EMTCT. Pretoria: NDoH, 2016. http://www.emtctthelastmile.co.za/ (accessed 11 August 2017). 14. Goga A, Dinh T, Jackson D, et al. Population-level effectiveness of maternal antiretroviral treatment initiation before or during the first trimester and infant antiretroviral prophylaxis on early motherto-child transmission of HIV, South Africa: Implications for eliminating MTCT. J Glob Health 2016;6(2):020405. https://doi.org/10.7189/jogh.06.020405 15. Massyn N, Peer N, English R. District Health Barometer 2015/6. Durban: Health Systems Trust, 2016. http://www.hst.org.za/publications/Pages/HSTDistrictHealthBarometer.aspx (accessed 17 August 2017). 16. Massyn N, Peer N, Padarath A, Barron P, Day C. District Health Barometer 2014/15. Durban: Health Systems Trust, 2015. https://www.health-e.org.za/wp-content/uploads/2015/10/Complete_ DHB_2014_15_linked.pdf (accessed 10 August 2017). 17. Rollins N, Little K, Mzoloa S, Horwood C, Newell M-L. Surveillance of mother-to-child transmission prevention programmes at immunisation clinics: The case for universal screening. AIDS 2007;21(10):1341-1347. https://doi.org/10.1097/qad.0b013e32814db7d4 18. Sherman G, Lillian R. Early Infant Diagnosis of HIV Infection in South Africa: 2008 to 2010. http:// www.nhls.ac.za/assets/files/EID_HIV_PCR_2008-2010.pdf (accessed 17 August 2017).

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

19. Goga A, Sherman G, Chirinda W, et al. Eliminating mother-to-child transmission of HIV in South Africa, 2002-2016: Progress, challenges and the last mile plan. In: Padarath A, Barron P, edS. S Afr Health Rev 2017:137-146. http://www.hst.org.za/publications/South%20African%20Health%20 Reviews/13_Eliminating%20mother%20to%20child%20transmission%20of%20HIV%20in%20South%20 Africa_2002%20to%202016_progress_challanges%20and%20the%20Last%20Mile%20Plan.pdf 20. Nicol E, Dudley L, Bradshaw D. Assessing the quality of routine data for the prevention of motherto-child transmission of HIV: An analytical observational study in two health districts with high HIV prevalence in South Africa. Int J Med Inform 2016;95:60-70. https://doi.org/10.1016/j. ijmedinf.2016.09.006 21. Joint United Nations Programme on HIV/AIDS (UNAIDS). 90-90-90 An ambitious treatment target to help end the AIDS epidemic. Geneva: UNAIDS, 2014. http://www.unaids.org/en/resources/ documents/2014/90-90-90 (accessed 16 July 2017). 22. De Cock K, El-Sadr W, Ghebreyesus T. Game changers: why did the scale up on HIV treatment work despite weak health systems. J Acquir Immune Defic Syndr 2011;57(Suppl 2):S61-S63. https://doi. org/10.1097/qai.0b013e3182217f00 23. Mhlongo O. Yes we can eliminate: A case of KZN last mile approach to EMTCT. 8th South African AIDS Conference, Durban, 13 - 15 June 2017. 24. Doherty T, Besada D, Goga A, Daviaud E, Rohde S, Raphaely N. “If donors woke up tomorrow and said that we can’t fund you, what would we do?” A health system dynamics analysis of implementation of PMTCT Option B+ in Uganda. Globalization Health 2017;13:51. https://doi. org/10.1186%2Fs12992-017-0272-2

S24

25. Car L, Van Velthoven M, Brusamento S, et al. Integrating prevention of mother-to-child HIV transmission programs to improve uptake: A systematic review. PLOS ONE 2012;7(4):e35268. https:// doi.org/10.1371/journal.pone.0035268 26. Nachega J, Uthman O, Anderson J, et al. Adherence to antiretroviral therapy during and after pregnancy in low-income, midddle-income, and high-income countries: A systematic review and meta-analysis. AIDS 2012;26(16):2039-2052. https://doi.org/10.1097/qad.0b013e328359590f 27. Tenthani L, Haas AD, Tweya H, et al. Retention in care under universal antiretroviral therapy for HIV-infected pregnant and breastfeeding women (‘Option B+’) in Malawi. AIDS 2014;28(4):589-598. https://doi.org/10.1097/qad.0000000000000143 28. Clouse K, Schwartz S, Van Rie A, Bassett J, Yende N, Pettifor A. “What they wanted was to give birth; nothing else”: Barriers to retention in Option B+ HIV care among postpartum women in South Africa. J Acquir Immune Defic Syndr 2014;67(1):E12-E18. https://doi.org/10.1097/qai.0000000000000263 29. African Network for Research & Training in Sexual & Reproductive Health & HIV. 2016 Guideline Update on HIV and Infant Feeding. http://repronetafrica.org/2016-guideline-update-on-hiv-andinfant-feeding/ (accessed 17 August 2017). 30. NDOH. 2013 Antenatal HIV prevalence survey South Africa. Pretoria: NDoH, 2016. https://africahealthnews. com/antenatal-hiv-prevalence-survey-south-africa-published/ (accessed 17 August 2017).

Accepted 30 October 2017.

March 2018, Vol. 108, (3 Suppl 1)

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

RESEARCH

Child mortality in South Africa: Fewer deaths, but better data are needed L J Bamford,1 FCPaed (SA), DrPH; N H McKerrow,2 FCPaed (SA), MMed (Paed); P Barron,3 BCom, FFCH (SA); Y Aung,4 MBBS, MPH National Department of Health, Pretoria, South Africa KwaZulu-Natal Department of Health; and Department of Paediatrics and Child Health, University of KwaZulu-Natal, Durban, South Africa 3 School of Public Health, University of the Witwatersrand, Johannesburg, South Africa 4 UNICEF, Pretoria, South Africa 1 2

Corresponding author: L J Bamford (lesley.bamford@health.gov.za)

South Africa is committed to reducing under-5 mortality rates in line with the Sustainable Development Goal (SDG) targets. Policymakers and healthcare service managers require accurate and complete data on the number and causes of child deaths to plan and monitor healthcare service delivery and health outcomes. This study aimed to review nationally representative data on under-5 mortality and the cause of deaths among children under 5 years of age. We also reviewed systems that are currently used for generating these data. Child mortality has declined substantially in the past decade. Under-5 mortality in 2015 is estimated at 37 - 40 deaths per 1 000 live births, with an estimated infant mortality rate of 27 - 33 deaths per 1 000 live births. Approximately one-third of under-5 deaths occur during the newborn period, while diarrhoea, pneumonia and HIV infection remain the most important causes of death outside of the newborn period. The proportion of deaths owing to non-natural causes, congenital disorders and non-communicable diseases has increased. However, many discrepancies in data collected through different systems are noted, especially at the sub-national level. There is a need to improve the completeness and accuracy of existing data systems and to strengthen reconciliation and triangulation of data. S Afr Med J 2018;108(3 Suppl 1):S25-S32. DOI:10.7196/SAMJ.2018.v108i3.12779

Reducing child mortality represents an important health sector and societal goal at a global[1] and national level.[2] Sustainable Development Goal (SDG) 3 calls for an â&#x20AC;&#x2DC;end to preventable deaths of newborns and children under five yearsâ&#x20AC;&#x2122;, which demands a reduction of under-5 mortality to <25 per 1 000 live births and the neonatal mortality rate to be <12 per 1 000 live births in every country by 2030 (SDG indicators 3.2.1 and 3.2.2, respectively).[3] Policymakers and healthcare service managers require accurate and complete data on the number and causes of child deaths to plan and monitor child health and child healthcare service delivery.[4,5] The preferred source of data for the monitoring of deaths and under-5 mortality rates is a civil or vital registration (VR) system that records births and deaths on a continuous basis, collects information as events occur and covers the entire population.[6] If registration coverage is complete and the systems function efficiently, the resulting child mortality estimates will be accurate, timely and available at national, sub-national and local levels. However, many countries, including South Africa (SA), remain without viable or fully functioning VR systems that accurately record all births and deaths.[7,8] The study aimed to review nationally representative data on the levels and trends of under-5 mortality rates, the causes of death among children under 5 years of age in SA, and current systems that report on child mortality on a national and sub-national level to identify gaps, and to make recommendations for improving the availability of complete, accurate and timely information on child deaths.

Methods

The following major national systems that present data on child deaths were reviewed and the key features of the systems were summarised and compared: the Vital Registration (VR) System of Statistics South

S25

Africa (StatsSA); the Rapid Mortality Surveillance (RMS) of the South African Medical Research Council (SAMRC), which is used by the National Department of Health (NDoH) to monitor trends in child mortality;[9] the District Health Information System (DHIS) of the NDoH; the United Nations (UN) Inter-agency Group for Child Mortality Estimation; the South African Demographic and Health Survey (SADHS) of StatsSA; and the population census of StatsSA. The currently available national and sub-national estimates of under5 mortality and causes of death in these systems were reviewed and compared. Where available, published data were used, while DHIS and VR data were extracted directly from the relevant databases. The comparisons allowed the authors to identify gaps and discrepancies between the different data systems, and these discrepancies are also described.

Results

The information systems and surveys that collect nationally representative data, or provide national estimates, on child deaths are summarised in Table 1. A brief description of each system/ survey is provided alongside its strengths and limitations. Each of the systems estimates or calculates child mortality numbers and rates using different data sources, different methods and different interpretations; taken together they provide an overall perspective on child mortality.

National estimates of under-5 deaths and mortality rates

Under-5 mortality rates in SA are shown in Fig. 1. The HIV epidemic during the late 1990s and early 2000s, with a peak in 2003, led to an increase in infant and child mortality. However, infant and under5 mortality rates have declined during the past decade. The UN figures show that under-5 mortality peaked in 2003 (at 78.1 per

March 2018, Vol. 108, (3 Suppl 1)

Demographic and Health Survey (DHS)[13]

Rapid Mortality Surveillance (RMS)[8]

System United Nations (UN) Interagency Group on Child Mortality Estimation[6]

S26 Methods are rigorous and standardised: this facilitates identification of trends and comparison with DHS data from other countries. Nationally representative. Provides information on the context of health and ill-health in terms of the socio-economic conditions. Information on all deaths is collected, irrespective of whether these deaths occurred within the health system, or were officially reported.

March 2018, Vol. 108, (3 Suppl 1)

continued...

Difficult and costly to conduct: as a result are only conducted on an intermittent basis. Data on under-5 mortality covers the preceding five years, and are not comparable with annual figures. Does not provide accurate estimates of child mortality at provincial and district levels. Data on neonatal mortality probably includes stillbirths; thus neonatal mortality appears artificially high[14,15]

Estimates include all under-5 deaths. Provides only national estimates (not disaggregated to Provides data on an annual basis with a sub-national level). lag period of 1 year. Data are adjusted based on an estimation of completeness of death notification. However, the levels of completeness for children are uncertain.[11]

Mortality rates

Limitations Depends on accuracy of primary data, as well as adjustments. For example, VR data adjusted for incompleteness of reporting are used. However, levels of completeness in children are uncertain.[11] Provides only national estimates (not disaggregated to sub-national level).

Advantages Estimates include all under-5 deaths Provides data on an annual basis with a lag period of ~9 months.

Type of data Number of deaths Mortality rates

The number of under-5 and infant deaths is derived from the number of children (under-5) who are removed from the National Population Register by the Department of Home Affairs, following notification of the death. These deaths are corrected for under-reporting of both births and deaths. The number of neonatal deaths is based on data from the DHIS (adjusted for under-reporting). Mid-year population estimates with an age distribution and size consistent with those of the 2001 and 2011 censuses are used to calculate mortality rates.[12] National household survey to identify Mortality rates demographic characteristics and define key health indicators. Should ideally be conducted every five years. Conducted in SA in 1998, 2003 and 2016.

Description Estimates are based on all available nationally representative population-based primary data on child deaths which are adjusted using standard methods. A statistical model is then applied to provide a smooth trend curve that averages over disparate estimates from the different data sources for a country, and extrapolates the model to a target year.[10] The RMS system aims to provide data based on death notification in a more timely manner than the routine reports provided by StatsSA.

Table 1. Data sources on child deaths in South Africa

RESEARCH

S27

March 2018, Vol. 108, (3 Suppl 1)

VR data include both death notification (notification of death by a health professional which includes cause of death) and death reporting (notification of death by a traditional leader or police officer without any indication of the cause of death.) The death notification/reporting forms need to reach a Home Affairs office. They are then sent to StatsSA where the data are captured, collated, analysed and published annually. All under-5 deaths occurring within the public health system are reported as part of the routine health information and management system.

Description All households are asked about child deaths in the 12 months preceding the census. The last census was conducted between 9th and 31st October 2011.

StatsSA = Statistics South Africa; NMR = neonatal mortality rate.

District Health Information System (DHIS)[18]

Vital Registration (VR) System[17]

System Population census[16]

Table 1. (continued) Data sources on child deaths in South Africa

Does not include details regarding cause of death (apart from data on deaths from diarrhoea, pneumonia and severe acute malnutrition). Does not include deaths outside of public healthcare facilities. Reporting may be incomplete (especially from larger hospitals)

Number of deaths by age category (NMR is calculated using hospital births as a denominator). Number of deaths owing to diarrhoea, pneumonia and severe malnutrition.

Data on the number of under-5 deaths are available for every healthcare facility in the public sector. Data are available timeously (6 weeks after the end of a month period).

Limitations May over-report child deaths due to double-counting (a child may be reported by more than one household) and inclusion of stillbirths and deaths which occurred more than 12 months prior to the census. Conducted once every 10 years. Not all deaths are captured. This may be because the death is not notified or reported, or because the death notification form does not reach a Department of Home Affairs office. Accurate estimations of the degree of completeness of reporting of child deaths are not available.[11] Data for reported deaths may be incomplete. Delay of ~2 years before data are published.

Advantages All households are included and therefore all under-5 deaths should be captured. Figures are corrected for underreporting. Number and cause Includes deaths in both the public and of death private health sectors, as well as those that occur outside the health sector. Includes data on age, gender, date, site and primary cause of death. StatsSA maintains a database on all deaths in children under 5 years of age

Type of data Number of deaths

RESEARCH

1 000 live births) before declining steadily to 40.5 per 1 000 live births in 2015. The RMS figures suggest a more rapid decline from 2008 to 2010, followed by a much slower decline to 37 per 1Â 000 live births in 2015. UN data estimate that the total number of under-5 deaths fell from 63 994 in 1990 to 41 930 in 2015.[19] The infant mortality rate followed a similar pattern as the under-5 mortality rate, with the UN reporting an infant mortality rate of 33.6 per 1 000 live births in 2015, while the RMS reported a figure of 27 per 1 000 live births. The 2016 SADHS reported an under-5 mortality rate of 42 per 1 000 and an infant mortality rate of 35 per 1 000 in the 5 years preceding the survey.[13] While newborn deaths are discussed in more detail elsewhere in this special issue,[20] the contribution of neonatal mortality to the overall under-5 mortality should be noted. A number of sources, i.e. UN estimates, RMS and DHIS reported neonatal mortality rates of ~11 - 12 per 1 000 live births in 2015. In contrast, the 2016 SADHS reported a neonatal mortality rate of 21 per 1 000. UN estimates and RMS data suggest that the proportion of under-5 deaths that occur in the neonatal period has increased in the past decade: UN estimates show an increase from 18% in 2005 to 27% in 2015 and RMS estimates suggest a similar pattern. In contrast, SADHS data suggest that newborn deaths accounted for 50% of under-5 deaths.[13] This figure is more closely aligned with the global estimate that 45% of under-5 deaths occur during the newborn period,[21] but may be artificially high owing to misclassification of stillbirths as neonatal death, which has been shown to occur during such surveys in other countries.[14,15]

Total number of under-5 deaths

The number of under-5 deaths reported through the VR system decreased from 37 908 in 2011 to 31 938 in 2015, with a consistent annual decrease in the deaths (Table 2). It should be noted that the VR database is released on an annual basis; the number of deaths in recent years may therefore

RESEARCH

Deaths per 1 000 live births, n

Under-five Rate: Under-5Mortality MR: SADHS SADHS Under-five Under-5Mortality MR: UNRate: UN Under-five Rate: Under-5Mortality MR: RMS RMS Infant IMR:Mortality SADHSRate: SADHS Infant Rate: UN IMR:Mortality UN

70 60 50 40 30

Infant IMR:Mortality RMS Rate: RMS

20 10 2014

2012

2010

2008

2006

2004

2001

2000

1998

1996

1994

1992

1990

Neonatal Mortality Rate: NMR: SADHS SADHS Neonatal Mortality Rate: NMR: RMS RMS NNR: UN Neonatal Mortality Rate:

Fig. 1. Child mortality rates in South Africa between 1990 and 2015 based on different sources.[8,13,19,21] (MR = mortality rate; SADHS = South African Demographic and Health Survey; UN = United Nations; RMS = Rapid Mortality Surveillance, IMR = infant mortality rate; NMR = neonatal mortality rate.)

through the 2011 census are shown in Table 3. More deaths were reported through the census than through VR for 2011; this is presumed to reflect incomplete VR, although the census may also have over-reported deaths (Table 1). The 54 250 child deaths reported in the 2011 census was lower than the UN estimate of 60 356 for 2011 (range 50 256 - 72 132).[19] The ratio of the difference between the two sources of death numbers ranged from 0.7 (in the Western Cape, which was an exception with more deaths notified in the VR system than were reported in the census) to 2.1 in the Eastern Cape (with more than twice as many deaths in the census compared with the deaths that were notified in the VR system). There were also large differences in KwaZulu-Natal and Mpumalanga, which, like the Eastern Cape, have large underserved rural populations where incomplete VR would be expected.[11,23]

Table 2. Number of under-5 deaths by province, VR: 2011 - 2015 Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga North West Northern Cape Western Cape South Africa

2011 3 678 3 754 8 330 7 194 4 204 2 976 3 830 1 222 2 462 37 908

2012 3 419 2 928 7 716 7 342 5 042 3 096 3 300 1 112 2 479 36 915

2013 3 242 2 886 7 718 6 506 4 707 2 725 3 588 1 295 2 280 35 094

2014 3 614 2 671 7 699 5 813 4 555 2 769 3 532 1 299 2 209 34 262

2015 3 240 2 356 7 348 5 372 4 426 2 597 3 171 1 068 2 319 31 938

VR = vital registration.

Table 3. Comparison of under-5 deaths reported through VR and the population census, by province, 2011[16] Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga North West Northern Cape Western Cape South Africa

Deaths in VR 3 687 3 754 8 330 7 194 4 204 2 976 3 830 1 222 2 462 37 908

Census 7 759 4 015 8 591 14 842 5 405 5 223 4 774 1 239 1 713 54 250

Census: VR ratio 2.10 1.07 1.03 2.06 1.26 1.75 1.25 1.01 0.70 1.43

VR = vital registration.

Deaths inside and outside of the health system

Just over half (50.9%) of all under-5 deaths in the VR system were reported to have occurred within health facilities in 2015 (Table 4). The lowest proportion was reported in the Eastern Cape (43.5%) and the highest proportion in the Northern Cape (67.7%). The number of VR deaths occurring in health facilities includes deaths that occur in both public and private health facilities. Therefore, the number of these deaths would be expected to be higher than the number of deaths reported through the DHIS, which collects data from public health facilities only. However, in 2015, there were 2 963 (18%) more in-facility deaths reported through the DHIS than through VR (Table 4). This ranged from 28% fewer deaths in the DHIS in the North West to 126% more deaths in the DHIS in the Eastern Cape.

Age distribution of under-5 deaths

increase slightly as the database is updated to include deaths that were recently registered. Similar downward trends in the number of deaths were evident at sub-national (provincial) level, although some inconsistencies are apparent. The number of deaths reported in the Free State and Gauteng declined consistently year-on-year over the 5-year period. However, in the other provinces, although the overall trend was downwards, there were interim increases. Limpopo was the only province in which more deaths were reported in 2015 than in 2011, although there has been a steady decline from 2012 - 2015. As noted in Table 1, not all deaths are captured in the VR system, and there is uncertainty regarding the degree of completeness of reporting for young children. Data on under-5 deaths collected

S28

The number and proportion of deaths by age category in 2015 as reported through the VR system and the DHIS are shown in Tables 5 and 6, respectively. The VR data show that a third of the reported deaths (33.4%) occurred in the neonatal period and three-quarters (75.8%) in the first year of life. The Western Cape reported the highest proportion of deaths in the neonatal period (40.1%) and the first year of life (82.3%), while the Eastern Cape reported the lowest proportion of deaths in the neonatal period (20.1%) and the highest proportion in the latter two age categories (47.8% and 32.2%, respectively). The DHIS data show that 11 979 deaths took place in the neonatal period (first 28 days of life). This was 12% more than the deaths notified through the VR system, and 37% more than the number of deaths that were recorded as having occurred in healthcare facilities

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Table 4. In-facility deaths reported to the VR system and the DHIS in 2015

Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga North West Northern Cape Western Cape South Africa

VR deaths occurring in healthcare facilities, n (%) 1 411 (43.5) 1 190 (50.5) 3 839 (52.2) 3 165 (58.9) 2 021 (45.7) 1 328 (51.1) 1 532 (48.3) 723 (67.7) 1 054 (45.5) 16 272 (50.9)

VR deaths 3 240 2 356 7 348 5 372 4 426 2 597 3 171 1 068 2 319 31 938

DHIS deaths, n 3 185 867 3 577 4 351 2 809 1 545 1 104 570 1 227 19 235

Ratio of DHIS deaths: VR deaths in healthcare facilities 2.26 0.73 0.93 1.37 1.39 1.16 0.72 0.79 1.16 1.18

VR = vital registration; DHIS = District Health Information System.

Table 5. VR under-5 deaths by age category, 2015 Â Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga North West Northern Cape Western Cape South Africa

Neonatal (0 - 27 days), n (%) 650 (20.1) 793 (33.7) 2 909 (39.6) 1 961 (36.5) 1 199 (27.1) 745 (28.7) 1 113 (35.1) 351 (32.9) 930 (40.1) 10 663 (33.4)

28 days - 1 year, n (%) 1 548 (47.8) 1 058 (44.9) 2 924 (39.8) 2 097 (39.0) 1 965 (44.4) 1 162 (44.7) 1 350 (42.6) 457 (42.8) 979 (42.2) 13 553 (42.4)

1 - 4 years, n (%) 1 042 (32.2) 505 (21.4) 1 515 (20.6) 1 314 (24.5) 1 262 (28.5) 690 (26.6) 708 (22.3) 260 (24.3) 410 (17.7) 7 722 (24.2)

Total, N 3 240 2 356 7 348 5 372 4 426 2 597 3 171 1 068 2 319 31 938

VR = vital registration.

Table 6. DHIS under-5 deaths in public health facilities, 2015 Â Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga North West Northern Cape Western Cape South Africa

Neonatal (0 - 28 days), n (%) 1 897 (59.6) 544 (62.7) 2 600 (72.7) 2 546 (58.5) 1 723 (61.3) 821 (53.1) 349 (61.2) 719 (65.1) 780 (63.6) 11 979 (62.3)

29 days - 1 year, n (%) 536 (16.8) 127 (14.6) 399 (11.2) 772 (17.7) 496 (17.7) 287 (18.6) 106 (18.6) 174(15.8) 121 (9.9) 3 018 (15.7)

1 - 4 years, n (%) 752 (23.6) 196 (22.6) 578 (16.2) 1 033 (23.7) 590 (21.0) 437 (28.3) 115 (20.2) 211 (19.1) 326 26.6) 4 238 (22.0)

Total, N 3 185 867 3 577 4 351 2 809 1 545 570 1 104 1 227 19 235

DHIS = District Health Information System.

through the VR system (Table 7). As noted above, under-reporting through the VR system is known to occur in rural areas[11,23] and so it is not unexpected that the difference was highest in provinces with large rural populations (Eastern Cape, Limpopo and KwaZulu-Natal). Furthermore, 5 673 deaths in infants between 29 days and one year of age were reported as having occurred in healthcare facilities through the VR system (Table 7), while only 3 018 deaths among children in this age category were reported through the DHIS (Table 6). This pattern was consistent across all provinces.

S29

Cause of death

UN and VR cause-of-death data for 2015 are shown in Figs 2 and 3, respectively. It should be noted that the underlying cause, i.e. the disease that precipitated the change of events that resulted in the death, is used when reporting on the cause of a death. HIV infection is considered to be the cause of death in all cases where this is noted on the death certificate (e.g. as a contributory cause). In contrast, malnutrition is only coded as the cause of death if this is listed as the underlying cause of death on the death certificate.

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Other 25.0% Neonatal conditions 19.3%

Other 25.0%

Neonatal conditions 19.3%

Sepsis 3.8%

Pneumonia 16.9%

Sepsis 3.8%

Pneumonia 16.9%

The proportional contribution of the leading causes of death in under5s for the period 2011 - 2015 is shown in Table 8. There has been an increase in the proportion of deaths owing to neonatal disorders, congenital abnormalities, malnutrition and non-natural deaths (Table 8). At the same time there has been a decline in the proportion of deaths attributed to diarrhoea (gastroenteritis) and tuberculosis, and minimal or no change in the proportion of deaths owing to pneumonia and HIV, or where the cause of death is ill-defined (Table 8).

Discussion

Congenital 8.6% Congenital 8.6% AIDS 8.7% AIDS 8.7%

Injury 9.0% Injury 9.0%

Diarrhoea 8.7% Diarrhoea 8.7%

Fig. 2. Cause of under-5 deaths based on 2015 United Nations estimates.[19] Ill-defined 14.7% Ill-defined 14.7% Neonatal disorder 29.6% Neonatal disorder 29.6% Other 16.0% Other 16.0%

HIV 1.2% HIV 1.2% Malnutrition 4.6%

Gastroenteritis 10.1%

Malnutrition 4.6% Congenital abnormalities 5.9%

Gastroenteritis 10.1%

Congenital abnormalities 5.9% Non-natural 7.9% Non-natural 7.9%

Pneumonia 9.9% Pneumonia 9.9%

Fig. 3. Cause of under-5 deaths, Vital Registration 2015.

The number of under-5 deaths, as well as under-5 mortality rates in SA, have declined in the past decade. Under-5 mortality is estimated at 37 - 40 deaths per 1 000 live births for 2015 with an infant mortality rate estimated at 27 - 33 deaths per 1 000 live births. Neonatal mortality accounts for approximately one-third of all under-5 deaths. UN and VR figures both suggest that, outside of the newborn period, pneumonia, diarrhoea and non-natural deaths are the leading causes of death. Declines in the number of deaths and case fatality rates associated with diarrhoea, pneumonia and severe acute malnutrition have contributed substantially to the decline in under-5 mortality. These are discussed in another article in this supplement.[24] It should be noted that VR data underestimate the contribution of HIV infection and malnutrition to under-5 mortality, albeit for different reasons. HIV infection is considered to be the cause of death in all cases were this is noted on the death certificate (e.g. as a contributory cause), and low rates of HIV infection reflect reluctance to record a diagnosis of HIV infection on the death notification form.[25] The UN estimates that 8.7% of deaths were due to HIV in 2015 is likely to be more accurate, although it remains below the figure of 19% estimated in the Second Burden of Disease study for 2012.[26]

Table 7. Under-5 deaths in health facilities reported through the VR system, 2015 Â Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga North West Northern Cape Western Cape South Africa

Neonatal (0 - 28 days), n (%) 438 (31.0) 518 (43.5) 2 029 (52.9) 1 521 (48.1) 842 (41.7) 552 (41.6) 252 (34.9) 705 (46.0) 649 (61.6) 7 512 (46.2)

29 days - 1 year, n (%) 616 (43.7) 459 (38.6) 1 247 (32.5) 1 049 (33.1) 692 (34.2) 508 (38.3) 293 (40.5) 542 (35.4) 267 (25.3) 5 673 (34.9)

1 - 4 years, n (%) 357 (25.3) 213 (17.9) 563 (14.7) 595 (18.8) 487 (24.1) 268 (20.2) 178 (24.6) 285 (18.6) 138 (13.0) 3 087 (19.0)

Total, N 1 411 1 190 3 839 3 165 2 021 1 328 723 1 532 1 054 16 272

VR = vital registration.

Table 8. Cause of under-5 deaths as reported through the VR system, 2011 - 2015 Cause Neonatal disorder Gastroenteritis Pneumonia Non-natural Malnutrition Congenital abnormality Tuberculosis HIV Other Ill-defined

2011 23.1 15.0 11.0 6.3 3.9 3.9 2.0 1.2 19.1 14.5

2012 25.0 12.3 11.2 7.1 3.9 1.6 1.0 -

2013 22.6 14.6 11.2 7.0 3.7 4.2 1.5 1.7 18.6 14.9

VR = Vital Registration; - = data not available.

S30

March 2018, Vol. 108, (3 Suppl 1)

2014 23.4 14.0 10.8 7.3 4.7 4.9 1.4 1.4 17.9 14.2

2015 29.6 10.1 9.9 7.9 4.6 5.9 1.1 1.2 14.9 14.7

RESEARCH

In contrast, malnutrition is only recorded as the cause of death if this is recorded as the underlying cause (and not the immediate or contributory cause) of death. Malnutrition, however, remains an important contributory cause of death as shown by data from child mortality audits which show that almost a third (30.9%) of children who die are severely malnourished.[27] As expected, as deaths from diarrhoea, pneumonia and SAM decline, deaths from other causes become more prominent.[28] These include deaths during the newborn period, deaths due to non-natural causes as well as deaths associated with congenital abnormalities and non-communicable diseases. Further reductions in child mortality will require ongoing efforts to reduce deaths from traditional causes, as well as a scale-up of interventions aimed at reducing deaths from these emerging causes. Non-natural deaths are particularly challenging, given that most of the interventions required to reduce them lie outside of the health sector. The high proportion of VR deaths from ill-defined causes reflect both poor completion of death notification forms by health professionals, as well as the fact that death reporting forms (which are completed by community leaders) do not capture cause-of-death information. A high proportion of child deaths occur outside of healthcare facilities. Just over half (50.9%) of the deaths reported through VR occurred outside of the healthcare system, while the DHIS was able to account for ~19 000 out of an estimated 40 000 under-5 deaths[19] in 2015. While violence and injury (occurring at home or in the community) account for some of these deaths, a number of recent studies suggest that the majority of caregivers of sick children recognise the need for medical care for their ill child, and seek appropriate care from a variety of providers including the public health sector; however, they are either discouraged by inadequate communication and care or do not have sufficient resources to allow follow-up visits when their child’s condition deteriorates or fails to improve.[29] Better VR data is important as it allows for counting of all deaths (irrespective of where they occurred) and thus has the potential to provide accurate disaggregated data that are required for monitoring health outcomes and health service delivery at local levels. The VR system should provide these data at sub-national and local level. Although under-5 mortality rates can be calculated using the number of registered births and VR deaths, incomplete reporting means that these figures are not sufficiently robust to allow for useful comparison at provincial and sub-national levels. Recommendations for improving the availability of more accurate data include: • Improvement in the quality of information on death certificates, especially regarding cause-of-death data. Improved training, better linkages to mortality audit processes and routine review of the accuracy and completeness of death notification forms by senior clinicians are all likely to contribute positively to this measure. • Improved completeness of the VR. This requires two complementary strategies. The first relates to ensuring that death notification forms reach the Department of Home Affairs (and are then captured into the VR system). Most public-sector hospitals have Home Affairs offices; these offices currently focus primarily on birth registration but have the potential to play an important role in improving the completeness of VR. The second strategy relates to ensuring that all deaths that occur outside of healthcare facilities are reported. Both strategies would require closer collaboration between health facilities, the Department of Home Affairs and StatsSA at local, provincial and national levels. • Better triangulation of data from various sources at all levels i.e. national, provincial and district levels. District Clinical Specialist

S31

Teams will soon be required to complete a Quarterly Death Report, which will triangulate data on maternal, newborn and under-5 deaths from various sources. This will provide a mechanism for ensuring that DHIS data, mortality audit data and data from death notification forms are reconciled and triangulated. Particular attention should be placed on ensuring concordance between the number and age disaggregation of in-facility deaths reported through the DHIS and the VR system. • Collaboration with StatsSA to ensure that censuses and other surveys collect relevant and accurate data on child deaths, and that these are triangulated with routine data sources. Where available, data on child deaths from other sources such as surveillance sites, mortuary data and registries of children with chronic diseases should be integrated with routine data systems. Attention should be paid to identifying good practices with a view to ensuring that there is scale-up of these practices. More research and/or investigation is required to explain and address the discrepancies in the number of newborn deaths reported through the different systems. There is also a need to better understand the extent of unreported community deaths.

Conclusion

Under-5 mortality rates in South Africa have fallen substantially in the past decade, and the SDG target of an under-5 mortality rate of less than 25 per 1 000 live births by 2030 is achievable. However, this will require ongoing declines in deaths from diarrhoea, pneumonia and HIV, as well as declines in deaths during the newborn period, and from emerging prominent causes of death (violence and injury, congenital disorders and non-communicable disorders). More attention will also need to be paid to understanding and addressing the high number of child deaths which occur outside of the healthcare system. The country is not yet able to rely on VR data to generate under-5 mortality rates (without adjustment). While UN and RMS estimates allow us to track under-5 mortality at a national level, the accuracy and completeness of existing information systems, especially the VR system, will need to be improved to ensure that accurate data are available at provincial and district levels. Acknowledgements. We thank Ms Koshen Moodley, who extracted the data from the Vital Registration database. Author contributions. NM and LB conceptualised the article. NM compiled figures and tables. LB and NM with assistance from PB wrote the article. YA reviewed the article. Funding. None. Conflicts of interest. None. 1. United Nations General Assembly. Transforming our World: The 2030 Agenda for Sustainable Development. New York: UN, 2015. http://www.un.org/ga/search/view_doc.asp?symbol=A/ RES/70/1&Lang=E (accessed 26 September 2017). 2. National Planning Commission (NPC). Our Future: Make it Work. National Development Plan 2030 Executive Summary. Pretoria: NPC, 2011. https://www.gov.za/sites/www.gov.za/files/Executive%20 Summary-NDP%202030%20-%20Our%20future%20-%20make%20it%20work.pdf (accessed 1 June 2016). 3. World Health Organization. SDG 3: Ensure healthy lives and promote wellbeing for all at all ages. Geneva: WHO, 2015. http://www.who.int/sdg/targets/en/ (accessed 30 September 2017). 4. Lo S, Horton R. Everyone counts – so count everyone. Lancet 2015;386(10001):1313-1314. https://doi. org/10.1016/S0140-6736(15)60305-1 5. Shibuya K, Gilmour S. Civil registration as a means to promote human security. Lancet 2015;386(10001):e14-e15. https://doi.org/10.1016/S0140-6736(15)60765-6 6. UNICEF, WHO, World Bank, United National Population Division. Levels and trends in child mortality 2014. New York: UNICEF, 2014. https://www.unicef.org/media/files/Levels_and_Trends_ in_Child_Mortality_2014.pdf. (accessed 30 September 2017). 7. Bradshaw D, Pillay-van Wyk V, Laubscher R, et al. Cause of death statistics for South Africa: Challenges and possibilities. Cape Town: Burden of Disease Research Unit, 2010. http://www.mrc. ac.za/bod/cause_death_statsSA.pdf (accessed 30 September 2017). 8. Dorrington RE, Bradshaw D, Laubscher R, Nannan N. Rapid mortality surveillance report 2014. Cape Town: South African Medical Research Council (SAMRC), 2015. www.mrc.ac.za/bod/ RapidMortalitySurveillanceReport2014.pdf (accessed 30 September 2017).

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

9. National Department of Health (NDoH). Health Data Advisory and Coordination Committee report. Pretoria: NDoH, 2012. http://www.health.gov.za/index.php/2014-03-17-09-09-38/reports/ category/100-2012rp# (accessed 30 September 2017). 10. Alkema L, New JR. Global estimation of child mortality using a Bayesian B-spline bias-reduction method. Ann Appl Stat 2014;8(4):2122-2149. https://doi.org.10.1214/14-AOAS768. 11. Joubert J, Rao C, Bradshaw D, Vos T, Lopez AD. Evaluating the Quality of National Mortality Statistics from Civil Registration in South Africa, 1997-2007. PLOS ONE 2013;8(5):e64592. https://doi. org/10.1371/journal.pone.0064592 12. Dorrington RE. Alternative South African Midâ&#x20AC;?year Estimates, 2013. CARe Monograph no 13. Cape Town: University of Cape Town, 2013. http://www.commerce.uct.ac.za/research_units/care/ monographs/monographs/mono13.pdf (accessed 15 November 2017). 13. National Department of Health, Statistics South Africa (StatsSA), SAMRC, ICF. South African Demographic and Health Survey 2016: Key Indicators. Pretoria: StatsSA, 2017. http://www.statssa. gov.za/publications/Report%2003-00-09/Report%2003-00-092016.pdf (accessed 30 September 2017). 14. Adewemimo A, Kalter HD, Perin J, Koffi AK, Quinley J, Black RE. Direct estimates of cause-specific mortality fractions and rates of under-five deaths in the northern and southern regions of Nigeria by verbal autopsy interview. PLOS ONE 2017;12(5):e0178129. https://doi.org/10.1371/journal. pone.0178129 15. Liu L, Kalter H, Chu Y, et al. Understanding Misclassification between Neonatal Deaths and Stillbirths: Empirical Evidence from Malawi. PLOS ONE 2016;11(12):e0168743. https://doi.org/10.1371/journal. pone.0168743 16. Statistics South Africa. Census 2011 Statistical release - P0301.4. Pretoria: StatsSA, 2012. https://www. statssa.gov.za/publications/P03014/P030142011.pdf 17. National Department of Health. First Triennial Report of the Committee on Morbidity and Mortality in Children Under 5 Years (CoMMiC) 2008 - 2010. Pretoria: NDoH, 2011. http://www.health.gov.za/ index.php/2014-03-17-09-09-38/reports/category/101-2011rp?download=195:first-triennial-report-ofthe-committee-on-morbidity-and-mortality-in-children-under-five-years. (accessed 12 August 2017). 18. National Department of Health. District Health Management Information System (DHMIS) Policy. Pretoria: NDoH, 2011.https://www.idealclinic.org.za/docs/policies/District%20Health%20 Management%20Information%20System%20Policy_2011.pdf (accessed 29 November 2017). 19. UNICEF. Under 5 mortality: Current Status and Progress. New York: UNICEF, 2017. https://data. unicef.org/topic/child-survival/under-five-mortality/ (accessed 30 September 2017).

S32

20. Rhoda N, S Velaphi S, Gebhardt G, Kauchali S, Barron P. Reducing Neonatal Deaths in South Africa: Progress and challenges. S Afr Med J 2018;108(3 Suppl 1):S9-S16. https//doi/org/10. 7196/SAMJ.2018. v108i3.12804 21. UNICEF, WHO, World Bank, United Nations Population Division. Levels and Trends in Child Mortality 2015. New York: UNICEF, 2015. http://www.who.int/maternal_child_adolescent/ documents/levels_trends_child_mortality_2015/en/ (accessed 4 July 2016). 22. Dorrington RE, Bradshaw D, Laubscher R, Nannan N. Rapid mortality surveillance report 2015. Cape Town: SAMRC, 2016. www.mrc.ac.za/bod/RapidMortalitySurveillanceReport2015.pdf (accessed 3 March 2017). 23. Nannan N, Dorrington R, Laubscher R, et al. Under-5 mortality statistics in South Africa: Shedding some light on the trends and causes 1997-2007. Cape Town: South African Medical Research Council, 2012. http://www.mrc.ac.za/bod/MortalityStatisticsSA.pdf (accessed 3 March 2017). 24. Bamford LJ, Barron P, Kauchali S, Dlamini NR. In-patient case fatality rates improvements in children under five: diarrhoeal disease, pneumonia and severe acute malnutrition. S Afr Med J 2018;108 (3 Suppl 1):S33-S37. https//doi/org/10.7196/SAMJ.2018.v108i3.12772 25. Nojilana B, Groenewald P, Bradshaw D, Reagon G. Quality of cause of death certification at an academic hospital in Cape Town, South Africa. S Afr Med J 2009;99(9):648-652. 26. Pillay-van Wyk V, Msemburi W, Laubscher R, et al. Mortality trends and differentials in South Africa from 1997 to 2012: Second National Burden of Disease Study. Lancet Glob Health 2016;4(9):e642-e653. https://doi.org/10.1016/S2214-109X(16)30113-9 27. Stephen CR. Saving Children 2012 - 2013: An Eighth Survey of Child Healthcare in South Africa. Pretoria: Tshepesa Press, 2016. http://www.childpip.org.za/images/stories/documents/saving_ children_2012-2013.pdf (accessed 26 September 2017). 28. Liu L, Oza S, Hogan D, et al. Global, regional, and national causes of under-5 mortality in 2000-15: An updated systematic analysis with implications for the Sustainable Development Goals. Lancet 2016;388(10063):2959-3086. https:// doi.org/10.1016/S0140-6736(16)31593-8 29. Sharkey AB, Chopra M, Jackson D, Winch PJ, Minkovitz CS. Pathways of care-seeking during fatal infant illnesses in under-resourced South African settings. Trans R Soc Trop Med Hyg 2012;106(2):110116. https://doi.org/10.1016/j.trstmh.2011.10.008

Accepted 11 December 2017.

March 2018, Vol. 108, (3 Suppl 1)

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

RESEARCH

Inpatient case fatality rates improvementsÂ in children under 5: Diarrhoeal disease,Â pneumonia and severe acute malnutrition L Bamford,1 FCPaed, DrPH; P Barron,2 BCom, FFCH; S Kauchali,1,3 FCPaed, MS, MPhil; N Dlamini,1 MMed (Paeds) National Department of Health, Pretoria, South Africa School of Public Health, University of the Witwatersrand, Johannesburg, South Africa 3 Department of Paediatrics and Child Health, College of Health Sciences, School of Clinical Medicine, University of KwaZulu-Natal, Durban, South Africa 1 2

Corresponding author: L Bamford (lesley.bamford@health.gov.za)

Data on the number of admissions and deaths in children aged under 5 years from diarrhoea, pneumonia and severe acute malnutrition are routinely collected through the District Health Information System. These data, and the associated case fatality rates, are available for all public sector hospitals in South Africa (SA), and can be compared over time, as well as across different settings. This article presents these data for the period 2011/12 - 2016/17. It reflects on the remarkable improvements in these case fatality rates, and the likely reasons for their declines across all provinces. The article concludes by identifying the actions that need to be taken to ensure that SA achieves the Sustainable Development Goal aim of ending preventable child deaths by 2030. S Afr Med J 2018;108(3 Suppl 1):S33-S37. DOI:10.7196/SAMJ.2018.v108i3.12772

Pneumonia, diarrhoea and malnutrition remain important causes of death in children aged under 5 years, both globally[1] and in South Africa (SA),[2] and contribute significantly to the under-5 mortality rate. In 2015, pneumonia and diarrhoea were the leading causes of under-5 deaths (excluding perinatal causes), accounting for 16% and 9%, respectively, of all under-5 deaths worldwide.[1] Data from the Second National Burden of Disease Study[2] show a similar picture in SA, with diarrhoea accounting for 16% and pneumonia for 12.3% of under-5 deaths. An additional 19.5% of deaths were due to HIV, and it is likely that many of these deaths were associated with pneumonia and diarrhoea. Diarrhoea, pneumonia and malnutrition coexist, with many children suffering from two or more of these conditions simultaneously. Almost half of all under-5 deaths are associated with undernutrition,[3] and children with severe acute malnutrition (SAM) are 6.3 times as likely to die from diarrhoea, and 8.7 times more likely to die from pneumonia, when compared with children with these diseases who are not malnourished.[4] Data from child mortality audits show that one-third of children under 5 years of age who died in SA hospitals had SAM, and a further 27% were underweight for their age.[5] Pneumonia, diarrhoea and SAM are preventable and treatable causes of child deaths. All three are diseases of poverty, and are closely associated with factors such as poor home environments, poor access to nutritious foods and lack of access to health services. Child survival interventions prevent deaths from these diseases by reducing the number of cases, decreasing their severity and ensuring that they are correctly managed. Key prevention strategies include optimal breastfeeding practices and adequate nutrition, immunisations, handwashing with soap and water and prevention of comorbidities (especially HIV infection). Provision of safe drinking water and basic sanitation is important to prevent diarrhoea, while reduced household air pollution is associated with a reduction in pneumonia.[6,7]

S33

Effective treatments are also available. Therefore, improving healthcare-seeking behaviour and effective case management at household and health-facility levels are important interventions for reducing deaths from these diseases.[6] The treatment of pneumonia depends on the child taking an appropriate course of antibiotics, with more severe cases requiring supplemental oxygen. Effective treatment of diarrhoeal disease rests on three key interventions: the administration of oral rehydration salt (ORS) solutions to prevent life-threatening dehydration; continued feeding; and zinc supplementation. Most treatment is thus provided at household and primary healthcare level, although access to hospital care is important in more severe cases, with rehydration usually requiring short-term parenteral infusion of fluids. The management of SAM is slightly more complicated, but correct case management based on nationally adapted World Health Organization guidelines[8] has been shown to reduce case fatality rates substantially.[9] A reduction in the number of deaths from these three diseases indicates improvements in the health and nutrition status of young children, and is an important outcome of child health and nutrition programmes implemented in the public health system at all levels, from community and household interventions through to the national level.

Methods

All public sector hospitals in SA are required to submit data to the District Health Information System (DHIS) on a monthly basis. These include data on the number of children aged under 5 years who are admitted with diarrhoea, pneumonia and SAM, as well as the number of deaths associated with these conditions. Case fatality rates for the three conditions are automatically calculated for each level of the health system (facility, district, provincial and national). For this study, data were extracted from the DHIS data files, which are stored in the National Department of Health, focusing on the data related to the three indicators of case fatality rates for diarrhoea,

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Table 1. Diarrhoea admissions, deaths and case fatality rates in children aged under 5 years, 2011/12 - 2016/17

1 041 (2.2) 873 (2.0)

Case fatality rate, %

6 20

1 500

1 000

6 4

500

Deaths Case fatality rate

7 /1

20 16

15 /

20 11

Deaths, n (CFR, %) 1 589 (13.1) 1 637 (12.7) 1 672 (11.3) 1 851 (11.6) 1 379 (8.9) 1 192 (8.0)

2 500

20 1

500

Admissions, N 12 107 12 904 14 842 15 911 15 515 14 940

Year 2011/12 2012/13* 2013/14 2014/15 2015/16 2016/17

/1 4

Table 3. SAM admissions, deaths and case fatality rates in children aged under 5 years, 2011/12 - 2016/17

1 000

Fig. 2. Pneumonia deaths and case fatality rates in children aged under 5 years, 2011/12 - 2016/17.

/1 2

Financial year

Deaths, n

1 500

2 000

Case fatality rate, %

500

Deaths Case fatality rate

20 11 /1 2 20 12 /1 20 3 13 /1 4 20 14 /1 20 5 15 /1 20 6 16 /1 7

Deaths, n

2 000

1 000

Case fatality rate, %

1 500

Deaths Case fatality rate

SAM = severe acute malnutrition; CFR = case fatality rate. *Western Cape data for 2012/13 are not available.

CFR = case fatality rate

2 500

2 000

47 640 44 120

4.5 4 3.5 3 2.5 2 1.5 1 0.5 0

2015/16 2016/17

Deaths, n (CFR, %) 1 541 (4.5) 1 526 (4.3) 1 769 (3.9) 1 513 (3.3)

2 500

Admissions, N 34 017 35 666 45 824 45 779

Deaths, n (CFR, %) 1 769 (4.1) 1 395 (3.8) 1 524 (3.5) 1 411 (2.9) 1 240 (2.3) 1 005 (2.0)

*Western Cape Province data for 2012/13 are not available.

20 1

Year 2011/12 2012/13 2013/14 2014/15

Admissions, N 43 138 36 431 43 402 48 365 53 290 50 439

Year 2011/12 2012/13* 2013/14 2014/15 2015/16 2016/17

The number of admissions and deaths of children under 5 years of age due to diarrhoea, pneumonia and SAM for the past 6 financial years in public health facilities are shown in Tables 1 to 3. Despite some minor variation, there is a relatively consistent pattern for all three diseases. Admissions for each disease increased between 2011/12 and 2015/16, and then decreased during 2016/17. In addition, the number of deaths, as well as the case fatality rates, declined consistently over the 6-year period. The number of admissions due to diarrhoea (Table 1 and Fig. 1) showed a steep rise in 2013/14. This can be attributed to the higher number of cases (and deaths) associated with rotavirus infection which occurred during that year.[10] The number of reported deaths declined by 43% (from 1 541 in 2011/12, to 873 in 2016/17), while the case fatality rate decreased from 4.5% in 2011/12, to 2% in 2016/17, representing a decline of 55.5%. DHIS data on admissions, deaths and case fatality rates for pneumonia are shown in Table 2 and Fig. 2. Overall deaths declined by 43% (from 1 769 in 2011/12 to 1 005 in 2016/17) over the 6-year period. The case fatality rate declined by 51.2%, from 4.1% in 2012/13 to 2.0% in 2016/17. Over the same time period, there was a general increase in the number of admissions, from 43 138 in 2011/12 to 50 439 in 2016/17 (although there was an anomaly in 2012/13, as Western Cape Province data for that year are not available). DHIS data on admissions, deaths and case fatality rates for SAM are shown in Table 3 and Fig. 3. The number of admissions of children with SAM increased consistently between 2011/12 and

Table 2. Pneumonia admissions, deaths and case fatality rates in children aged under 5 years, 2011/12 - 2016/17

Results

2014/15, before declining in the latter 2 years. The number of deaths showed a similar pattern, with the number of deaths increasing

Deaths, n

pneumonia and SAM for the financial years 2011/12 to 2016/17. These data were extracted into an Excel (Microsoft, USA) file, disaggregated down to provincial level and compiled into relevant tables and figures. There was no sampling and all the relevant data were used in the analysis.

Financial year

Fig. 1. Diarrhoea deaths and case fatality rates in children aged under 5 years, 2011/12 - 2016/17.

S34

Fig. 3. Severe acute malnutrition deaths and case fatality rates in children aged under 5 years, 2011/12 - 2016/17.

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

9 8 7 6 5 4 3 2 1 0

Financial year

9 8 7 6 5 4 3 2 1 0

20 1

6/ 1

6 15 /1

/1 5

20 14

13 /1

13 20 12 /

1/ 1 20 1

Province: Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga Northern Cape North West Western Cape

Case fatality rate, %

Fig. 4. Diarrhoea case fatality rates in children aged under 5 years per province, 2011/12 - 2016/17.

Financial year

Fig. 5. Pneumonia case fatality rates in children aged under 5 years per province, 2011/12 - 2016/17. Western Cape data for 2012/13 are not available. 9 8 7 6 5 4 3 2 1 0

17 16 / 20

15 /

5 /1

20 14

13 /

13 12 / 20

11 / 20

Province: Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga Northern Cape North West Western Cape

Case fatality rate, %

these rates are largely the results of a decrease in the numerator, i.e. the number of deaths, rather than a decrease in the denominator, i.e. the number of admissions, which have remained consistently high.

Limitations of DHIS data

While the quality of DHIS data has improved over time, ongoing efforts are required to ensure that data are complete and accurate. DHIS reporting is well-established in most district hospitals, and reporting by secondary, tertiary and central hospitals has also improved substantially in recent years. It is likely that most deaths from diarrhoea and pneumonia are recorded, as well as deaths associated with clinically apparent SAM. Collecting data on all admissions proved more challenging for some hospitals; thus increases in the number of admissions recorded over time may reflect either a true increase, or better reporting. Only deaths which occur in health facilities are reported. However, in the absence of reliable reporting of the cause of child deaths through vital registration,[11] DHIS data provide useful information on child mortality trends.

Discussion

/1 7

/1 6

/1 5

/1 3

12 20

11 20

/1 4

Province: Eastern Cape Free State Gauteng KwaZulu-Natal Limpopo Mpumalanga Northern Cape North West Western Cape

/1 2

Case fatality rate, %

annually to 2014/15, with a rapid decline of 36% occurring in the past 2 years. Apart from a small increase between 2013/14 and 2014/15, the SAM case fatality rate declined consistently, with an overall decline of 38.9% over the 6-year period. Provincial trends for the three diseases are shown in Figs 4 - 6. Case fatality rates declined in all provinces across all three diseases over the 6 years. The Western Cape has consistently reported case fatality rates far below those in other provinces. Across the board

Financial year

Fig. 6. Severe acute malnutrition (SAM) case fatality rates in children aged under 5 years per province, 2011/12 - 2016/17. Western Cape data for 2012/13 are not available.

S35

The number of admissions associated with each of the three diseases shows a fairly consistent trend â&#x20AC;&#x201C; increasing between 2011/12 through 2015/16 before declining in 2016/17. While this may reflect a real increase in the number of children with these diseases, it is more likely to reflect more complete reporting and/or improved access to inpatient care. Given the increased number of admissions, the decrease in the number of deaths associated with each of the diseases is particularly encouraging. Case fatality rates have consistently declined year on year for all three diseases (with the exception of SAM in 2014/15) over the 6Â years. It should be noted that case fatality rates are not corrected for case mix. The declines are therefore likely to reflect a combination of improved case management of sick children, and a different case mix, with children presenting to public health facilities both earlier in the course of their illness and/or with less severe illness. As shown in Figs 4 - 6, case fatality rates have declined in all provinces. The consistently low case fatality rates reported by the Western Cape reflect both the high number of admissions and low number of deaths reported by the provinceâ&#x20AC;&#x2122;s hospitals. There is a need to ensure that all child deaths are reported (especially by central hospitals), and to identify practices and other factors that contribute to the low case fatality rates with a view to (where possible) replicating these factors elsewhere. The decline in the number of child deaths associated with all three diseases can be attributed to a number of factors, with the relative contribution of the different factors being difficult to quantify. The most obvious reason is the dramatic decline in HIV prevalence in young children as a result of the successful scale-up of the prevention of mother-to-child transmission of HIV, with transmission rates declining from 8% in 2008 to 2.6% at 6 weeks in 2012/13,[12] with a further estimated decline to 1.5% in 2015.[13] As a result, fewer children are at risk of developing (and dying from) diarrhoea, pneumonia and SAM. Increased coverage of other key child survival interventions has also contributed to the reduction in deaths. Immunisation against rotavirus and pneumococcal diseases was introduced into the routine immunisation schedule in 2009, and coverage levels equivalent to those of more traditional vaccines were rapidly achieved.[14] Case-control studies demonstrated that immunisation with these two vaccines was associated with reductions in hospitalisation for rotavirus-associated

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

diarrhoea[15] and presumed bacterial pneumonia,[16] while invasive pneumococcal disease among children in SA fell substantially.[17] Rates of exclusive breastfeeding in infants under 6 months of age remain modest (32% in 2016,[18]) but have increased substantially from the 8% reported in 2003.[19] Non-breastfed children are 11 times more likely to die of diarrhoeal disease than exclusively breastfed children,[6] so improvements in the proportion of children who breastfeed can play a significant role in further declines in admissions and case fatality rates for all three diseases. Fortification of basic foodstuffs and routine supplementation have contributed to a decline in micronutrient deficiencies in children under 5 years of age. Vitamin A deficiency among children under 5 decreased from 63.6% in 2005 to 43.6% in 2012, while anaemia and iron deficiency anaemia in children under 5 decreased by 63.0% and 83.2%, respectively, during the same period.[20,21] Care seeking and home management of these diseases have also improved, with 88% of children with symptoms of pneumonia being taken to a health facility or practitioner, and 73% of children with diarrhoea receiving recommended home-made fluids, and 51% fluid made from an ORS packet.[18] The quality of case management at health facilities and hospitals is not routinely measured, and the role of improved case management in reducing child deaths is thus difficult to determine. However, the steady decline in case fatality rates suggests that improved case management in health facilities has contributed to the decrease in the number of deaths. Efforts to improve case management of children with diarrhoea, pneumonia and SAM through the implementation of standard treatment guidelines[8,22,23] are likely to have contributed to the decline in case fatality rates, as have efforts to improve clinical governance through appointment of District Specialist Support Teams.[24] The impact of improved access to social support, through the child support grant (CSG), and improved access to basic services, is difficult to measure. There have been across-the-board improvements in children’s access to basic services, although access remains far from universal (Table 4). Likewise, receipt of a social grant has been shown to promote positive nutritional, educational and health outcomes, especially in young children.[25] By 2014, close to 12 million children were receiving a CSG, with a further 470 015 children receiving a foster child grant and 131 040 children receiving a care dependency grant.[26]

Efforts to improve the quality of data

Current efforts to improve the quality of data are focused on ensuring that all hospitals collect admission and discharge data using standardised tools (including a standardised paediatric ward register), and that there is full alignment between the child mortality audit and DHIS data. More attention also needs to be paid to correctly identifying all children with SAM; it is likely that many uncomplicated cases are currently missed owing to failure to accurately measure and classify the nutritional status of children admitted with other diseases (especially diarrhoea and pneumonia). Better recognition of children with SAM and earlier appropriate case management should reduce the number of deaths associated with the condition. This, together with recognition of more uncomplicated cases (where mortality is lower), is likely to result in a further decline in the case fatality rate.

Efforts to further reduce deaths

While the reduction in the number of child deaths is encouraging, under-5 mortality rates remain higher in SA than in other uppermiddle-income countries such as Brazil, Mexico, Peru and China.[28] Deaths in the post-neonatal period account for a higher proportion

S36

Table 4. Access to key basic and social services, 2002 v. 2014[27] Proportion lacking service Children living in households where there is reported child hunger Children living in households without basic sanitation Children living in households without water on site Children living in overcrowded households Children living far away from their health facility

2002

2014

30.6

12.3

54.8

25.6

41.2 26.6 37.1

31.4 18.4 21.5

of deaths than in countries with comparable mortality rates, with diarrhoea and pneumonia accounting for a high proportion of these deaths. This suggests that further improvements can be anticipated, providing that access to the child survival interventions described earlier is fully scaled-up. The Western Cape figures also suggest that most other provinces can still make significant improvements in the reduction of hospital deaths from the three diseases. While significant progress has been made in reaching children with the package of child survival interventions, many gaps and inequities persist. Further gains are likely to depend on ensuring that these interventions reach all children, especially the most vulnerable groups. This will require: • Continued improvements in the reduction of mother-to-child transmission of HIV and a reduction in the HIV prevalence of pregnant women, as well as ensuring that all HIV-infected children are identified early, receive antiretroviral therapy and are virally suppressed. • Improvements in infant and young child feeding, especially increasing rates of exclusive breastfeeding until 6 months old, and improvement in complementary feeding practices in children aged 6 - 24 months. • Increasing coverage of preventive and promotive services, especially immunisation. • Earlier access to appropriate healthcare when sick, through improving household care-seeking behaviour and ensuring access to care. • Improved case management in health facilities, including primary healthcare facilities and hospitals. Early identification and correct management of children with growth faltering and moderate malnutrition is particularly important. • Addressing the social determinants of health, especially ensuring household food security, increasing access to basic services, particularly water and sanitation, and ensuring that all eligible children receive a child support or other grant. • Community case management of diarrhoea and pneumonia by community health workers has been shown to reduce under-5 mortality,[29] and consideration should be given to introducing this in parts of the country with high mortality rates, especially in remote rural areas where access to health services may be difficult.

Conclusion

The sustained decline in the number of deaths among children under 5 years of age in public sector hospitals from diarrhoea, pneumonia and SAM is encouraging. The majority of the remaining deaths associated with these diseases are preventable. Further improvements in coverage of key child health and nutrition interventions, accompanied by improvement in the quality of care children receive in public health facilities, will be required for SA to achieve the Sustainable Development Goal target of ending preventable child deaths by 2030.

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Acknowledgements. We thank Ms Ronelle Niit for extracting the data from the DHIS database. Author contributions. LB and PB conceptualised the article. LB wrote the first draft. PB, SK and ND reviewed and edited the article. Funding. None. Conflicts of interest. None. 1. United Nations Children’s Fund. UNICEF Data: Monitoring the Situation of Children and Women. New York: UNICEF, 2016. https://data.unicef.org/topic/child-survival/under-five-mortality/ (accessed 21 July 2017). 2. Msemburi W, Pillay-van Wyk V, Dorrington RE, et al. Second national burden of disease study for South Africa: Cause-of-death profile for South Africa, 1997 - 2012. Cape Town: South African Medical Research Council (SAMRC), 2016. https://www.mrc.ac.za/bod/SouthAfrica2012.pdf (accessed 12 July 2017). 3. Black RE, Victora CG, Walker SP, et al. Maternal and child undernutrition and overweight in lowincome and middle-income countries. Lancet 2013;382(9890):427-451. https://doi.org/10.1016/ S0140-6736(13)60937-X 4. Black RE, Allen LH, Bhutta ZA, et al. Maternal and child undernutrition: Global and regional exposures and health consequences. Lancet 2013;371(9608):243-260. https://doi.org/10.1016/S01406736(07)61690-0 5. Stephen CR. Saving Children 2012 - 2013: An eighth survey of child healthcare in South Africa. Pretoria: Tshepesa Press, 2016. http://www.childpip.org.za/images/stories/documents/saving_ children_2012-2013.pdf (accessed10 January 2017). 6. United Nations Children’s Fund. Pneumonia and Diarrhoea: Tackling the Deadliest Diseases for the World’s Poorest Children. New York: UNICEF, 2012. https://www.unicef.org/eapro/Pneumonia_and_ Diarrhoea_Report_2012.pdf (accessed 15 June 2013). 7. World Health Organization, United Nations Children’s Fund. Global Action Plan for Prevention and Control of Pneumonia. Geneva: WHO, 2009. https://www.whqlibdoc.who.int/hq/2009/WHO_FCH_ CAH_NCH_09.04_eng.pdf (accessed 15 June 2013). 8. National Department of Health, South Africa. Integrated Management of Children with Acute Malnutrition in South Africa: Operational guidelines. Pretoria: NDoH, 2015. 9. Ashworth A, Chopra M, McCoy D, et al. WHO guidelines for management of severe malnutrition in rural South African hospitals: Effect on case fatality and the influence of operational factors. Lancet 2004;363(9415):1110-1115. https://doi.org/10.1016/S0140-6736(04)15894-7 10. Page N, Mapuroma F, Seheri M. Rotavirus Surveillance Report, South Africa, 2013. Communicable Diseases Surveillance Bulletin 2014;12(4):108-113. National Institute of Communicable Diseases. http://nicd.ac.za/assets/files/Rotavirus%20Surveillance2013.pdf (accessed 12 July 2017). 11. Bamford L, McKerrow NH, Barron P, Aung Y. Child mortality in South Africa: Fewer deaths but better data are needed. S Afr Med J 2018;108(3 Suppl 1):S25-S32. https://doi.org/10.7196/SAMJ.2018. v108i3.12779 12. Goga AE, Jackson DJ, Singh M, Lombard C, for the SAPMCTE [South African Prevention of Mother to Child Transmission Evaluation] study group. Early (4 - 8 weeks postpartum) Population-level Effectiveness of WHO PMTCT Option A, South Africa, 2012 - 2013. Pretoria: South African Medical Research Council and National Department of Health of South Africa, 2014. http://www.mrc.ac.za/ healthsystems/SAPMTCTEReport2012.pdf (accessed 12 July 2017). 13. Joint United Nations Programme on HIV/AIDS. 2015 Progress Report on the Global Plan towards the Elimination of New HIV Infections among Children and Keeping their Mothers Alive. Geneva: UNAIDS, 2015. http://www.emtct-iatt.org/wp-content/uploads/2015/12/UNAIDS-2015ProgressReport-on-the-Global-Plan.pdf (accessed 12 July 2017). 14. Madhi SA, Bamford L, Ngcobo N. Effectiveness of pneumococcal conjugate vaccine and rotavirus vaccine introduction into the South African public immunisation programme. S Afr Med J 2014;104(3 Suppl 1):228-234. https://doi.org/10.7196/SAMJ.7597

S37

15. Groome MJ, Page N, Cortese M, et al. Effectiveness of monovalent human rotavirus vaccine against admission to hospital for acute rotavirus diarrhoea in South African children: A case-control study. Lancet Infect Dis 2014;14(11):1096-1104. https://.doi.org/10.1016/S1473-3099(14)70940-5 16. Madhi SA, Groome MJ, Zar H, et al. Effectiveness of pneumococcal conjugate vaccine against presumed bacterial pneumonia hospitalisation in HIV-uninfected South African children: A casecontrol study. Thorax 2015;70(12):1149-1155. https://doi/10.1136/thoraxjnl-2014-206593 17. Von Gottberg A, de Gouveia L , Tempia S, et al. Effects of vaccination on invasive pneumococcal disease in South Africa. New Engl J Med 2014;371(20):1889-1899. https://doi/10.1056/NEJMoa1401914 18. National Department of Health, Statistics South Africa, South African Medical Research Council, ICF. South African Demographic and Health Survey 2016: Key Indicators. Pretoria, South Africa and Rockville, Maryland, USA: NDoH, StatsSA, MRC, ICF, 2016. http://www.statssa.gov.za/publications/ Report%2003-00-09/Report%2003-00-092016.pdf (accessed 11 June 2017). 19. National Department of Health, South Africa, Medical Research Council, OrcMacro. South African Demographic and Health Survey 2003. Pretoria: NDoH, 2007. https://dhsprogram.com/pubs/pdf/ FR206/FR206.pdf (accessed 12 July 2017). 20. Shisana, O, Labadarious D, Rehle T, et al. South African National Health and Nutrition Examination Survey (SANHANES-1). Cape Town: HSRC Press, 2013. http://www.hsrc.ac.za/uploads/pageNews/72/ SANHANES-launch%20edition%20(online%20version).pdf (accessed 4 October 2013). 21. Labadarios D, Swart R, Maunder EM, et al. The National Food Consumption Survey – Fortification Baseline (NFCS-FB-1): South Africa, 2005. Pretoria: National Department of Health, 2007. 22. National Department of Health, South Africa. Standard Treatment Guidelines and Essential Medicines List. Pretoria: NDoH, 2013. http://www.health.gov.za/index.php/standard-treatment-guidelines-andessential-medicines-list?download=867:hospital-level-paeds-2013-edition-version-02-pdf (accessed 12 July 2017). 23. National Department of Health, South Africa. Pocketbook of Hospital Care for children: Management of Common Childhood Illnesses. Pretoria: NDoH, 2016. http://www.health.gov. za/index.php/hiv-aids-tb-and-maternal-and-child-health/category/160-child-and-schoolhealth?download=1720:pocketbook-of-hospital-care-for-children (accessed 12 July 2017). 24. Feucht U, Marshall C, Kauchali S, et al. Innovations in the clinical care of mothers and children in South Africa: The contribution of district clinical specialist teams. S Afr Med J 2018;108 (3 Suppl 1):S38-S43. https://doi.org/10.7196/SAMJ.2018.v108i3.12808 n 25. Department of Social Development, South African Social Security Agency, United Nations Children’s Fund. The South African Child Support Grant Impact Assessment: Evidence from a survey of children, adolescents and their households. Pretoria: UNICEF South Africa, 2012. https://www.unicef.org/ southafrica/SAF_resources_csg2012s.pdf (accessed 12 December 2016). 26. Delany A, Jehoma S, Lake L (eds). South African Child Gauge 2016. Cape Town: University of Cape Town, 2016. http://www.ci.uct.ac.za/sites/default/files/image_tool/images/367/Child_Gauge/2006/ Book_Child_Gauge_2016_lowres.pdf (accessed 12 December 2016). 27. Sadan M, Delany A. Social assistance for children: Looking back, thinking forward. In Delany A, Jehoma S, Lake L (eds). South African Child Gauge 2016. Cape Town: University of Cape Town, 2016. http://www.ci.uct.ac.za/sites/default/files/image_tool/images/367/Child_Gauge/2006/Book_Child_ Gauge_2016_lowres.pdf (accessed 12 December 2016). 28. United Nations Children’s Fund. Countdown to 2015 Maternal, Newborn and Child Survival: A Decade of Tracking Progress for Maternal, Newborn and Child Survival: The 2015 Report. New York: UNICEF, 2015. http://countdown2030.org/documents/2015Report/Countdown_to_2015_final_ report.pdf (accessed 11 July 2016). 29. World Health Organization, United Nations Children’s Fund. Joint Statement Integrated Community Case Management. An equity-focused strategy to improve access to essential treatment services for children. New York: UNICEF, 2012. http://www.who.int/maternal_child_adolescent/documents/ statement_child_services_access_whounicef.pdf (accessed 11 July 2016).

Accepted 26 September 2017.

March 2018, Vol. 108, (3 Suppl 1)

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

RESEARCH

Innovations in the clinical care of mothers and children in South Africa: The contribution of district clinical specialist teams U Feucht,1,2,3 PhD; C Marshall,3,4 MMed (Community Medicine); S Kauchali,4,5 MPhil; P Barron,6 FFCH (SA); L Slavin,7 FCPaed (SA); S Bhardwaj,7 MD; Y Pillay,4 PhD Department of Paediatrics, School of Medicine, Faculty of Health Sciences, University of Pretoria, South Africa Tshwane District Health Services, Gauteng Department of Health, South Africa 3 Research Centre for Maternal, Fetal, Newborn and Child Health Care Strategies, Faculty of Health Sciences, University of Pretoria, South Africa 4 National Department of Health, Pretoria, South Africa 5 Department of Paediatrics and Child Health, Nelson Mandela School of Medicine, College of Health Sciences, University of KwaZulu-Natal, Durban, South Africa 6 School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa 7 United Nationa Children’s Fund (UNICEF), Pretoria, South Africa 1 2

Corresponding author: U Feucht (ute.feucht@up.ac.za)

The contribution of the District Clinical Specialist Teams (DCSTs) to improving maternal and child health outcomes in South Africa, through strengthening the four pillars of clinical governance, is reflected in innovative work presented at a ‘Promising Practices’ symposium and at various conferences. Of the 24 identified DCST innovations, 21% reflected the clinical effectiveness pillar, 17% clinical risk management, 41% staff development, and 21% user-related considerations. In order to ensure scale-up, the submitted best practices/ innovations were reviewed using the World Health Organization quality standards and ExpandNet parameters for likely scalability. Here we describe one case study from each pillar, illustrating the contribution of the innovations to improved patient outcomes. The development and scale-up of innovations needs to be institutionalised and must include effective support and action from the relevant health managers. S Afr Med J 2018;108(3 Suppl 1):S38-S43. DOI:10.7196/SAMJ.2018.v108i3.12808

In 2010 concern was raised that South Africa (SA) was falling short in achieving the Millennium Development Goals. Despite being a middle-income country, high levels of maternal and child mortality persisted, notwithstanding progressive improvements in the inherited inequitable health system.[1] As a major contributor to the high mortality rates, HIV was now being addressed in line with international best practice. However, poor quality of clinical care remained a significant challenge towards achieving improved outcomes, despite relatively high government expenditure.[2] The District Clinical Specialist Team (DCST) is an initiative established in SA in 2011 to improve the maternal, neonatal, child and women’s health (MNCWH) outcomes through enhanced clinical governance.[3] This is defined as a ‘framework through which … organisations are accountable for continually improving the quality of their services, and safeguarding high standards of care by creating an environment in which excellence in clinical care will flourish’.[4] Four pillars form the basic framework for the work of the DCSTs in the SA clinical

Pillar 1

Pillar 2

Pillar 3

Pillar 4

Clinical effectiveness

Clinical risk management

Reliable implementation of evidence-based protocols for every patient, every time

Providing safe care without harm through analysis of and learning from adverse events and deaths

Professional development and management

Create demand and improve accountability for MNCWH

Staff provision, development and adequacy through professional development and management

User perspectives and community-based factors influencing use of services and behaviour of patients and caregivers

Fig. 1. Pillars of clinical governance for the District Clinical Specialist Teams.[5] (MNCWH = maternal, neonatal, child and women’s health.)

governance model (Fig. 1). DCSTs were recruited at the primary care level in all districts. Teams consist of three doctor-nurse pairs (obstetrician and advanced midwife, paediatrician and paediatric nurse, family physician and primary healthcare nurse) together with an anaesthetist. DCSTs are in a unique position to develop best practices or innovations.[6] The main aim of this paper is to highlight the contribution of the DCSTs to improving maternal and child health outcomes in SA through strengthening the four pillars of clinical governance, as illustrated by four case studies selected from among 24 best practices/innovations submitted by DCST members. These are

S38

March 2018, Vol. 108, (3 Suppl 1)

reviewed against standards of quality and predictors of successful scale-up, in addition to outlining likely paths to successful implementation. The paper concludes with recommendations on the processes that are essential to the implementation and scale-up of the innovations.

Methods

Identification and categorisation of innovation

To encourage the sharing of innovations, the National Department of Health (NDoH) sent a request for concepts to all DCST members. Twenty of the 42 responses received were selected by a panel and

RESEARCH

subsequently presented at a Promising Practices symposium in 2016. Four additional practices were later identified through conference presentations. Innovations were then categorised into the four pillars of clinical governance, based on their primary purpose (Table 1).

Criteria for review

Innovations were then reviewed against the World Health Organization (WHO) quality standards and WHO attributes for potentially scalable innovations. These are shown in Table 2.[7,8]

Results

Quality standards

All innovations directly or indirectly supported the delivery of evidence-based care (Standard 1), reflecting the initial selection process. They were a response to a problem, based on actionable information (Standard 2), from the routine information system and from more granular data obtained from clinical audits. Referrals and continuity of care were mentioned in 11 out of 24 submissions (Standard 3), while communication with patients (n=9), respectful care (n=3) and provision of emotional support (n=3) were mentioned less frequently (Standards 4 - 6). Staff competence and organisation (Standard 7) was a component of all practices, while essential health facility infrastructure (Standard 8) was mentioned in 11 cases. Most innovations also highlighted the importance of a referral chain and of sector co-ordination.

Scale-up potential

The innovations reflected credible and feasible approaches to important problems, therefore demonstrating positive predictors of scalability. Documentation of impact, however, was variable at the time of presentation. Identification of the primary user was of importance, as the authors considered those innovations with clear benefit for clinicians (the primary focus group of the DCSTs) to be more scalable than those necessitating management buy-in and action, given the variability in management authority and accountability in different settings. In about a third of cases, the innovation was fairly complex, requiring expert support, while a few needed actual policy changes.

Case studies illustrating the four pillars of clinical governance

Four case studies were chosen to be presented in this article as examples of the innovations reviewed, as they illustrate the four pillars and meet most of the relevant quality standards and predictors for scalability. Pillar 1: Reliable implementation of protocols for gestational hypertension Problem statement and response

High numbers of maternal deaths due to hypertension were recorded in the Amathole District in 2013. At 58% of deaths (n=7/12 deaths), this was three times the national average of 15%.[9] Contributory factors, as identified by the local DCST, included: • High risk antenatal care (ANC) for gestational hypertension was available at district hospitals, but lack of early recognition, inappropriate classification, non-adherence to protocols, infrequent patient follow-ups and high defaulter rates occurred. • Eclampsia and severe pre-eclampsia cases were inappropriately transferred from clinics to district hospitals, instead of directly to tertiary hospitals, by ambulance crews who also lacked the necessary skills to ensure safe transport. • Challenges were experienced in the mentoring of clinicians, as the patient-held maternity case records were not available and problems were only identified at post-natal audits. The DCST conducted workshops to train clinic nurses and hospital maternity staff on the clinical management of hypertension in pregnancy. Doctor-nurse teams were identified as champions at high-risk ANC clinics to audit hypertension management and inform the DCST of mentoring and support needs. Quality assurance managers conducted monthly scoring of antenatal records at delivery sites based on DCST guidance, with feedback given to clinics at review meetings. The referral system for patients with severe gestational hypertension was revised to fast-track access to level 3 hospital care for seriously ill patients, bypassing district hospitals. The central innovation was a facility-held tracer card developed to follow up and monitor fidelity to clinical guidelines on gestational

Table 1. Number and percentages of innovations per clinical governance pillar Clinical governance pillar Clinical effectiveness: Use of evidence-based protocols Clinical risk management: Safe care without harm Professional development and management Create demand and improve accountability for MNCWH

From DCST symposium 4 4 9 3

Other source 1 0 1 2

Number of innovations, n (%) 5 (21) 4 (17) 10 (41) 5 (21)

DCST = District Clinical Specialist Teams; MNCWH = maternal, neonatal, child and women’s health.

Table 2. Criteria used to review innovations[7,8] Quality standards 1. Evidence-based care and management of complications 2. Health information systems enabling the use of data for action 3. Application of referral systems 4. Effective communication responding to needs 5. Respectful clinical care 6. Provision of emotional support sensitive to needs 7. Competent staff 8. Health facility with optimal physical environment

S39

Predictors of scalability 1. Credible/based on sound research 2. Observable/potential users can see results 3. Relevant to addressing persistent or pertinent problems 4. Advantage over existing practices (potential users perceive benefits as higher than costs of implementation) 5. Easy to understand and implement 6. Compatible with user values, norms and practices and aligned with national programmes 7. Testable

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

hypertension. Concomitant training of clinic supervisors, MNCWH programme managers and maternity operational managers was done. Scale-up of the checklist was done with a focus on healthcare facilities where deliveries took place. Scores obtained during audits were then used to build dashboards, resulting in structured feedback to clinicians and managers. Outcomes

As part of the intervention, local audits were conducted to gain an in-depth understanding of the clinical problems and to measure the impact of the intervention. The institutional Maternal Mortality Rate (iMMR) dropped from 83/100 000 (2013) to 44/100 000 (2015) after the intervention, with maternal deaths due to hypertension reducing from 58% (7/12 deaths; 2013) to 33% (2/6 deaths; 2014) to 20% (1/5 deaths; 2015). Maternal deaths at the level of the tertiary hospital reduced from 18 deaths (2014) to 10 deaths (2015) to 9 deaths (2016), indicating that the rerouting of ambulances transporting seriously ill patients did not lead to an increased mortality at tertiary level. Review findings from the Perinatal Problem Identification Program (PPIP) indicate that perinatal deaths due to hypertension reduced from 16.4% (2013) to 6.9 % (2014), with stillbirth rates improving from 17/1 000 (2013) to 13.3/1 000 (2014). The improvements were sustained during 2015 (14.1/1 000); however, a slight increase in stillbirths in one of the district hospitals was noted, triggering more intensified clinical oversight and training.[10] Implications and scalability

Improving clinical care and systems for the management of gestational hypertension has great potential impact if implemented at scale. These innovations meet the quality standards relating to the implementation of evidence-based protocols, referral systems and the use of actionable information, as well as the human resources standard. This initiative is a good candidate for future scale-up. Clinical audit and health system strengthening are part of DCSTsâ&#x20AC;&#x2122; scope and capacity. Action by managers is, however, critical for effective implementation. Universal introduction of the checklist would require further refining and testing to determine benefits and processes in different settings before a policy decision is made. Pillar 2: Managing clinical risk across a catchment area to reduce maternal deaths Problem statement and response

The Saving Mothers Report (2011 - 2013)[11] noted high maternal mortality ratios (MMRs) in the eThekwini Metro: 197.6/100 000 live births, compared with 171.1/100 000 for KwaZulu-Natal Province and 158.3/100 000 for SA as a whole. Approximately 5 000 deliveries per month occurred across the Metro, with 101 deaths in 2014. However, the DCST found that staff in the district had no consolidated understanding of the reasons for the maternal deaths, including main causes and avoidable factors identified during mortality review meetings. Importantly, regional and tertiary hospitals did not report their deaths to district management or the DCST. Of relevance is the fact that there are two tertiary, four regional and four districtlevel hospitals in the district, as well as over 100 primary healthcare clinics and 16 midwife obstetric units, emphasising the need for a coordinated approach to maternity care. The district manager, assisted by the DCST, convened a meeting attended by all hospital and nurse managers and heads of obstetric departments. The need to conduct maternal mortality meetings within 72 hours of each death was emphasised. Since April 2015, the DCST produces a quarterly metro-wide report covering all levels of

S40

care, including regional and tertiary hospitals, across the catchment area. It includes the number of deaths and relevant mortality rates, referrals between facilities and from other districts, and information on causes of death and avoidable factors. The report is presented at the district quarterly Perinatal Mortality and Morbidity Meeting and guides the quality improvement plans which address the main avoidable factors and other associated problems. Examples include improvement of clinical care at primary healthcare level, implementing a dedicated form to capture relevant HIV information in maternity charts, and auditing of all cases of ruptured uteri. Outcomes

As part of the intervention, the DCST scrutinised local audit data to identify challenges and measure the impact of quality improvement interventions. Major discrepancies between the audit data and data from the District Health Information System (DHIS) were noted, which were likely linked to underreporting on the DHIS. During 2014 there were 101 recorded institutional maternal deaths (iMMR of 173.8/100 000 live births), with scanty details available on causes of deaths and avoidable factors. By 2016, maternal mortality reporting had greatly improved, with 71 deaths counted and an iMMR of 120/100Â 000. The level of healthcare at which the deaths were occurring was now well mapped: 29% at tertiary hospitals, 51% at regional hospitals, 14% at district hospitals and 6% at clinics. There were fewer unknown causes of death, and medical and surgical disorders were now the most common cause of death (24%), rather than primary obstetric causes, with non-pregnancy related infections and hypertension contributing 18.3% each to the mortalities. (Audit data supplied by T Ibrahim, eThekwini DCST Obstetrician).[10] Since 2015, seven maternal mortality reports have been generated, and death reporting systems have been streamlined and improved to reduce under-reporting. Facilities have a more than 90% fidelity with maternal mortality reporting within 72 hours, with top management and DCST present at reviews. Action plans are then drawn up, and feedback to level 1 facilities is done by the DCST, if necessary. Responsibility and accountability for the implementation of action plans, however, remains a constant challenge. Implications and scalability

Learning from every maternal death to improve the management of clinical risk through specific and local problem identification across a catchment area has the potential to rapidly reduce maternal mortality. This innovation is focused on strengthening the health system across an entire catchment or referral area to enable implementation of protocols. This addresses the first three quality standards as well as those related to staff practice and essential resources. Dissemination of this process through other DCSTs could easily be achieved. No additional resources are needed. However, the key requirement of management accountability for each death, in order to ensure effective action, is the critical success factor. Pillar 3: Staff development through effective paediatric clinical skills training Problem statement and response

Review of child deaths using the Child Problem Identification Programme (CHIP) in the Ugu District revealed local gaps and weaknesses in paediatric district hospital care. After further investigations, the DCST concluded that the knowledge base was lacking, as the same modifiable factors came up repeatedly. Staffing constraints were impacting on monthly outreach visits by the regional hospital paediatricians and on training opportunities for district hospital doctors at the regional hospital. Staff training courses

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

on all relevant child health programmes were being conducted, often requiring extended (3 - 11 days) attendance. However, the DCST assessed these to have limited impact, with suboptimal practices persisting and limited cascading of knowledge to colleagues who did not have the opportunity to attend the trainings. In response to the identified gaps, the DCST paediatrician developed an on-site customised modular training programme. Training modules were delivered to clinical staff and hospital management through short, 2-hour training sessions run four times during a day-long visit, permitting all staff (including night shift and after-hours non-paediatric doctors) to attend without significantly compromising patient care. Sessions included a theoretical component and a focus on practical skills and procedures, for example intubation. The setting was small group teaching, with repetition and assessment in an interactive environment, where questions were asked and weaknesses identified. Future suggestions were to include local nurses and doctors in the process of tailoring the training content, to email theory prior to sessions, to use e-learning platforms, and to conduct baseline knowledge tests. Within the district health services, the DCSTâ&#x20AC;&#x2122;s role needs to include such training as a core business. Major challenges remain with non-permanent staff such as sessional doctors, knowledge gaps in foreign-trained doctors, as well as the continued high turnover of staff, and structured follow-up of clinical practice after completion of training modules is needed. Outcomes

Staff members embraced the training approach and contributed to the topics for the training agenda. Clinical patient management improved, as observed during discussions at local CHIP meetings. Onsite DCST support ward round visits showed entrenchment of evidence-based paediatric practice, with improved understanding of the essentials of paediatric care by the entire medical team at the hospital, with improved oversight by the DCST and teamwork within the district. As illustrated, improved outcomes in terms of childhood mortality were achieved in one of the district hospitals after intensive modular training was conducted. CHIP data show that the hospital had 12 childhood deaths during the first 6 months of 2016, which was reduced to 7 childhood deaths during the same period in 2017. The in-hospital modifiable factors identified on audit also decreased from 10 to 1 during the corresponding periods. A decrease in number and better quality of intensive care referrals was also noted. (Audit data supplied by R Naidoo, Ugu DCST Paediatrician; Ugu District CHIP, personal communication.) Implications and scalability

Effectively targeted innovations to improve clinical skills of staff at facility level are critical, with the local implementers noting that change was affected by focussing on specific problems and not just common clinical conditions. Most of the quality standards are met. This innovation falls within the DCSTsâ&#x20AC;&#x2122; scope and could easily be replicated, given the local availability of a DCST paediatrician. Required changes include the review of a narrow interpretation of the DCST job description, restricting time spent on teaching. Additionally, DCSTs need updates on best clinical practice as well as teaching skills, with assistance given in terms of modern teaching methods, and relevant equipment and internet connectivity being of importance. The use of e-learning platforms is a potential success factor, although the importance of face-to-face contact remains in order to ensure effective motivation and support of local staff members. At a health system level, deficits in clinical skills of doctors and nurses point towards a serious problem in the level of basic training which

S41

requires interventions. There are also issues in the prerequisites for contracting of sessional and foreign-trained doctors, which require involvement of health professional bodies. Pillar 4: Addressing the determinants of severe acute malnutrition (SAM) Problem statement and response

An increase in SAM cases in the Tshwane Metro was noted in 2015, where the staff in one sub-district were particularly concerned about the number of admissions and deaths in SAM children, as well as SAM readmissions after discharge. The need for a multi-disciplinary targeted intervention to improve outcomes resulted in the formation of a sub-district Malnutrition Working Group, established in April 2016 under the leadership of the DCST and the sub-district dietician, using the local district hospital as the focal point. Team members included staff from the district hospital (paediatric services, emergency department, nutrition, social worker, HIV and tuberculosis services), district health services (nutrition, DCST, primary healthcare facilities, ward-based outreach teams, health promotion services, HIV and MCWNH programmes), social services as well as colleagues from the neighbouring province. Geospatial mapping of SAM admissions was done to pinpoint community hotspots for interventions, and contributory factors were identified using household questionnaires. Additionally, the team considered case studies and participated in morbidity and mortality meetings, with discussions on avoidable factors contributing to SAM deaths. This led to the development of locally adapted guidelines and protocols to address specific problems. In addition to meetings, the team used now widely available technology (email and WhatsApp groups) to facilitate patient referrals and sharing of information. Challenges identified included inadequate referral systems and acute care in the hospital emergency department, in-hospital constraints to care (overcrowding) and nutritional supplementation stock-outs, especially for cross-boundary patients. Untreated HIV infection and HIV exposure were identified as contributors to malnutrition. The working groupâ&#x20AC;&#x2122;s leaders ensured the coordination of the diverse group to maintain momentum in implementing improvements. The involvement of the district management team was also important, especially in engaging on cross-boundary issues and with other relevant government departments. Outcomes

Achievements hinge around the establishment of a communication network of relevant staff members in the sub-district, as well as the neighbouring province. Emergency clinical care of SAM children improved, with better triaging and initial patient management in the emergency department. Sub-district malnutrition guidelines were developed, printed and distributed, and relevant on-site training done, including training of the ward-based outreach teams. Lost-tofollow-up SAM children remained challenging, therefore emphasis was placed on building linkages between hospital staff, primary healthcare clinics, community care workers and social services. Monitoring and evaluation included collecting local granular data on SAM admissions and case fatality rates, and also strengthening of the DHIS processes. The nutrition coordinator now attends monthly sub-district data verification meetings to ensure submission of clinically verified data. The impact of the clinical intervention, which started at sub-district level and expanded to improvement of SAM care in the entire district, on improvement of data quality is shown in Table 3. In 2014/15, only five hospitals submitted SAM data to the DHIS. This expanded to all nine hospitals in 2016/17. As a result of improved data quality, there was a paradoxical increase in

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Table 3. Inpatient SAM admissions and case fatality rates within healthcare facilities in the Tshwane District*

Facility name Jubilee Hospital Odi Hospital Pretoria West Hospital Tshwane District Hospital Bronkhorstspruit Hospital Mamelodi Hospital Kalafong Hospital Dr George Mukhari Hospital Steve Biko Academic Hospital Total

Level of care

District

2014/5 Inpatient Inpatient SAM SAM admissions, n deaths, n No data 90 4

4.4

2015/6 Inpatient Inpatient SAM SAM admissions, n deaths, n 60 3 63 5

iCFR (%) 5.0 7.9

Inpatient SAM admissions, n 113 64

iCFR (%)

7 25

10.0

12.0

17.0

1.2

No data Regional

No data 16.6

No data Tertiary/ central

224

401

No data 33

16.2

No data

2016/7 Inpatient SAM deaths, n 19 2

iCFR (%) 16.8 3.1

7.6

192

6.3

150

4.7

8.2

6 393

2 29

33.0 7.4

41 534

4 41

9.8 7.7

DHIS= district health information system; SAM= severe acute malnutrition; iCFR= institutional case fatality rate. *Data obtained from the District Health Information System.

the number of SAM admissions. Jubilee Hospital, the district hospital used as the node for the sub-district malnutrition group, progressed from having no SAM admissions recorded on DHIS to being the district hospital with the highest numbers in terms of SAM admissions and deaths. This highlights the critical role of DCST onsite support visits to identify clinical problems, without solely relying on DHIS data for decision making and planning of interventions. Implications and scalability

Fragmentation of health services was noted as a big risk to holistic patient care. This innovation met most of the quality standards mentioned above. It did not require anything outside of routine clinical care. The improvement in quality was achieved through greatly enhanced multi-disciplinary team work using a subdistrict approach. Local replication of the model has already started in two additional sub-districts in the Tshwane District, with more to follow. However, wider scale-up would require a specific effort to entrench a collaborative work ethic, with clear communication and buy-in from stakeholders in different settings.

Conclusion and recommendations

DCSTs are generating innovations focused on key service delivery improvements, with notable outputs, and with documented reductions in institutional deaths and fatality rates over time, although in confined areas. Although clearly multi-factorial, with concurrent implementation of a range of evidence-based interventions to achieve these reductions, the improved processes and skills that these DCST innovations introduced are obvious enabling success factors.[12] The impact of the clinical interventions on the DHIS data quality is also evident, and highly beneficial, as this has the potential to lead to better decision making within the health system based on accurate data. Encouraging more innovation and problem-solving skills is critical for DCSTs and for health professionals in general. Some

S42

identified weaknesses in the currently reviewed initiatives include inadequate problem formulation and measurement of outputs and outcomes. Quality standards that were not addressed, including respect and dignity, communication and emotional support and, to a certain extent, in the management of essential physical resources, could be specifically targeted. Adoption and implementation of innovations at scale cannot be assumed, and without a pathway to testing and scale-up, the impact will remain localised, even though the potential is clearly evident. To take advantage of innovations developed by the DCSTs, an agile and effective scale-up and dissemination approach is needed, based on the attributes of scalability and the characteristics of the SA health service itself. The team leads of the featured innovations, described the critical importance of ensuring managerial approval and support as a prerequisite for implementation, even at small scale. Moving forward, the most promising innovations, with clear and encouraging results at local level, need to be supported towards scale-up. This scale-up strategy must include reflection on the barriers and drivers of actual implementation, and draw on the insights from the field of implementation science, which has identified competency/ability, organisation/systems and leadership as drivers of effective implementation.[13] Encouraging the development and identification of innovations to successfully address the challenges facing the SA health system, and ensuring their scale-up, should become an accepted norm. Experienced and well-trained health workers, like the DCSTs, are in an ideal situation to contribute towards this process. Acknowledgements. The authors thank all DCST staff members who contributed by supplying case studies and innovations. For the featured case studies, the following persons and districts are acknowledged: Dr Sibongile Mandondo, DCST Obstetrician, Amathole District; Dr Tasnim Ibrahim, DCST Obstetrician, Ethekwini District; Dr Romola

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Naidoo, DCST Paediatrician, Ugu District; Ms Danielle Reyneke, subdistrict dietician; and Prof Ute Feucht, DCST Paediatrician, Tshwane District. Author contributions. Conceptualisation: all authors; data review and analysis: CM, UF, SK, PB; writing: UF, CM, SK, PB. Funding. None. Conflicts of interest. None. 1. United Nations (UN). The Millennium Development Goals Report 2010. New York: UN, 2010. http:// www.un.org/millenniumgoals/pdf/MDG%20Report%202010%20En%20r15%20-low%20res%20 20100615%20-.pdf (accessed 16 August 2017). 2. Naledi T, Barron P, Schneider H. Primary healthcare in SA since 1994 and implications of the new vision for PHC re-engineering. S Afr Health Rev 2011:17-28. 3. National Department of Health (NDoH). District Clinical Specialist Teams in South Africa. Ministerial Task Team Report, Pretoria: NDoH, 2011. http://www.health.gov.za/index.php/2014-08-15-12-55-04/ category/100-2012rp?download=186:district-clinical-specialist-teams-in-south-africa-ministerialtask-team-report (accessed 16 August 2017). 4. Scally G, Donaldson LJ. Clinical governance and the drive for quality improvement in the new NHS in England. BMJ 1998;317:61-65. https://doi.org/10.1136/bmj.317.7150.61 5. NDoH. Handbook for District Clinical Specialist Teams. Pretoria: NDoH, 2014. https://www.health-e. org.za/wp-content/uploads/2015/06/Handbook-for-DCSTs.pdf (accessed 16 August 2017).

S43

6. Voce A, Bhana R, Monticelli F, et al. District Clinical Specialist Teams. S Afr Health Rev 2014:45-58. 7. World Health Organization (WHO). Standards for Improving Quality of Maternal and Newborn Care in Health Facilities. Geneva: WHO, 2016. http://www.who.int/maternal_child_adolescent/documents/ improving-maternal-newborn-care-quality/en/ (accessed 16 August 2017). 8. World Health Organization (WHO). ExpandNet. Nine Steps for Developing a Scaling-up Strategy. Geneva: WHO, 2010. http://expandnet.net/PDFs/ExpandNet-WHO%20Nine%20Step%20Guide%20 published.pdf (accessed 16 August 2017). 9. National Committee for Confidential Enquiry into Maternal Deaths. Saving Mothers 2011 - 2013: Sixth report on the Confidential Enquiries into Maternal Deaths in South Africa: Executive Summary. http://www.kznhealth.gov.za/mcwh/Maternal/Saving-Mothers-2011-2013-Executive-Summary.pdf (accessed 16 August 2017). 10. Priorities in Perinatal Care Association of South Africa. Perinatal priorities: Proceedings 2017. https:// www.perinatalpriorities.co.za/proceedings-database/ (accessed 25 September 2017). 11. National Committee for Confidential Enquiry into Maternal Deaths. Saving Mothers 2011-2013: Sixth Report on the Confidential Enquiries into Maternal Deaths in South Africa: Short Report. https:// www.health-e.org.za/wp-content/uploads/2016/05/Saving-Mothers-2011-2013-short-report.pdf (accessed 16 August 2017). 12. Oboirien KO, Harris B, Eyles J, et al. Understanding roles, enablers and challenges of District Clinical Specialist Teams in strengthening primary healthcare in South Africa. S Afr Health Rev 2015:45-56. 13. Wutzke S, Benton M, Verma R. Towards the implementation of large scale innovations in complex healthcare systems: Views of managers and frontline personnel. BMC Res Notes 2016;9:327-331. https://doi.org/10.1186/s13104-016-2133-0.

Accepted 26 September 2017.

March 2018, Vol. 108, (3 Suppl 1)

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

RESEARCH

Implementation of strategies to improve programme effectiveness lead to an improvement in maternal and child health outcomes in South Africa S Bhardwaj,1 MD, MPH; R Pattinson,2 FCOG (SA), PhD; S Kauchali,3,4 FCPaed (SA); N Dlamini,3 MD, MMed; C Marshall,3 FCPaed(SA); M van der Merwe,5 RD (SA), MSc (Nutr); P Barron,6 BCom, FFCH (SA) United Nations Children’s Fund (UNICEF), Pretoria, South Africa South African Medical Research Council and University of Pretoria, Pretoria, South Africa 3 National Department of Health, Pretoria, South Africa 4 Department of Paediatrics and Child Health, University of KwaZulu-Natal, Durban, South Africa 5 Mpumalanga Department of Health, Nelspruit, South Africa 6 School of Public Health, University of the Witwatersrand, Johannesburg, South Africa 1 2

Corresponding author: S Bhardwaj (sbhardwaj@unicef.org)

Over the last few years, the South African (SA) National Department of Health has led the design, implementation and monitoring of key strategies to improve maternal, newborn, and child health outcomes. The strategies have resulted in the reduction of maternal, under-5 and infant mortality rates in SA. Here, we examine the strategies using a framework to understand the determinants and actions that influence the quality of care, coverage and resultant programme effectiveness in saving the lives of mothers and children. Three case studies from the field, namely, the Essential Steps In Managing Obstetric Emergencies, undernutrition in young children and breastfeeding are illustrated through applying the framework and demonstrating how improved quality of care and coverage achieved better health outcomes for pregnant women and children under 5 years of age. These case studies highlight different aspects of the framework, including improving healthcare workers’ skills, ensuring implementation of standard protocols and strengthening management accountability within facilities and across the districts. We also highlight how these aspects collectively improved the overall programme impact. The paper concludes that the framework would be strengthened through addition of critical cross-cutting aspects related to targeting and sustainability of actions. S Afr Med J 2018;108(3 Suppl 1):S44-S49. DOI:10.7196/SAMJ.2018.v108i3.12812

In order to have an impact at the population level, key evidence-based interventions should reach high population coverage levels, ensuring that they reach all those in need, and be delivered at high quality, meeting the standards of care.[1] The World Health Organization (WHO) defines quality of care as ‘the extent to which healthcare services provided to individuals and patient populations improve desired health outcomes’. In order to achieve this, healthcare must be safe, effective, timely, efficient, equitable and people-centred.[2] Quality of care is also a key component of the right to health and the route to equity and dignity for women and children, and it is essential to deliver healthcare services that meet the quality criteria.[3] Improving universal health coverage is also influenced by the accessibility, affordability and utilisation of the services. Fig. 1 outlines the factors that influence programme effectiveness, including the stakeholders (individuals, the community), the enablers (healthcare managers) and the doers (healthcare providers). The framework also highlights how the various actions that influence coverage and quality of care, e.g. knowledge and the use and accessibility of the facilities, contribute to the overall impact on saving lives. There has been progress in the reduction of maternal, under-5 and infant mortality rates in South Africa (SA). The SA Demographic and Health Survey of 2016 (SADHS 2016) observed a decline in the under-5 mortality rate, from 59 deaths per 1 000 live births (2008) to 42 deaths per 1 000 live births (2016), and infant mortality from 45 deaths per 1 000 live births (2008) to 35 deaths per 1 000 live births (2016).[4] The SADHS 2016 also reported that almost all live

S44

Activities

Responsibility for actions

Health knowledge Coverage

Community Use of healthcare facilities Accessible healthcare facilities Transport between facilities

Saving lives

Enablers (healthcare managers) Adequately equipped facilities Adequately staffed facilities Appropriate skills

Quality of care

Doers (healthcare providers) Appropriate attitude

Fig. 1. Conceptual framework showing factors influencing programme effectiveness. [3]

births occurred in healthcare facilities, and that 97% of deliveries were attended by skilled healthcare workers (midwives or doctors). Since 2009, the National Department of Health (NDoH) has led the design, implementation, and monitoring of key programme strategies and interventions for improving maternal, newborn and child health outcomes. The paper examines selected interventions in the programme effectiveness framework to understand key components for achieving coverage and quality of care in saving lives.

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Methods

We reviewed reports and strategy documents to identify interventions led by the NDoH over the last 5 years to reduce maternal and child health mortality. Specifically, we examined the interventions that focused on improving the quality of care for pregnant women and children under 5 years of age. Our aim was to select a few interventions and evaluate them against the programme effectiveness framework (Fig. 1), to understand the critical factors for success and to inform replicability and scale-up in SA. The criteria used to select the interventions for analysis in this paper were: 1. Timeline – implemented within the timeframe of the national maternal, newborn, child and women’s health and nutrition strategy (2011 - 2016). 2. Governance, including strategic oversight and leadership – driven by the NDoH and not by a partner organisation. 3. Scale – implemented across an entire province or more than one province – and not in select facilities and communities only. 4. Resources – utilised existing resources in terms of human resources (no additional persons employed for implementation). 5. Quality of care activities – specific focus on improving quality of care and not only focused on coverage and achieving targets. 6. Results – interventions that showed improvement in results of maternal and child survival. Some of the reviewed interventions included policy changes to incorporate new evidence for HIV treatment regimens, infant feeding in the context of HIV, strengthening the use of data to understand gaps in programme implementation, reducing missed opportunities for linkage and retention in care and focusing on the implementation of standard treatment guidelines for the management of malnutrition. There were also examples of the development of multisectoral implementation plans to strengthen nutrition interventions and to ensure improvement in clinical care and governance through the district clinical specialist teams (DCSTs), a group of skilled healthcare practitioners who are specifically appointed to this end. Although all the above do contribute to overall improvement in maternal and child health outcomes, we selected interventions that met the criteria listed here for the purposes of this paper. The case studies were then examined against the programme effectiveness framework, to understand the key components and facilitators for success. They also identified additional factors that were critical to the success of the intervention beyond the components of the framework.

Results

Three interventions covering maternal and child health and survival programmes were selected and examined against the framework. For

maternal health, we looked at a 12-district, two-step intervention aimed at improving obstetric skills and practices, i.e Essential Steps in Managing Obstetric Emergencies (ESMOE). For child health, we examined a province-wide, multipronged intervention that reduced the case fatality rate for severe acute malnutrition (SAM) ,and a nationwide, multipronged intervention to increase exclusive breastfeeding. Achieving better maternal health outcomes through ESMOE saturation training and simulation fire drills as a quality improvement strategy With almost all deliveries in SA occurring in health facilities, improving the quality of care provided at the facilities in a timely, consistent, safe and equitable manner should be the key focus in achieving better maternal health outcomes. The recent Saving Mothers Report[5] identified the fact that the poor management of emergency obstetric cases was associated with maternal deaths, and recommended that it was essential to address gaps in healthcare workers’ knowledge, skills and practice in order to reduce preventable deaths. Following this recommendation, the NDoH developed, tested and scaled up the ESMOE training programme, in collaboration with the SA Medical Research Council.[6] In 2012, 12 ‘most-in-need’ districts were identified, and >80% of the healthcare professionals that were involved in maternity care were trained in ESMOE. A ‘before-and-after’ assessment of the training indicated that all staff categories had improved their knowledge and skills. However, the baseline knowledge and skills were low in all categories.[7] This indicated the need for continuing professional development for all levels of healthcare professionals. The effect of the training on the health system was monitored by assessing the number of emergency-care signal functions a healthcare facility could perform at the start of the project and 1 year after the saturation training had been completed. A repeat baseline assessment was conducted at all the sites, using the same tools as the original baseline assessment. There were significant improvements in the number of signal functions the community healthcare centres (p<0.01) and district hospitals (p=0.05) could perform 1 year after the training, compared with the baseline (Table 1). However, in two areas (performing assisted deliveries and manual vacuum aspiration for spontaneous incomplete miscarriages) there was very little improvement in the community health centres (CHCs).[6] This was due to Nursing Council regulations, which state that these activities are not within the scope of practice of professional nurses with midwifery qualifications.[8] As most of the CHCs are staffed with professional nurses with midwifery as part of their training, and not professional advanced midwives, there was very little change in these aspects. The impact of the scale-up of the ESMOE programme was monitored by assessing the number of maternal deaths from January 2011 to the start of saturation training in each district, starting in October 2012 (the ‘before’ period), and from after the saturation

Table 1. Number of signal functions at baseline and repeat baseline assessment Community health centres (n=51) Number of functions 9 8 7 6 5 4 3

Baseline, % N/A N/A 0 5.9 49.0 94.1 100

Repeat baseline, % N/A N/A 3.9 9.8 64.7 100 100

District hospitals (n=62) Baseline, % 46.8 80.6 91.6 96.8 100 100 100

N/A = not applicable.

S45

March 2018, Vol. 108, (3 Suppl 1)

Repeat baseline, % 62.9 82.3 90.3 96.3 100 100 100

RESEARCH

training to the end of 2016 (the ‘after’ period). There was a significant reduction of 29.3% in all maternal deaths, and a 17.5% reduction in direct maternal deaths after the training compared with numbers before the training (Table 2). There was also a significant reduction in deaths due to bleeding at or after caesarean delivery. The 12 intervention districts had a 29.3% reduction in maternal deaths, compared with 5% for the other 40 districts (p<0.0001). Implementing an emergency obstetric-care training programme in which 80% of the healthcare professionals are trained was thus associated with significant improvements in the knowledge and skills of the healthcare professionals, improvement in the provision of emergency services and a reduction in maternal mortality. Sustaining and sharpening obstetric skills Sustaining the knowledge and skills obtained in the ESMOE programme was the next priority. An average district hospital sees approximately one woman every 2 years who dies due to obstetric haemorrhage or complications of hypertension (Table 3).[9] In other words, these conditions are rare in these institutions, and even more so in CHCs. To maintain knowledge and skills to manage these women it is essential to practice by performing emergency drills. A training programme was integrated into the ESMOE programme to teach facilities how to conduct emergency drills. This training was called Emergency Obstetric Simulation Training (EOST). The NDoH recommended that all maternity units (irrespective of the size of the unit) should perform at least one emergency obstetric drill per month. These drills were recognised for continuing professional development (CPD) points if there was a clear record of them

being performed. A record of having performed the drill had to be submitted to a central office. Facilities that performed regular drills, as evidenced by their submission for CPD points, were analysed, and a significant correlation was found between the reduction in maternal mortality and the number of drills. Those districts where drills were conducted and recorded regularly had an 18.4% reduction in mortality, whereas those who did not only had a 5.2% reduction in mortality (Table 4).[10] The training in emergency drills continues to be expanded throughout the country in all maternity units. The aim of the training is for every maternity unit to have staff trained in emergency care, and to conduct regular emergency drills. Analysing this intervention against the programme effectiveness framework shows that the focus was on improving the quality of care and ensuring appropriate skills in the healthcare providers, i.e. the doers, in the framework. However, a critical aspect of success was the focus on sustaining the skills, by institutionalising their implementation and linking these to incentives, i.e. the CPD points system. The incorporation of sustainability as an important crosscutting component in the programme-effectiveness framework is further highlighted in the discussion section of the paper.

Strategies to improve under-5 survival of by improving nutritional outcomes

There are no direct population estimates of the contribution of undernutrition or acute malnutrition to under-5 mortality in SA. However, clinical mortality audit data routinely collected in public hospitals indicates that at least 50% of under-5 deaths in hospitals

Table 2. Comparison of all maternal deaths and direct maternal deaths before and after ESMOE and EOST saturation training Live births 645 477 602 856

Before After

All MD* 1332 880

iMMR (all) 206.4 146.0

Direct MD† 628 475

Direct iMMR 97.30 78.79

ESMOE = Essential Steps towards Managing Obstetric Emergencies; EOST = Emergency Obstetric Simulation Training; MD = maternal deaths; iMMR = institutional maternal mortality ratio. *Risk ratio (RR) 0.707; 95% confidence interval (CI) 0.65 - 0.77; p<0.0001. † RR 0.825; 95% CI 0.73 - 0.93; p=0.0015.

Table 3. Distribution of births per facility per year

CHC* DH RH PT NC

Facilities, n ND 256 48 16 8

Number of pregnant women who died after giving birth, per level of care NPRI OH HT 125 166 139 70 23

95 116 81 42 14

112 111 87 61 20

Births/year 172 984 392 861 239 269 74 066 58 497

Average births/facility ND 1 535 4 985 4 629 7 312

NPRI = non-pregnancy-related infections; OH = obstetric haemorrhage; HT = hypertensive disorders in pregnancy; CHC = community health centre; ND = not determined, DH = district hospital; RH = regional hospital; PT = provincial tertiary hospital; NC = national central hospital. *The number of CHCs performing deliveries is unknown and varies from year to year, and many perform <10 deliveries per year.

Table 4. Effect of EOST exercises Submitted data for CPD points No Yes Total

Districts, n* 25 20 45*

Average iMMR 2011 - 2013 (before EOST) 134.8 174.3 152.4

Average number of drills per district 0 145 64

Average iMMR 2014 - 2015 (after EOST) 127.8 142.3 134.3

Average difference 7.0 32.1 18.1

Percentage decrease 5.2 18.4† 11.5

EOST = Emergency Obstetrics Simulation Training; CPD = continuing professional development; iMMR = institutional maternal mortality ratio. *Only 45 districts were included, as the Western Cape had their own mechanism of obtaining CPD points for the drills and the Amathole District and Buffalo City were treated as a single entity in 2011 when the study began. † Paired t-test p<0.004.

S46

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Improving inpatient quality of care of children with severe acute malnutrition (SAM) All children aged under 5 should be assessed, with growth monitoring and plotting of the growth curve at every visit at the primary-care level, and referred, where necessary, to higher levels of care. Improved, timely and accurate case detection of all forms of acute malnutrition, accompanied by better data reporting, will result in more SAM cases that access early nutritional treatment, care and support. Early access to treatment would deliver better outcomes in these cases. To reduce overall inpatient under-5 mortality, the NDoH prioritised the development and implementation of quality improvement plans to decrease the case fatality rate for SAM. The plans included the formation of a multidisciplinary management team led by senior hospital management; improving active case detection in casualties and inpatient care units; strengthening healthcare workersâ&#x20AC;&#x2122; skills to treat and manage sick SAM cases by improving adherence to standard treatment guidelines; and improving data quality through better recording, reporting and responding to avoidable factors contributing to SAM deaths. The inpatient SAM case fatality rates (iCFRs) in SA provide a good composite proxy measure of the quality of care provided to sick SAM inpatients. Fig. 2 shows the trends in the iCFRs from 2009 to 2017, and there is a clear systematic reduction in the SAM case fatality rate from 2009 to 2016, with a significant reduction observed in the last 3 years of intensive implementation of the quality improvement plan; this effort targeted facilities with the largest number of SAM deaths. To illustrate the impact of the national intervention, we describe a case study from the province of Mpumalanga. The case fatality rate in Mpumalanga province between 2011/12 and 2014/15 ranged between 15% and 25%. Fig. 3 shows the trends in the SAM iCFRs in under-5 children from 2011/12 - 2016/17 in Mpumalanga. Following an initial reduction in the provincial and district SAM iCFRs from 2011 to 2013, an increase in the rates occurred during the 2014/15 financial year, to 19.1%, with 233 children dying from SAM-related mortality. In order to address this high mortality rate, the province developed a turnaround

SAM iCFR, %

25 20

19.3

16.4

13.3

12.7

11.3

11.6

8.9

8.0

2015

2016

5 0 2009

2010

2011

2012

2013

2014

Year

Fig. 2. Trends in the SAM inpatient CFRs between 2009 and 2017. (SAM = severe acute malnutrition; iCFR = inpatient case fatality rate.)

S47

Mpumalanga Ehlanzeni Gert Sibande Nkangala

40 SAM inpatient CFR, %

were associated with underweight status, and just over 30% were attributable to severe acute malnutrition (SAM).[11] The SADHS 2016 showed that there has been a remarkable improvement in exclusive breastfeeding, from 7% in 1998 to 32% in 2016.[4] However, this still means that two out of three children are not exclusively breastfed and miss the benefits of this practice. The sections below describe the two selected interventions that led to improvement in child survival outcomes, and examine these interventions against the programmeeffectiveness framework.

35 30 25 20 15 10 5 0 2

1 01

2 01

4 01

3 01

5 01

6 01

Fig. 3. Provincial- and district-level SAM inpatient CFR for under-5 children in Mpumalanga: 2011/12 - 2016/17. (SAM = severe acute malnutrition; CFR = case fatality rate. Data obtained from the District Health Information System.)

strategy with support the NDoH, implementing partners and the University of Pretoria. Fig. 4 highlights the different activities implemented over a period of 2 years (2015 - 2017). These activities were aimed at improving early case detection, referral and appropriate treatment and management of SAM cases, as needed. An analysis of the interventions against the programme-effectiveness framework highlighted the following aspects: 1. A focus on coverage, with activities in the framework including: a. Health promotion to improve health knowledge and use of facilities by communities. b. Improving access to healthcare facilities and transport â&#x20AC;&#x201C; these important aspects were facilitated by the enablers (healthcare managers). 2. A focus on quality of care, with activities in the framework including: a. Adequately equipped facilities and adequately staffed facilities. This was done by enablers (healthcare managers). b. Appropriate skills and appropriate attitudes â&#x20AC;&#x201C; ensured by doers (healthcare providers). All the above was made possible through management oversight and using evidence-based planning involving a multidisciplinary team. The interventions collectively contributed to an overall provincial reduction in SAM mortality to 146 deaths during 2015/16 (a decrease of 12.5% in the iCFR), which was subsequently reduced to 83 deaths (a further decrease of 7.8% in the iCFR) during 2016/17. The most notable reduction in the iCFR (from 21.9% during 2014/15 to 9.3% in 2016/17) and the number of inpatient SAM deaths (from 82 deaths during 2014/15 to 19 deaths during 2016/17) occurred in the Gert Sibande District. Creating an enabling policy and implementation environment to improve the coverage of breastfeeding Improving infant and young-child feeding practices requires concerted efforts towards improving breastfeeding rates in SA. Breastfeeding remains the undisputed preferred feeding option for optimal growth and development of children. The breastfeeding series published in the Lancet in 2016 reinvigorated interest in breastfeeding and, in addition to the well-known advantages of

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

Provincial SAM training Commercial feeds on contract Gert Sibande MRU established Provincial consultative workshop and action plan development Provincial action plan implemented Provincial monitoring and support visits Health promotion intervention on SAM Referral to SASSA for SRD implemented in Gert Sibande District Provincial supplementation protocol implemented District SAM advocacy session District action plan implemented National support visit – Embhuleni hospital CMVC procured to prevent interruption in supply of therapeutic feeds SAM screening campaign – Albert Luthuli subdistrict SAM training for dieticians National support visit – Piet Retief Hospital 2014 Apr

Jul

1/9/201 - 5/9/2014 1/9/2014 - 31/3/2017 1/11/2014 - 31/3/2017 19/3/2015 - 21/4/2015 1/5/2015 - 31/3/2017 25/8/2015 - 8/12/2015 1/9/2015 - 31/12/2015 1/11/2015 - 31/3/2017 1/12/2015 - 31/3/2017 27/1/2016 1/2/2016 - 31/3/2017 29/3/2016 1/5/2016 - 23/6/2016 10/5/2016 - 23/6/2016 7/6/2016 11/10/2016 - 12/10/2016

Oct

2015

1/4/2014 CFR 14.7%

Apr

Jul

Oct

2016

31/3/2015 CFR 21.9%

Apr

Jul

Oct

2017

31/3/2016 CFR 18.2%

2017 31/3/2017 CFR 8.3%

Interventions related to availability of therapeutic feeds Strategic planning and capacity building Monitoring and support interventions Community-based interventions

Fig. 4. Timeline of interventions aimed at improving the integrated management of under-5 children with severe acute malnutrition in Mpumalanga between 2014 and 2017. (SAM = severe acute malnutrition; MRU = monitoring and response unit; SASSA = South African Social Security Agengy; SRD = social relief of distress; CMVC = combined mineral vitamin complex; CFR = case fatality rate.)

breastfeeding, also highlighted the long-term breastfeeding outcomes of improved intelligence and protection against the development of obesity and diabetes.[12] The World Bank has identified breastfeeding as the most lucrative return on investment, and noted the additional benefits to the mother, including fertility protection and reduced risk of the development of breast and ovarian cancer.[13] Historically, breastfeeding became less common in high-income countries during the 20th century, and this had a ripple effect in lower- and middle-income countries. Breastmilk substitutes (infant formula) were viewed as modern and prestigious. SA did not escape this trend, and it was exacerbated by the provision of free infant formula as part of the prevention of mother-to-child transmission programme. The detrimental effects of this practice were well documented. In a laudable attempt to reposition breastfeeding as a child survival strategy, and to do away with formula feeding, SA adopted the Tshwane Declaration on Breastfeeding in August 2011, with highlevel political support from the Minister of Health.[14] The declaration reaffirmed the fact that breastfeeding rates in SA were extremely low. In addition to calling for the prohibition of free formula, it called for the establishment of human-milk banks, the implementation of the MBFI (Mother Baby Friendly Hospital Initiative) and Kangaroo Mother Care. Currently, there are 37 breastmilk banks, and 407 out of 545 (75%) public health facilities with maternity beds are accredited as part of the MBFI. In December 2016, SA cohosted the World Breastfeeding Conference. The deputy Minister of Health launched the country’s first 3-year Breastfeeding Campaign during the opening ceremony of the conference.[15] The interventions are beginning to bear fruit, as evidenced by the cumulative exclusive breastfeeding rate, which has increased from 7% in 1998 to 32% in 2016.[4] SA now wants to accelerate this rate, and has set an ambitious target of increasing its 14-week exclusive breastfeeding rates to 70% in 2019. Reviewing the national scale-up of the promotion of exclusive breastfeeding over the last few years, with interventions by the community, enablers and doers based on the programme-

S48

effectiveness framework, showcases the multiplicity of actions to influence coverage and quality of care. The milk banks and the MBFI scale-up ensured that there were adequately equipped facilities and staff with appropriate skills and attitude, while the campaigns and community engagement focused on actions at the community level and improving health knowledge and the use of healthcare facilities.

Discussion

The interventions highlighted in this article demonstrate how selected interventions have contributed to the reduction in maternal and child deaths through improvement in the quality of care and coverage. We found that access to services (ensured by the enablers) and their use (by individuals) is an important component of overall quality of care. The sustained use of the services (based on improved actions and attitudes of healthcare providers) was an important component of coverage and quality. We do know that provision of equipment and adequate staffing are critical aspects of quality of care. These were not covered in the case studies, except for the issue of staffing, as highlighted in the maternal-health example. The lack of adequate staffing is a barrier to providing emergency obstetric care, and it is important to find a solution to ensure a sustained improvement in effective coverage.[16] While key actions focused on coverage and quality of care are covered in all three interventions, we found limitations in the programme-effectiveness framework. In relation to the ‘doers’, the article does identify challenges with appropriately skilled staff (noting the low baseline knowledge and skills levels with respect to obstetric emergencies) and the reliable implementation of protocols by staff (in relation to inadequate treatment of severely malnourished children). Basic training needs to ensure that graduates are competent. In addition, there are large numbers of existing staff, foreign-trained and sessional doctors in the country, which means that initiatives such as ESMOE, aided perhaps by new technologies such as e-learning, will be needed at a larger scale and with a guaranteed level of quality for sustainability in generating results.

March 2018, Vol. 108, (3 Suppl 1)

RESEARCH

In relation to the community and the individual, health knowledge appears to be variable, and therefore adequate childcare practices, especially the sustaining of breastfeeding, remain challenging. The establishment of a multisectoral forum to address the multifaceted problems of under- and over-nutrition is important. However, the low (though improving) breastfeeding rates and relatively high levels of stunting all point to the need for more effective actions in this area. The observed successes have been a consequence of important initiatives in policy and programme design, including supporting rollout and implementation of the initiatives. However, an important point to note is the high level of political commitment to and technical oversight of these initiatives in SA, which was demonstrated by the setting up of ministerial advisory committees for each target group, i.e. mothers, newborns and children. These committees are guided by passionate and knowledgeable expert-advocates, who are able to analyse information and recommend the latest evidencebased action. They benefit from the direct support and dedication of the Minister of Health. The framework does not, however, account for key cross-cutting issues, such as political commitment, allocation of resources and overall management oversight, that are critical to ensuring sustainability for both coverage and quality of care. Furthermore, the importance of focusing the actions by using relevant data and targeting the initiatives to areas that are most in need of positive impacts on survival is seen in the examples described in this article. The emphasis on understanding the evidence and ensuring that data collected are analysed, interpreted and used for action has been a major thrust across the programmes. In addition, understanding the context and building on recommendations in the midterm review of the maternal, newborn child and women’s health programmes in 2014[17] led to the focus on clinical governance issues and improving local motivation, management skills and efforts to maximise partnerships with academia, professional bodies and partner organisations. Efforts were made to ensure that local-level data resulted in actions tailored to the local context, as seen in the emphasis and prioritisation of specific activities in the phased scale-up of the ESMOE training programme and the case study from Mpumalanga. In SA, clinical experts were placed at the district level to analyse the local situation and develop targeted solutions. Initiatives such as ESMOE and EOST, and improved SAM case management, were possible because of the maximal utilisation of these skilled cadres. The use of the framework would benefit by adding a fourth dimension of cross-cutting components. These include evidencebased targeting and the use of data for action, management oversight towards institutionalising and sustaining actions, continuous feedback loops through monitoring and mentoring actions, forging and sustaining partnerships, leveraging on existing resource and partner networks, and resource allocations. All of these were highlighted in the examples described in the article and contributed to the success of the interventions and overall impact on improving maternal- and child-health outcomes.

Conclusion

Significant improvements in programme effectiveness, with better coverage and quality of care, were achieved. This has resulted in a reduction in the number of deaths and an improvement in early identification of weaknesses and consequent change of practices. However, more needs to be done towards sustaining these efforts and institutionalising these practices.

S49

The global agenda of the Sustainable Development Goals (SDGs) for 2030 on maternal and child health carries the slogan ‘Survive, thrive and transform.’[18] In SA, the new Integrated Plan for Reducing Mortality in Mothers, Newborns and Under-fives has a strong focus on reducing mortality in healthcare facilities that underpins the goals of the ‘thrive’ component of the SDGs. Reflecting on the enhanced programme effectiveness framework highlights the importance of interventions that address the social determinants of healthcare, especially among the most vulnerable, and sustained improvements in client experiences of healthcare service delivery, including focusing on the empowerment of mothers, caregivers and staff at facilities. Acknowledgements. The authors wish to acknowledge SAMRC for the support of the Maternal and Infant Health Care Strategies Unit and the provincial Department of Health for supporting the quality improvement plans. Author contributions. RB, PB, SB, SK conceptualised the paper. RB wrote the initial draft and different sections were written by SK, MV, ND, CM. SB collated and finalised the paper, PB and SK supported review, editing and submission of the first draft. PB, SB, SK, CM, RB, ND provided significant inputs to the paper. SB finalised and edited the paper, SK and PB reviewed and edited the manuscript. Funding. None. Conflicts of interest. None. 1. Tanahashi, T. Health service coverage and its evaluation. Bull World Health Organ 1978;56(2):295-303. 2. World Health Organization. Quality of Care. Geneva: WHO, 2017. 3. Save the Children, United Nations Children’s Fund, National Department of Health, South African Medical Research Council/University of Pretoria Maternal and Infant Health Care Strategies Unit. Every Death Counts. Saving the lives of mothers, babies and children in South Africa. Cape Town: SAMRC, 2008. http://www.mrc.ac.za/bod/edcrptfinal.pdf (accessed 1 February 2018). 4. Statistics South Africa. South Africa Demographic and Health Survey 2016: Key Indicators Report, 2016. Pretoria: StastSA, 2017. 5. National Committee for Confidential Enquiries into Maternal Deaths. Saving Mothers 2005 - 2007: Fourth Report on Confidential Enquiries into Maternal Deaths in South Africa. Pretoria: NCCEMD, 2008. http://www.health.gov.za/index.php/2014-08-15-12-55-04/category/1012011rp?download=190:saving-mothers-2005-2007-fourth-report-on-confidential-enquiries-intomaternal-deaths-in-south-africa-full-report (accessed 1 February 2018). 6. Frank K, Lombaard H, Pattinson R. Does completion of the Essential Steps in Managing Obstetric Emergencies (ESMOE) training package result in improved knowledge and skills in managing obstetric emergencies? S Afr J Obstet Gyn 2009;15(3):94-99. 7. South African Medical Research Council. Essential Steps in Managing Obstetric Emergencies (ESMOE) – Emergency Obstetric Simulation Training (EOST) Scale-up Programme: Report to National Department of Health. Pretoria: SAMRC, 2016. 8. National Department of Health. Nursing Act, 2005 (Act No. 33 of 2005). Regulations relating to the approval of and the minimum requirements for the education and training of a learner leading to registration in the category midwife. Government Notice No. R786. 2014. 9. National Committee for Confidential Enquiries into Maternal Deaths. Saving Mothers 2011 - 2013: Sixth Report on Confidential Enquiries into Maternal Deaths in South Africa. Pretoria: NCCEMD, 2014. http://www.kznhealth.gov.za/mcwh/Maternal/Saving-Mothers-2011-2013-short-report.pdf (accessed 1 February 2018). 10. Pattinson R. Correlation between numbers of EOST drills reported for CPD points and change in iMMR per district. Priorities in Perinatal Care Conference, Mpekweni Holiday Resort, Eastern Cape, South Africa, 7 - 10 March 2017. https://www.perinatalpriorities.co.za/proceedings-database/ Presentation (accessed 1 February 2018). 11. Stephen CR, Bamford LJ. Saving Children 2010 - 2011: A Seventh Survey of Child Health Care in South Africa. Pretoria: Tshepesa Press, Medical Research Council and Centers for Disease Control and Prevention, 2013:1-142. 12. Rollins NC, Bhandari N, Hajeebhoy N, et al. Why invest, and what it will take to improve breastfeeding practices? Lancet 2016;387(10017):491-504. https://doi.org/10.1016/S0140-6736(15)01044-2 13. Shekar M, Kakietek J, Eberwein JD, Walters D. An Investment Framework for Nutrition: Reaching the Global Targets for Stunting, Anemia, Breastfeeding, and Wasting. Washington: The World Bank, 2017. 14. National Department of Health. The Tshwane Declaration of Support for Breastfeeding in South Africa. S Afr J Clin Nutr 2011;24(4):214. 15. 2nd World Breastfeeding Conference. 11 - 14 December 2016, Johannesburg, South Africa. http:// worldbreastfeedingconference.org/agenda.html (accessed 1 February 2018). 16. Donabedian A. The quality of care. How can it be assessed? JAMA 1998;260(12):1743-1748. https:// doi.org/10.1001/jama.1988.03410120089033 17. National Department of Health. Mid-term Review of Strategic Plan for Maternal, Newborn, Child and Women’s Health and Nutrition in South Africa 2012 - 16. Pretoria: NDoH, 2016. http:// childhealthpriorities.co.za/conference-presentations/2014/Review%20of%20the%20National%20 MNCWH%20and%20N%20Strategy%202012%202016%20S%20Bardwaj%20H%20Saloojee.pdf (accessed 1 February 2018). 18. World Health Organization. Global Strategy for Women’s, Children’s and Adolescents’ Health, 2016-2030. Geneva: WHO, 2017.

Accepted 10 October 2017.

March 2018, Vol. 108, (3 Suppl 1)

NOTES

S50

March 2018, Vol. 108, (3 Suppl 1)

NOTES

S51

March 2018, Vol. 108, (3 Suppl 1)

NOTES

S52

March 2018, Vol. 108, (3 Suppl 1)