Faculty of Medicine, Suez Canal University Clinical Epidemiology Unit
Measurements And Bias Badr Mesbah Prof. of Pediatrics
Measurements when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind; it may be the beginning of knowledge.
Lord Kelvin, a distinguished physicist of the 1800s.
Measurements Definition: To express a concept or characteristic of a group of objects in term of classes (qualitative, categorical, or nominal) or numbers( quantitative)
Measurements Why does one perform measurement? - To form a clearer definition of the concept or object to gain a more profound understanding and to be able to perform further operations or calculations - To test a hypothesis
Accuracy of Measurements The main goal in measurement is to achieve high accuracy : - High validity (low systematic error) - high reliability (low random error) on measuring the exposure factors and outcomes of interest without bias and errors or to minimize them to the least as possible
Validity Definition: - the extent to which a technique measures what it purports to measure. - The degree to which the results of a measurement correspond to the true state or truth Synonym: Conformity
Validity A valid measurement should have two characteristics: - High Sensitivity - High Specificity
Sensitivity – the ability of a test to correctly identify those who have the disease – Measures the true positive. – Proportion of diseased persons with positive test result. – Positivity in disease. – Ability to include person with a disease.
– A test with high sensitivity will have few false negatives
Specificity the ability of a test to correctly identify those who do not have the disease
Measures the true negative Proportion of healthy persons with negative test result. Negativity in health. Ability to exclude person without a disease.
– A test that has high specificity will have few false positives
Sensitivity and specificity Example: If glycosuria is taken as a test for diabetes, and it was positive in only 26% of true diabetics and it was absent in 98% of non diabetics What is the sensitivity and specificity of glycosuria as a test for diabetes?
Sensitivity and specificity Diabetes
present
Present
Absent
26
2
28
74
98
172
100
100
200
a b c d
glycosuria
absent
Sensitivity = a / (a + c)
Specificity = d / (b + d)
Reliability
The level of agreement between replicate measurements. A reliable measurement is one that has nearly the same value each time it is measured Synonym:
• Precision, repeatability, reproducibility, consistency
Reliability ď Ź Depends on - Subject (biological) variations (random, systematic) - Observer variation: Within observer (random) Between observer (Systematic)
Reliability and Validity Reliability Definition The degree to which a variable has nearly the same value when measured several times Best way to Comparison among assess repeated measures
Validity The degree to which a variable actually represents what it is supposed to represent Comparison with a reference standard
Reliability and Validity Reliability Value to study Threatened by
Increase power to detect effects
Validity Increase validity of conclusions
Mostly random Mostly systematic error error contributed by : contributed by : The observer The subject The instrument
The observer The subject The instrument
Reliability and Validity
.... .
A
B
. . . .
C
.. . . .
A- High reliability, low validity B- Poor reliability, fair validity C- Excellent reliability and validity D- Poor reliability and validity
D
. . .. .
Reliability and Validity Frequency
A
B
C
A- Valid and reliable B- Valid but not reliable C- Not valid but reliable D- Not valid and not reliable
DD Measurement
True value
Bias ď Ź
A preference or an inclination ď Ź Bias refers to an effect at any stage of an investigation or interference tending to produce results which depart systematically from the true value
Classification of Bias Selection
Bias Information Bias Confounding
Intervention Bias
Selection Bias ď Ź
It occurs when there is a systematic difference between the characteristics of the people that are selected for the study and those that are not. ď Ź Selection biases affect the applicability and usefulness of findings and make it impossible to generalize the results to all patients with the disorder of interest.
Selection Bias ď Ź
Example: - Volunteer bias: If those who volunteer for the study differ from those who refuse participation, the results are affected. - Non-response bias: when those who do not respond to take part in a study differ in important ways from those who respond.
Selection Bias
Example: - Loss to follow up bias: when those who remain in the study differ from those “lost,” in terms of personal characteristics and outcome status. - Survivor bias: when the study base consist of those people who have survived long enough to become members of the base
Information Bias ď Ź
ď Ź
A distortion in the estimate of association between risk factor and disease that is due to systematic measurement error or misclassification of subjects on one or more variables, either risk factor or disease status. It occurs if data used in the study are inaccurate or incomplete, thus influencing the validity of the study conclusions.
Information Bias
Inaccurate measurement of study variables can lead to bias Sources of inaccurate measurement: • subject error – error within the individual for any reason, e.g. imperfect recall of past exposures • Instrument error – e.g. equipment not properly calibrated, wording of question • Observer error – error in use of instrument or recording
Information Bias ď Ź
Example:
- Recall bias: Recall or memory bias may be a problem if outcomes being measured require that subjects (cases and controls) recall past events.
- Diagnosis bias: This bias occurs when the disease being investigated is more likely to be detected in people who are under frequent medical surveillance than those receiving routine medical attention.
Information Bias ď Ź
Example: - Lead time bias: when diagnosis of a condition is made during its latency period, leading to a longer duration of illness.
Intervention (exposure) bias ď Ź
This type of bias involves differences in how the treatment or intervention was carried out or how subjects were exposed to the factor of interest.
Intervention (exposure) bias ď Ź
Example: - Compliance bias: Compliance bias occurs when differences in subject adherence to the planned treatment regimen or intervention affect the study outcomes. - Proficiency bias: Proficiency bias occurs when treatments or interventions are not administered equally to subjects. This may be due to skill or training differences among personnel or differences in resources or procedures used at different sites.
Intervention (exposure) bias ď Ź
Example: - Contamination bias: occurs when control group subjects inadvertently receive the intervention or are exposed to extraneous treatments, thus potentially minimizing the difference in outcomes between the cases and controls.
Confounding ď Ź
A situation in which the measure of effect of exposure on disease is distorted because of the association of the study factor with other factors that influence the outcome.
ď Ź
These other factors are called confounders
Confounding A
B
C
B
Confounding ď Ź
A variable is a confounder if: 1. it is an independent risk factor (cause) of disease 2. it is unevenly distributed among the exposed and non-exposed 3. it is not on the causal pathway between exposure and disease
Confounding Confounders are true causes of disease, whereas biases are artifacts
Strategies for Dealing with Bias ď Ź
The causes of bias can be related to: - The manner in which study subjects are chosen. - The method in which study variables are collected or measured. - The attitudes or preferences of an investigator. - The lack of control of confounding variables.
Strategies for Dealing with Bias
The key to decreasing bias is to identify the possible areas that could be affected and change the design accordingly.
Minimizing selection bias
Selection of study subjects – clearly define study population in time and place – use sampling techniques that result in choosing groups from the same population – use methods that result in high recruitment rates Assignment of exposure status – use random allocation (RCTs) Study subjects omitted from analysis – use methods to minimize loss to follow up – review non-respondents
Minimizing information bias
Use valid reliable tools to measure all study subjects Train staff and monitor their use of research tools Regular quality checks of research tools Blinding of study subjects and assessors Subjects in C-C study unaware of study hypothesis Consider sub-study to determine validity and reliability of measurements
Controlling for confounders ď Ź
Confounding is the only type of bias that can be prevented or adjusted for, provided that confounding was anticipated and the requisite information was collected.
Controlling for confounders
At the design phase • restriction (limit the study to people with one level of the potential confounder) • stratified allocation within risk strata • matching • randomization – deals with known and unknown
Controlling for confounders
At the analysis phase • demonstrate comparability • data analyzed in strata of the confounder • statistical modeling – useful to control for multiple confounders
Thank you