Chapter 10: Bias

Selection bias

Arises from systematic differences in selecting and following study groups, in which association among study subjects differs from association among eligible source population

What is an example of recall bias in a cohort study?

If exposed subjects are more or less likely than unexposed subjects to recall and report subsequent diseases in a retrospective cohort study

Interviewer bias

a type of information bias that is a systemic difference In soliciting, recording, or interpreting information that occurs in studies using in person or telephone interviews. It can occur in every type of epidemiologic study

In what types of studies can information bias occur?

In both retrospective and prospective studies

T or F

Selection bias only occurs in retrospective cohort studies or in case control studies-False (selection bias can also occur in prospective cohort and experimental studies from differential loss to follow up)

Confounding

the mixing of effects between the exposure, the disease, and a third variable that is termed a confounder, which distorts the relationship between exposure and disease

what are the methods used to avoid selection bias?

using the same criteria for selecting cases and controls; high participation rates; using a variety of methods to trace study subjects; taking diagnostic and referral practices into account when designing a study to minimize any systemic differences between the cases and controls

How do you determine if the misclassification biases results toward or away from the null?

Calculate the odds ratio based on the completely accurate data, and compare it to the odds ratio of the biased study under question and compare the two odds ratios. So, if for example, the Accurate and true odds ratio is 4.0 and the odds ratio calculated for the Study that experienced Misclassification is 10.0, we'd consider it biased away from the null

What can epidemiologists do to reduce misclassification?

1. Improve the accuracy of the collected data, 2. Validation-corroborating the collected data using several sources, 3. Switching to a better information source for the exposure or disease, 4. Define the exposure and disease using sensitive and specific criteria

What can investigators do to avoid interviewer bias?

1. Mask interviewers to the subjects disease (in case control studies) or exposure status (in experimental studies and cohort studies); 2. Design standardized questionnaires consisting of close ended, easy to understand questions with appropriate response options; 3. Instruct interviewers on appropriate questioning techniques

What are the different types of information bias?

1. Recall bias, 2. interviewer bias, 3. differential and non-differential misclassification

What are the methods used to minimize recall bias?

1. the most common method is to select a diseased control group; 2. design a structured questionnaire so that information is obtained efficiently and exposure ascertainment is complete and accurate, 3. Usage of Self-administered questionnaire or audio computer assisted self interviewing (ACASI) for socially sensitive topics (i.e. Alcohol, illicit drug use sexual behaviors); 4. Rely on biological measurements and pre-existing records for study data

Recall bias typically occurs in what type of study?

A traditional case-control study with non diseased controls (although recall bias can also occur in cases who fail to recall details on sensitive exposures, such as drug and alcohol use); in either scenario, the direction of the bias is determined by which group has less accurate recall

Recall bias

A type of information bias that occurs when there is a differential level of accuracy in the information provided by compared groups; an example can occur in case control studies if cases are more or less likely than controls to recall and report prior exposures

What is a method investigators can use to determine if recall bias is present within their groups of cases and controls (in a case-control study)?

Ask subjects about their knowledge of the study hypothesis. Those with knowledge about the study hypothesis are thought to be more prone to recall bias than those without such knowledge. Thus, Investigators draw their main conclusions from the study subjects without knowledge of the hypothesis

Why is selection bias more likely in case control and retrospective cohort studies?

Because the exposure and outcome have occurred by the time the subjects are selected (They are retrospect)

Bias

a systematic error in the design or conduct of study that leads to an erroneous association between the exposure and disease

when the direction of the bias is away from the null, this means that the error causes the true measure association to be what?

Overestimated

What does it mean for results to be biased away from the null?

This means that the true relationship between the exposure and the disease is probably weaker than observed (as seen with studies that exhibit differential bias)

Nondifferential misclassification of a dichotomous exposure biases results away from or toward the null?

Toward the Null

Information bias

Type of bias that arises from a systematic difference in the way that the exposure or outcome is measured between compared groups. 3 key features Include : 1. it occurs after the subjects have entered the study, 2. pertains to how the data are collected, 3. often results in incorrect classification of participants as either exposed or unexposed or as diseased or not diseased

when the direction of bias is toward the null, this means that the error causes the true measure association to be what?

Underestimated

Healthy worker effect

a form of selection bias that occurs in proportional mortality ratio (PMR) and standardized mortality ratio (SMR) cohort studies, in which the general population, which consists of both healthy and ill people, is selected for comparison to a relatively healthy working population, which can lead to this "Healthy worker" cohort being Inaccurately found to have a lower mortality rate

What are specific types of selection bias?

control selection bias, healthy worker effect (in occupational cohort studies), self selection bias, differential loss to follow up (can impact prospective cohort and experimental studies), Differential surveillance, diagnosis, or referral of subjects according to exposure and disease status

Misclassification (i.e. measurement error)

likely the most common form of bias in epidemiologic research, it is An error in the classification of the exposure or the disease (i.e an exposed individual is classified as unexposed or vice versa, or a diseased individual is classified as non diseased or vice versa)

What are the two types of misclassification?

nondifferential and differential

What is the difference between nondifferential loss to follow up and differential loss to follow up?

nondifferential losses occur on one axis (the disease or the exposure) they're unrelated to the other axis and thus Will not affect the relative measure of association. Differential loss to follow up refers to losses that are related to both exposure and the disease, and thus differential loss to follow up can bias relative and absolute measures of association either upward or downward

Differential misclassification

refers to errors on one axis (exposure or disease) that are related to the other axis (exposure or disease)

Nondifferential misclassification

refers to errors on one axis that are unrelated to the other axis, for example, if there is an error in exposure classification, it occurs with equal likelihood among diseased and non diseased individuals; Overall it biases results toward the null