Major biases in pharmacoepidemiological research include selection bias, information bias, and confounding. Selection bias can occur through referral bias, self-selection bias, prevalence study bias, and protopathic bias. Information bias includes non-differential and differential misclassification. Differential misclassification involves recall bias and detection bias. Confounding involves covariates related to disease development that may incorrectly attribute an effect to drug exposure. Accurately defining exposure timing and avoiding these biases is important for valid pharmacoepidemiological research.
2. Major objective of pharmacoepidemiological research is to analyze
the relationship between ‘exposure’ (drug or disease) and outcome of
interest (usually health status of the patients).
Exposure (drug or disease):
• Not a stable phenomenon (varies w/ individuals, communities,
etc…);
• Can be associated with factors related to the ‘outcome of interest’
• E.g.:
indication (for prescribing the drug) vs QoL after drug
usage;
therapeutic regimen vs patient compliance;
drug usage based on publicity;
drug usage vs natural course of the disease during
treatment;
4. Challenge of pharmacoepidemiological research is to obtain an
accurate estimate (without error), of the relationship between
exposure and health status in the context of ‘outcome of interest’.
[Why accurate estimate and not accurate value?]
2 types of errors:
• Random error is related to precision and reliability.
• Systematic error is related to validity and bias.
[Accuracy is the absence of both random and systematic errors].
5.
6. BIAS
Any systematic error in an epidemiological
study that results in an incorrect estimate
of the association between exposure and
outcome of interest (risk of the disease
and/or patient’s QoL).
7. 1981 – National Childhood Encephalopathy Study (NCES)
Results presented by researchers Alderslade and Miller;
A nationwide case–control study conducted in the UK by the
Committee on Safety of Medicines and the Joint Committee on
Vaccination and Immunization.
Research Question: Is there any possible association between DTP
vaccine and neurologic disorders, leading to subsequent
developmental deficit later?
Findings:
Risk of a severe acute neurologic „event‟ was significantly increased
(241 cases) within seven days following DTP vaccination.
[RR 2.3; 95%(CI) 1.4–3.2],
One year later….
• 7 (2.9%) of the 241 cases either died or had a developmental deficit .
• Only 3 of the controls died or had developmental deficit.
8. These results were used in many court trials by parents of disabled
children who were seeking compensation.
BUT…..
Credibility of the study was compromised by suspicions of „BIAS‟.
Numerous potential biases were identified and were responsible
(either fully or partially), for the results observed.
Referral bias: Physicians were aware of the study objectives and this
might have influenced their referral of cases and increased the
apparent relative risk.
Information bias:
• Interviewers were not blinded to study objectives, subjects‟ clinical
status;
• Date of onset of the neurological disorder was occasionally
difficult to establish (potentially increasing the apparent relative
risk).
9. Protopathic bias: possible presence of some other subclinical
neurological disease prior to vaccination, could have falsely
increased the relative risk.
Lack of precise „disease‟ definitions and inclusion criteria not related
to the DTP vaccine, caused misinterpretation of results (Reye‟s
syndrome, hypsarrhythmia, or acute viral encephalopathies).
Issues in study design can affect the validity of results in
pharmacoepidemiology research.
Pharmacoepidemiology studies may be affected by particular biases
more often than other epidemiologic studies.
10. BIASES IN PHARMACOEPIDEMIOLOGICAL RESEARCH
3 general categories:
Selection bias (related to the recruitment of study subjects or loss of
study subjects in follow-up)
Information bias (related to the accuracy of information collected on
exposure, health status);
Confounding (covariates or effect modifiers related to the
pathophysiology of disease development, whereby one factor (or
several factors acting together) can produce an observed effect that
may be incorrectly attributed to the exposure of interest).
11. SELECTION BIAS (Sample distortion bias)
Due to the selection (inclusion) of particular groups of subjects into
the study who ‘differ in characteristics’ from those in the target
population, causing distortion of the measurement of an effect
(outcome);
4 types of selection bias:
Referral bias
Self-selection bias
Prevalence study bias
Protopathic bias
12. REFERRAL BIAS
Can occur if the reasons for a physician referring a patient to the
study are related to the patient‟s exposure to (use of) the drug.
Can be problematic when an illness presents in such a manner that
an accurate diagnosis is not always obtained.
This bias can occur if the physician(s) involved in the study
refer/refers patients just to increase the number of subjects in
the study.
E.g., In the NCES study, it was noted that some physicians included
subjects who were already suffering from neurological disorders
before even receiving the DTP vaccine.
WHY DOES THIS HAPPEN????
Payment for physician‟s involvement in the study;
„Enthusiasm‟ to prove/disprove an association;
13. E.g.1,
• Study objective: Study of the possible association between
NSAIDs use and mild non-bleeding GUs.
• Site: Hospital „XYZ‟ with 2 groups of patients:
• Group-1: 1000 patients on regular NSAIDs therapy presenting
w/ abdominal pain (may be more likely to be suspected of
having a GU).
• Group-2: 10 patients with similar abdominal pain who are not
using NSAIDs
• Group-1 patients are more likely to be tested for GU than
Group-2.
• A study in Hospital XYZ using all Group-1 patients will show a
strong, but biased, association between NSAIDs use and mild
non-bleeding GUs (whether it be a cohort study or case-control
study).
WHY??????
14. Possible Referral Biases:
• If it’s a Cohort study:
Including only group-1 Patients;
Including group-2 patients into the cohort for the purpose of
increasing the number of subjects.
• If it’s a Case-control study: Including all group-1 patients vs all
group-2 patients;
15. E.g.2, Association b/n DVT and oral contraceptives
• The association b/n DVT and oral contraceptives is already well
known.
• The use of oral contraceptives is a vital factor in this study.
• Exposed cases (women on oral contraceptives) may be more
likely to be tested for DVT than women not on oral
contraceptives.
• Earlier studies reporting a positive association b/n drug (oral
contraceptives) and disease (DVT) actually initiated the referral
bias phenomenon.
*** Identifying the potential for referral bias in initial stages of any
study is important for that study, as well as for future similar
studies.
16. SELF-SELECTION BIAS
May occur when study participants themselves decide to
participate in, or to leave a study (based on drug exposure
effects, change in health status of participants, personal
reasons).
So, the association observed in the study sample may not
be representative of the real association in the source
population.
This bias is very important in case–control studies or cohort
studies.
17. Self-selection bias (contd’.)
E.g., Association b/n drugs used during pregnancy and birth
defects
in 3 groups of mothers
• Group-1: mothers of („affected‟ children) who used medications
during pregnancy.
• Group-2: mothers of („normal‟ children) who used medications
during pregnancy.
• Group-3: mothers („normal‟ children) who did not use any
medications during pregnancy
• Group-1 will be more willing to participate in the study than
Groups 2 and 3.
• Solution: Systematically identify and recruit all eligible cases
(for both groups).
18. Losses to follow-up (study participants dropping out) in
longitudinal studies can also induce bias, if those who drop out
belong to the inclusion criteria.
Probable Solution: Use population-based registries
19. PREVALENCE BIAS
A type of selection bias that may occur in case–control studies when
prevalent cases (rather than new cases) are selected for a study.
Prevalence is proportional to both incidence and duration of the
disease (But, it is related more to the duration of the disease
rather than to the incidence).
In a group of incidence cases, significant association with prevalent
cases might not be confirmed.
Recruiting only incident cases with recent documented data is
relevant only to disease incidence, not to prevalence.
Selecting only prevalent cases will not give an accurate description of
the current epidemic scenario.
WHY DO WE NEED BOTH PREVALENT AND INCIDENT CASES IN
SUCH STUDIES?
WHAT IS „POINT PREVALENCE‟ AND „PERIOD PREVALENCE‟?
20. Protopathic bias
Feinstein (1985) – This bias may occur “if a particular treatment or
exposure was started, stopped, or otherwise changed because of the
baseline manifestation (outcome) caused by another disease
or other factor.”
If some other disease or risk factor produces the same
symptoms or signs that the researcher is analyzing, the bias
induced is protopathic.
E.g.,
• Study objective: Association between blood in stool as an indicator
for
colorectal cancer
• BUT… excessive use of aspirin can also cause blood in stool
• Haemorrhoids can also cause blood in stool.
21. Information and Misclassification Bias
Errors can even occur if cases in a study are classified with
regard to their exposure and disease status…..
• Unexposed people may be considered exposed and vice-
versa.
• Health status may also be incorrectly classified.
This type of error may lead to a „Misclassification Bias‟.
It equally affects both case–control and cohort studies.
22. Information and Misclassification Bias (contd’.)
Non-differential misclassification:
• When the misclassification error occurs randomly (i.e.,
independent of the knowledge regarding exposure–outcome
relationship).
• Mostly occurs if study instrument is not very reliable;
• Even if study instrument is reliable, it may give erroneous
results if the researcher does not convey the meaning properly
(e.g. to illiterate subjects).
• It may lead to variation (either increase or decrease) in the
strength of the association between exposure (drug) and
outcome (bias towards either the null or alternate
hypothesis)
23. Exposure timing:
Inaccuracy in properly defining the exposure time can result in
information bias leading to…
a) a non-significant association overall, even when there is a very
strong association between the drug and the outcome, within a
specific time window (short time period).
• E.g., Anaphylactic reactions occur rapidly after drug exposure,
very high risk during this short time period, and null after this
initial period.
24. Exposure timing (contd’.)
b) a significant association overall, when there is not a very strong
association between the drug and the outcome, within a
specific time window.
• E.g., In a study aiming to correlate smog exposure and RTIs,
inaccurately stating the time respondents are exposed to smog
can result in misclassification.
25. c) The risk mostly decreases with time (‘survivor effect’).
• E.g., Sometimes, chronic long-term users of NSAIDs are likely
to be at a lower risk of GI bleeding than new users, due to
„survivor effect‟.
d) Sometimes, the risk steadily increases with time, due to the
‘cumulative effect’ of drug exposure
• E.g., risk of myocardial toxicity after prolonged use of
doxorubicin;
• The differences in c) and d) are due to individual variations and
their varied responses to the „exposure‟.
26. Differential misclassification:
When the error is due to being influenced by knowledge of the
exposure (drug / disease) and the outcome status.
• During data collection in case-control and cohort studies,
knowledge of the exposure (drug and/or disease) influences the
quality of the information collected.
2 situations: Differential recall bias and Differential detection
bias
27. Differential Recall bias:
• Mostly seen in cross-sectional and longitudinal studies;
• In case–control studies, cases and controls may have a
selective memory of their past exposures.
• E.g., In studies of birth defects, mothers with an impaired child
may give a more valid and complete report of their exposure to
drugs during pregnancy as a result of devoting more time to
contemplating the cause of the birth defect.
28. Differential detection bias:
• Can affect either cohort or case–control studies.
• In case–control studies: occurs when the procedures for exposure
assessment is more thorough among potential cases than controls.
• In cohort studies: occurs mostly due to difference in the follow-up for
detecting adverse events.
• E.g., Study Objective: Association between postmenopausal
hormonal supplements and risk of CV diseases and/or cancers
(breast or endometrial)
• The respondents are sourced by sources like online media, AV
advertisements, or word-of-mouth.
• Women taking postmenopausal hormonal supplements are likely to see
their doctors more often than other women. They are more likely to be
examined for breast or endometrial cancer, or risk of CV disease.
• This may lead to an excess number of diagnosed diseases in the „cases‟
group (women who took postmenopausal hormonal supplements) and a
falsely elevated risk.
29. References:
Textbook of Pharmacoepidemiology; 2007 ed. Editors:
Brian L. Strom and Stephen E. Kimmel. Publishers: John
Wiley & Sons Ltd. The Atrium, Southern Gate, Chichester,
West Sussex PO19 8SQ, England(England) ; ISBN 978-0-
470-02925-1
Understanding Pharmacoepidemiology; 1st Edition. Eds.
Yang Y, West-Strum D. McGraw Hill, New York, 2011.
Modern Epidemiology; 3rd Edition. Eds. Rothman KJ,
Greenland S, Lash TL. Lippincott Williams & Wilkins,
Philadelphia, 2008.