Biases & Misclassifications

Question 1

Bias

Answer 1

Systematic (non-random) error in study design or conduct leading to erroneous error

Distorts the relationship b/w exposure & outcome

Question 2

What are the 2 main categories of Biases?

Answer 2

1) Selection-related: how did the patients in the study get elected?
2) Measurement-related (information/observation: how does the researcher get info?

Question 3

Selection Bias

Answer 3

How study subjects are selected generates differences in groups OR a study population does not adequately represent the population (so they’re not representative
(Selection related category)

Question 4

Healthy-worker bias

Answer 4

Can easily be seen in prospective cohort studies - when reporting only on the people who are currently working, but not questioning or reporting the people who are not working
(Selection-related category)

Question 5

Self-Selection/Participant (Responder) bias

Answer 5

Ppl who wish to participate/respond (volunteer) may be different in some way to those that don’t volunteer of self-select (non-responder bias) to participate
(Selection-related category)

Question 6

Recall (reporting) Bias

Answer 6

Exposed or dz subjects may have a greater sensitivity for recalling their hx (better memory, easier to remember is more severe) or amplify/exaggerate their responses

(Measure/information - subject related category)

Question 7

Hawthorne Effect

Answer 7

Ppl who modify their behavior bc they are a part of a study and know they are being watched

(Subject related - Measurement/information/observ category)

Question 8

Compliance/adherence bias

Answer 8

Groups being studied have different compliance/adherence with study protocol/tx; so groups are not equal bc of their compliance

(Subject related- info/measurement category)

Question 9

Lost-to-follow up/Withdrawal (Differential Attrition) Bias

Answer 9

Difference in groups due to participants withdrawal or difficulty in follow up rates OR there are differences in those that stay in the study and those who quit

(Subject related- info/measurement category)

Question 10

Contamination Bias

Answer 10

Contaminating placebo/control pt with a similar drug as to the tx group, so they are not truly untreated or unexposed

(Subject related-info/measurement category)

Question 11

Interviewer/Proficiency Bias

Answer 11

A difference in collecting data on part of the researchers OR
Interviewers knowledge may influence structure or tone - which may influence the response OR
Interventions/tx are not applied equally b/w groups
- conscious or unconscious actions of interviewer

(Observer related -info/measurement category)

Question 12

Diagnosis/Surveillance (Expectation) bias

Answer 12

Expectations of their pt’s responses OR
Different evaluation, classification, dx, or observation b/w study groups

(Observer-related -info/measurement category)

Question 13

Lead-time bias

Answer 13

A benefit from healthcare screening due to early detection of dz despite an unchanged clinical outcome or timeline 
Extra time (knowing patients have the dz) afforded by early detection (screening) = lead time 
But they may be finding ppl who already had the dz and the only thing keeping them alive is the intervention 

(Screening related -info/measurement category)

Question 14

Publication bias

Answer 14

Only publishing results that are positive or desired by public, rather than including all messages (even the neg ones)

Question 15

Misclassification Bias

Answer 15

Error in classifying either dz and/or exposure status)
2 forms = Non-differential and differential

(Measurement/information bias category)