Bias

Front 1

Selection bias:

Back 1

Concern is that subjects have different probability of being selected according to exposures or outcomes of interest, creating a biased measure of association.

Front 2

3 points relating to selection bias:

Back 2

Systematic difference in characteristics between those who are selected and retained for study and those who are not.
When characteristics of those chosen for the study differ systematically from those in the target population.
When the study and comparison groups are selected from different populations.

Front 3

Name the common types of selection bias:

Back 3

Volunteer bias.
Response bias.
Berkson's bias.
Prevalence-Incidence bias.
Exclusion bias.
Healthy worker effect.
Losses to follow up.

Front 4

Volunteer bias:

Back 4

People who volunteer for studies are different from those who don't.
Volunteers are likely to be better educated and more health conscious.
May be associated with behavioural and lifestyle exposure levels.

Front 5

Response bias:

Back 5

A type of volunteer bias. When those who respond to a questionnaire are systematically different from teh non-responders.

Front 6

Berkson's bias:

Back 6

Affects case control studies that use hospitalized controls and/or cases. Combination of exposure and outcome may increase the probability of hospitalization. If hospital based controls/cases have different exposures than population based controls, OR will be biased (over or underestimated).

Front 7

Solution to Berkson's bias:

Back 7

Select controls from teh general pop or people whose disease is unrelated to exposure.

Front 8

Prevalence-Incidence bias

Back 8

Affects cross-sectional and case control studies.
Study selection may exclude those with mild disease ( resolved by sampling time) / those who died quickly.
Consequently, the 2 may select disease survivors. - prevalent cases.

Front 9

Exclusion bias:

Back 9

Different exclusion criteria are applied to cases and hospital-based controls.

Front 10

Solution to exclusion bias:

Back 10

Exclusion criteria should be related to reason for admission to the hospital rather than prior history of disease.

Front 11

Healthy worker effect:

Back 11

Affects occupational cohort studies. Outcomes in workers are compared to the general popualtion. Workers tend to be healthier. The general pop includes healthy and un healthy people. Likely a negative bias will result.

Front 12

Losses to follow up:

Back 12

Affects cohort and experimental studies. Final sample could end up being very systematically different from original target pop.If probability of LTFU is greater for certain exposure and outcome groups, association will be biased.

Front 13

Minimising selection bias:

Back 13

Need for an accurate smapling frame to select a representative sample of the target population.
Diligent efforts to maximize response rate and prevent loss to follow up.

Front 14

Information bias:

Back 14

Error in classifying subjects with regard to their exposure or outcome status OR differential quality (accuracy) of information between compared groups.

Front 15

Information bias results in:

Back 15

Misclassification.

Front 16

Synonyms for info bias:

Back 16

Measurement / observation bias.

Front 17

Misclassification of exposure:

Back 17

Recall bias.
Interviewer bias

Front 18

Misclassifciation of outcome:

Back 18

Surveillance bias.

Front 19

Non- differential misclassification:

Back 19

Exposure classification is unrelated to the occurrence or presence of the disease.
Disease classification (ascertainment) is unrelated to exposure.

Front 20

Result o f^:

Back 20

Will always bias results towards the null.

Front 21

Differential misclassification:

Back 21

Exposure is classified differently for those with and without the disease.
Disease is classified differently for those with and without the exposure.
Result: Yields a biased result either away from or towards the null.

Front 22

Name the types of differential misclassifciation:

Back 22

Recall bias, interviewer bias, medical surveillance bias.

Front 23

Recall bias:

Back 23

Affects case control studies.
Cases remember past exposures differently than controls–If they spend more time reflecting on past exposures they will over-estimate exposure status–If they are elderly or have memory problems, they will under-estimate exposure status, particularly for distant events.

Front 24

Interviewer bias:

Back 24

Affects any type of epidemiologic study.
When researchers are aware of the outcome, the awareness may influence how they solicit, record, or interpret exposure information, or vice versa.

Front 25

Medical surveillance bias:

Back 25

If a population is monitored over time, disease ascertainment may be better in the monitored population than in the general population.
If clinical visits (monitoring) are associated with the exposure, sub-clinical cases are more likely to be detected among those with the exposure than those without the exposure. ??

Front 26

Possible solutions to medical surveillance bias:

Back 26

Only use cases and controls that have undergone similar detection procedures.
Concurrent prospective study - ascertain cases as they occur through regular medical surveillance surveillance regardless of exposure status. (Mask exposure status when ascertaining outcome).
Obtain information on type of medical care received to determine whether frequency of medical care may affect diagnosis of disease, can stratify by this variable4) Stratify by disease severity in the analysis (if exposed cases more likely to be diagnosed in early stages).

Front 27

Limits to generalizability:

Back 27

Study restriction and effect modification.

Front 28

EM:

Back 28

Not a source of error, but a natural phenomenon.