Lecture 7, part 3

Question 1

More details about misclassification bias

Answer 1

Error in the classification of the exposure or the dz
May be differential or non-differential

Question 2

How to create comparability with respect to tx experience in experimental studies

Answer 2

Use of placebo

Question 3

How to create comparability in cohort studies with respect to the exposure experience

Answer 3

Not possible

Question 4

How to create comparability in experimental studies with respect to participants

Answer 4

Randomization

Question 5

How to create comparability in cohort studies with respect to participants

Answer 5

Choice of comparison group

Question 6

How to create comparability in experimental studies with respect to post-randomization outcome observations

Answer 6

Masking/blinding

Question 7

How to create comparability in cohort studies with respect to outcome observations

Answer 7

Use objective measures
Mask outcome assessors to exposure status

Question 8

How are subjects followed in a cohort study after baseline info is collected?

Answer 8

Longitudinally over a period of time (days, months, yrs)

Question 9

What can investigators track in following the cohort?

Answer 9

Details of exposure changes
Development of outcome
Loss to f/u

Question 10

What calculations can be made in the f/u of the cohort?

Answer 10

Incidence rates for each of the exposure groups
-Numerator: number of events
-Denominator: dz-free observation time for each subject

Question 11

Problems with loss to f/u

Answer 11

Reduces sample size
Introduces bias

Question 12

How is a reduction in sample size a problem?

Answer 12

Reduces ability to detect an association

Question 13

How does LTFU introduce bias?

Answer 13

Participants LTFU may differ from those who remain in study (esp an issue if losses related to exposure, outcome, or both)
If large LTFU, validity of study results may be severely compromised

Question 14

How to encourage continued participation

Answer 14

Maintain regular contact (via mailings, personal contact, study newsletters, etc)
Offering incentives for study participation
Length of study influences retention strategies

Question 15

How to locate lost participants

Answer 15

Collection of baseline info to help locate participants as study progresses
Contact friends, relatives, medical providers
Vital stats records, driver’s license, voter registries, USPS/phone directories
Multiple sources of info can be used to obtain complete f/u info

Question 16

Goal of analysis of cohort studies

Answer 16

Compare dz occurrence in the exposed and unexposed groups

Question 17

Measures of analysis in cohort studies

Answer 17

Cumulative incidence or incidence rates

Question 18

Unit used in analysis

Answer 18

Often person-years of observation

Question 19

What should be controlled for in analysis?

Answer 19

Known or potential confounding via stratification and statistical analysis

Question 20

Advantages of cohort studies

Answer 20

Facilitate study of rare or unusual exposures
Can study multiple outcomes/effects of a single exposure
Can directly estimate incidence rates
Temporal sequence of exposure to dz relation is clear
Bias is less of a problem than in case-control or cross-sectional studies

Question 21

How is bias less of a problem in cohort studies than in case-control or cross-sectional studies?

Answer 21

Avoid selection bias at enrollment
Can decrease bias by selecting objective measures

Question 22

Disadvantages of cohort studies

Answer 22

Inefficient for rare outcomes because you may not observe outcome/dz
Can be very expensive and time-consuming
Need good records and comparison group
-Esp important if retrospective
LTFU
Changes in exposures and/or diagnostic methods over time may bias results

Question 23

Why can cohort studies be expensive and time consuming?

Answer 23

Large sample size
Multiple sites and researchers involved
Long f/u periods

Question 24

High costs in cohort studies are associated with what?

Answer 24

Prospective designs
Low incidence
Long induction periods

Question 25

Lower costs in cohort studies are associated with what?

Answer 25

Using existing monitoring systems
Using historical cohorts (retrospective designs)
Using general pop as non-exposed