Lecture 21 and 22: Cohort Studies Flashcards
What is a cohort?
cohort= a group of subjects (people animals etc) that has a defined characteristic in common
-In epi studies the characteristic of interest is the exposure status
-Selection of study subjects in cohort studies is based on exposure status
What is the general design of cohort studies?
-Groups of disease-free individuals to start
-Classify based on exposure (were they or were they not exposed)
-Follow the pre-determined time period (overtime and sale later)
-Compare development of disease in exposed group to that of non-exposed group
What are the 2 approaches to participant selection?
1 select based on exposure status
2 Select defined population (Disease -) prior to identifying exposure history
What is the difference between prospective and retrospective?
Prospective cohort study (more common)
-follow E+ and E- forward in time from present (t=0) to a pre-determined end point in the future
Retrospective cohort study
-“historical” cohorts established from past records
-Follow-up can be present day OR more recent past records
EX 2004 records-2022 follow up or 2004 records - 2010 follow up
What can we analyze in cohort studies?
-Measures incidence of disease (proportion of those that get disease ie new cases)
-If a positive association exists b/w the E and O we would expect the incidence of disease among the E+ group to be greater than the incidence of disease in E- group (so smokers likely to see more cases of lung cancer than non-smokers)
-Able to calculate ALL MOA and MOE
What are the advantages of cohort studies?
-Good for rare exposures (bc condition is people who are already exposed)
-Can study several outcomes
-Temporal sequence established (prospective) may be more difficult in retrospective studies
-Provides incidence data (new cases)
-No recall bias (with prospective studies)
What are disadvantages of cohort studies?
-Bad for rare outcomes
-Can study only one or a few exposures
-Bias (all study types) selection, info, and confounding
-Often expensive and time consuming
-Can have long follow-up time (loss to follow up or changes in diagnostic methods over time may lead to biased results)
How does selection bias relate to cohort studies?
-loss to follow-up is a major concern especially in prospective cohort studies (can be caused by death. losing track of time, moving etc, if those who left the study are diff than those who remain then will have selection bias and affect results) generally the # of withdrawals increases as the length of the study increases
-Also non-response bias can be an issue if multiple assessments / timepoints are used during follow-up period
How would one address selection bias?
-Increase incentives for participants
-reduce costs incurred by patients
-provide tangible benefits (financial, education etc)
-Provide clear support and care
-ETHICAL NOT COERCIVE
-Can also increase frequency of follow-ups but balance bc dont want to do it too much
How does info bias relate to cohort studies?
Misclassification bias can occur in different forms
-Changes in screening and diagnostic methods over time
-If the quality and extent of info differs b/w exposure groups (especially in retrospective studies that rely on past records)
-Unintentional bias introduced in data analysis and interpretation by person analyzing the data
-Unclear case definitions for E or O status
-Recall bias may be an issue in retrospective cohort studies
How would one address info bias?
-Try to ensure equal follow-up for both groups
-Regular, periodic observation with EQUAL rigour
-Blinding
-Clear case definitions for both E and O status
How does confounding bias relate to cohort studies?
-Need to be aware of and collect info on major confounding variables
-Addressing confounding bias
-exclusion/restriction (study subjects all possess same level if major CFV ie sex, age group)
-matching (id major CFV then select E- subjects so that they have the same level of the CFV as the E+ subjects)
-analytic control (id major CFV and control for them using stratification or multivariable regression) and that they
What are some ways to help ensure validity in cohort studies?
3 major areas where specific attention should be paid to help ensure validity
1. Selection of E+ and E- groups
-Selection is based on exposure status, try to ensure samples are comparable with respect to characteristics other than E factor of interest, representative of target pop, but be wary of confounding
-Careful and explicit definitions
-All subjects must be disease neg at the start
-Can have multiple E- groups and gradient of E+ groups
2. Follow-up of cohorts
-Biggest challenge in cohort studies
-Regular follow-up of BOTH exposure groups
-Strive for minimal, non-differential withdrawal
3. Objective diagnosis of disease status
-Involves the accurate diagnosis of disease
-Clear explicit case definitions
-Blinding with respect to E can again be used in the diagnosis if disease/assessment of the outcome, during the analysis
What is the framingham Study?
-Epi study in 1948, that took place in a small town looking at heart disease and E status among men and women 30-62 years of age
-Was a cohort study using the type 2 selection, bc they selected a defined population prior to identifying exposure history AND they were D- at the seat of the study (a test may be needed to identify) and then put into E+ or E- groups
How long was the framingham study follow up period?
-Initial follow-up period was 20 years, exams were completed every 2 years and daily surveillance of hospitalizations for new coronary events (bc they only had 1 hospital it was easy to follow up)
