Cohorts:
- Fixed
- Dynamic
Fixed Cohort: no individuals enter the population after the start of follow up
Dynamic Cohort: individuals may enter the population any time during the follow up period
Stable /Stationary Population: if the size and distribution of characteristics remain constant during the follow-up period.
Transformation of a Dynamic Population
-If f/u information is available on every member of a dynamic population, we can artificially transpose it to a fixed cohort.
-Often needed for analyzing data
-Pool all the individuals at time 0
Assumption = lack of secular trends regarding characteristics affecting outcome of interest
-Secular Trends
- Change in incidence over time that is caused by environmental factors, host susceptibility, etc (i.e., vaccine development)
- A change in a secular trend = change in characteristics of recruits, significant changes in relevant exposures / treatments due to environmental, governmental, etc factor
- May introduce bias
Cohort study
-Selecting the Unexposed group
- The unexposed group needs to be AS similar as possible to the exposed group with respect to confounders
- if there is no relationship between E + O, the rate of disease in the 2 groups should be equal
COUNTERFACTUAL IDEAL - the ideal unexposed group = exactly the same individuals in exposed group had they not been exposed (impossible)
-Researchers do their best to eliminate/reduce bias so that the control group is as similar as possible to the exposed group
Measurement of Exposures
- Direct
- Surrogate
Direct measurement - Biological measurements, air sampling, water sampling, subject interviews, use of proxies
Surrogate measurement - Work records, questionnaires, medical records
*measure of exposure takes place at baseline only or throughout the follow up period
Steps - Conducting a Prospective Cohort Study
- Define cohort, screen, invite eligible subjects
- Obtain baseline exposure measurements
- Follow cohorts for disease
- May obtain additional exposure measurements over time (ex, dietary intake)
- Analyze disease risk according to exposure status and estimate associations
Steps - Conducting a Retrospective Cohort Study
- Define cohort - ex, all employees who worked >1 yr between Jan 2005 - Dec 2011
- Obtain retrospective exposure measurements
- review work records for job descriptions, survey subjects, etc. - Obtain retrospective disease/mortality data for the period of the study
- Mortality data = death certificates
- Morbidity data = interview, records of hospital dx - Analyze outcome frequency by cohorts and estimate association
Selecting Study Subjects
Obtaining Exposure Data
Subject Selection:
- general population: whole population in an area, or representative sample
- special group: occupational group/professional group, exposure groups, as in exposure to physical/chemical agent
Obtaining exposure data:
-personal interviews, mailed questionnaires, review of records, medical exams, environmental surveillance
- use info to classify cohorts as exposed / non exposed and by degree of exposure (sub-classification of cohorts)
Temporal Sequence of Disease: Length of Follow-Up
Induction period - time between exposure to a specific risk factor and the initiation of the disease
Latent Period - time between biologic onset of disease and disease detection - as in, appearance of sx, or positive dx tests
- hard to separate in research practice - commonly use “empirical induction” = induction + latency (?)
- Follow Up
- Analysis
Follow -Up:
- Obtain data about outcome (disease or death)
- Some loss to f/u = inevitable, due to death, migration, change of occupation
- Drawback to cohort studies
Analysis:
- Calculation of Incidence Rates among cohorts
- Estimation of Risk/Rate ratio
Strengths of a Cohort Study
- Less Temporal Ambiguity: exposure status is measured before disease is detected
- Disease status cannot influence measurements of exposure status, esp. in prospective
- Factors associated with selection of subjects can’t bias exposure effect estimate
- Can study several diseases/outcomes for each exposure - and certain outcomes may be ID’d after start of follow up
* Prospective Cohort Studies = gold standard of observational studies
Limitations of a Cohort Study
- Expensive, time consuming due to long follow up periods
- Inefficient - statistically and economically, for studying rare disease or diseases with long latent periods
=large study populations required for sufficient power - Loss to F/U reduces the effective sample size and can lead to biased exposure effect estimates
