the common objective between cohort and case control studies
to investigate the association between exposures and disease
cohort characteristics
follows a group moving forward
start with groups by exposure status
key aspect is they are followed over time to see what their outcome is
confident in temporality sequence (exposure comes first)
what are cohort studies good at measuring
incidence rates, and multiple outcomes from a single exposure
when to use a cohort study
when an exposure is RARE (key), but disease rate is higher in those exposed
When there is good evidence of both the exposure and the disease of interest
When the cohort is stable and follow-up is feasible.
When you want to study multiple outcomes from a single exposure.
When ample resources and time are available.
an epidemiological factor that cohort studies are good for
temporal
direction is always exposure-> outcome
types of cohort studies
Prospective (concurrent):
Start now, follow into the
future
Retrospective (historical):
Use past records to follow up
to present
Ambidirectional:
Combination of both
Comparison groups
Internal comparison: one cohort with subgroups by exposure status (e.g., smokers vs. non-smokers)
External comparison: two or more cohorts compared
General population (as a study group)
- Whole population in a defined area
- Representative sample
types of outcomes
discrete events (single i.e. mortality/first diagnosis, or multiple i.e. recurrent disease episodes)
continuous outcomes (repeated overtime i.e. blood pressure)
composite outcomes (combination of multiple endpoints into a single measure)
Special groups
(as a study group)
- Occupational/professional groups (e.g., Nurses’ Health Study)
- Geographically defined groups (e.g., Framingham Heart Study)
- Risk groups (e.g., San Francisco Men’s Health Study)
- Exposure-defined groups — physical, chemical, biological
agents or specific events (e.g., Japanese Atomic Bomb
Survivors, 9/11 FDNY Workers Cohort
why does follow up matter
Ensures accurate outcome ascertainment
Minimizes loss to follow-up, which can bias results
Maintains study validity and statistical power
sources of outcome data
Active follow-up
* Mailed questionnaires
* Telephone calls or personal interviews
* Periodic medical examinations
Passive follow-up
* Linkage to medical records, hospital discharge databases, death registries
Cohort Studies
how can we get bias?
loos to follow-up/attrition bias (a form of selection bias)
information bias (misclassification of exposure, misclassification of outcome)
confounding
measures of association (cohort studies)
Mostly incidence rates:
Risk ratio (from cumulative incidence)
- Rate ratio (from incidence rates)
- Risk/rate differences
- Hazard ratio (from time-to-event/survival) analysis)
- Odds ratio (possible, but mainly for rare outcomes or logistic regression
measures of frequency (cohort studies)
Cumulative incidence (risk)
* Incidence rate (person-time rate
case-control study features
Compare groups with a known outcome (cases) and without it (controls)
Look backward in time to determine prior exposure status
Measure and compare the odds of exposure between cases and controls
Exposure status is determined after the outcome has occurred → temporal sequence is less
certain
Can study multiple exposures for the same outcome
Efficient for rare outcomes or those with a long latency period
case-control direction of time?
backwards
when to use a case control study
When the outcome/disease is rare (low incidence in the population)
Long induction or latent period (e.g., cancer, cardiovascular disease)
When little is known about the disease and you need to explore potential risk factors (e.g., early
studies of AIDS).
When resources, time, or funding are limited (more efficient than a prospective cohort study)
When exposure data are expensive or difficult to obtain prospectively.
When rapid, early evidence is needed to guide public health action.
When studying multiple exposures (risk factors) for the same outcome.
types of case control study
Population-based: Cases and controls are drawn from the same well-defined population (e.g.,
residents of a city, members of a health plan).
- Hospital-based: Cases and controls selected from patients in the same hospital or clinic.
- Nested case-control : Cases and controls identified from within an existing cohort study.
- Case-cohort: Controls are a random sample of the original cohort, compared with all cases in
the cohort. - Matched: Each case is paired with one or more controls with similar characteristics (e.g., age,
sex, location) to reduce confounding. - Unmatched: Cases and controls selected independently without pairing on specific variables.
Case-Control Studies
what has to happen when determining the case group for case-control studies
Use precise, consistent diagnostic criteria to avoid misclassification.
describe the control group in case control studies
Individuals from the same source population that produced the cases, but who do not have the
outcome of interest.
Purpose: To estimate the exposure distribution in the population from which the cases arose.
types of controls
Population controls – e.g., community sampling, voter lists, registries; ensure same base population
as cases.
Hospital controls – patients admitted for conditions unrelated to the exposure of interest (e.g.,
orthopedic or surgical admissions in a smoking–MI study).
Friends/relatives – similar background and environment; relatives help control for genetic
confounding
§ Counting, which is a key activity of epidemiologists, includes three steps
□ (1) developing a definition of disease
□ (2) instituting a mechanism for counting cases of disease within a specified population
□ (3) determining the size of that population.
Majors forms of depression treatment
cognitive-behavioural therapy
Interpersonal psychotherapy
pharmacotherapy
