Community medicine chapters
Cross-sectional, case-control and cohort studies
The main observational study designs, how each is done, what it measures, its strengths and weaknesses, the biases that affect it, and how to pick the right design from an exam scenario.
- A cross-sectional study looks at exposure and disease at one point in time; it measures prevalence but cannot show which came first.
- A case-control study starts with disease (cases vs controls) and looks back at exposure; it is quick and suits rare diseases, gives the odds ratio, but suffers recall bias.
- A cohort study starts with exposure and follows people forward to see who develops disease; it gives incidence and relative risk and shows temporality, but is slow, costly and loses people to follow-up.
On this page
Overview
In observational studies the researcher observes what happens, without assigning the exposure. In experimental studies the researcher assigns the exposure (see randomised controlled trials).
The main observational designs are ecological, cross-sectional, case-control and cohort studies.
Hierarchy of evidence
From strongest to weakest (for questions about treatment):
- Systematic reviews and meta-analyses of RCTs.
- Randomised controlled trials.
- Cohort studies.
- Case-control studies.
- Cross-sectional studies, ecological studies.
- Case series and case reports.
- Expert opinion.
Descriptive studies
- Case report: one unusual patient.
- Case series: several similar patients; no comparison group.
- Ecological study: compares groups (for example countries or districts) rather than individuals, using routine data. Quick and cheap but prone to the ecological fallacy: a link seen between groups may not hold for individuals.
Cross-sectional study
Design: exposure and outcome are measured at the same time in a defined population (a "snapshot").
Measures: prevalence (of disease, risk factors or behaviours).
Uses: planning services, surveys (for example the Pakistan Demographic and Health Survey, National Nutrition Survey), generating hypotheses.
Survey methods
- Questionnaires (self-administered or interviewer-administered).
- Interviews (structured, semi-structured).
- Examination and measurement (BP, height, blood tests).
- A representative sample chosen by a proper sampling method (see sampling).
- Pilot testing of questionnaires; trained interviewers.
Strengths: quick, relatively cheap; good for prevalence; can study many exposures and outcomes at once.
Weaknesses: cannot show temporality (which came first); not suited to rare diseases; prevalence favours long-lasting cases (survival bias).
A survey of how many adults have hypertension now is a cross-sectional (prevalence) study.
Case-control study
Design: start with people who have the disease (cases) and people who do not (controls); look back to compare their past exposure.
| Cases (disease) | Controls (no disease) | |
|---|---|---|
| Exposed | a | b |
| Not exposed | c | d |
Measure: odds ratio = ad ÷ bc (see measures of risk). It cannot measure incidence.
Choosing controls: similar to cases except for the disease (hospital or community controls). Matching (for example by age and sex) controls for those factors.
Strengths:
- Quick and cheap; needs few subjects.
- Best for rare diseases and diseases with a long latent period.
- Can study many exposures for one disease.
Weaknesses:
- Recall bias: cases may remember past exposures differently from controls.
- Selection bias in choosing controls.
- Cannot measure incidence or relative risk directly; temporality may be unclear.
Rare disease → case-control study. Rare exposure → cohort study.
A nested case-control study takes its cases and controls from within an existing cohort.
Cohort study
Design: start with people free of the disease, classified by exposure (exposed vs unexposed); follow them forward in time to see who develops the disease.
Measures: incidence in each group, relative risk and attributable risk.
Types:
- Prospective (concurrent) cohort: exposure is measured now; people are followed into the future. Example: the Framingham Heart Study.
- Retrospective (historical) cohort: uses existing records to identify past exposure, then traces outcomes up to the present (for example factory workers' employment records). Quicker, but depends on the quality of records.
Strengths:
- Shows temporality (exposure clearly comes before outcome).
- Measures incidence and relative risk directly.
- Best for rare exposures; can study several outcomes of one exposure.
- Less recall bias.
Weaknesses:
- Time-consuming and expensive; large numbers needed (not suited to rare diseases).
- Loss to follow-up (attrition bias).
- Exposure status may change over time.
Comparison
| Feature | Cross-sectional | Case-control | Cohort |
|---|---|---|---|
| Starting point | A population at one time | Outcome (disease) | Exposure |
| Direction | Snapshot | Backward | Forward |
| Measures | Prevalence | Odds ratio | Incidence, relative risk |
| Best for | Planning, prevalence | Rare diseases | Rare exposures, temporality |
| Main problem | No temporality | Recall bias | Cost, time, loss to follow-up |
Bias and confounding
- Bias (systematic error):
- Selection bias: the groups differ in ways other than the exposure (for example healthy worker effect, volunteers).
- Information (measurement) bias: errors in measuring exposure or outcome (recall bias, interviewer bias, misclassification).
- Confounding: a third factor is linked to both the exposure and the outcome and is not on the causal pathway, making an association look stronger or weaker than it is. Example: coffee drinking appears linked to lung cancer because coffee drinkers smoke more; smoking is the confounder.
Confounding is controlled by randomisation (in trials), restriction, matching, stratification and multivariable analysis.
A confounder is associated with both exposure and outcome and is not an intermediate step between them.
Most questions describe a study and ask the design or its measure. Look at the starting point: outcome (case-control), exposure (cohort) or a single time point (cross-sectional). Also expect: best design for rare diseases vs rare exposures, odds ratio vs relative risk, recall bias, loss to follow-up, ecological fallacy, and the definition of a confounder.
Quick revision
- Cross-sectional: prevalence; snapshot; no temporality.
- Case-control: start with disease, look back; rare diseases; odds ratio; recall bias.
- Cohort: start with exposure, follow forward; incidence and RR; rare exposures; costly.
- Retrospective cohort uses past records.
- Ecological: group data; ecological fallacy.
- Confounder: linked to both exposure and outcome.
Written by The House Job from Excel Community Medicine and Park's Textbook of Preventive and Social Medicine (named in the UHS PGET syllabus).
FAQ
Which study design is best for studying a rare disease?
A case-control study.
Which study design measures incidence and relative risk?
A cohort study.
What is recall bias?
A bias in which cases remember or report past exposures differently from controls, common in case-control studies.
What is a confounder?
A factor associated with both the exposure and the outcome that is not on the causal pathway and distorts the apparent association.