Cross-Sectional Study
A snapshot — measure exposure and outcome at the SAME moment, in one population, with no follow-up.
An observational study that assesses exposure and outcome simultaneously, at a single point in time, in a defined population — a freeze-frame. Because it counts existing cases, its natural output is prevalence (the proportion who have the outcome now), not incidence.
One snapshot, both measured together → prevalence, but timing is unknown.
A population photographed once: who is exposed and who has the outcome, side by side. You can describe how common the outcome is and whether exposure and outcome co-occur — but the single time-point gives no information about which appeared first.
Read it as a still image, not a film. An association tells you exposure and outcome travel together at that instant. Because nothing was followed over time, you cannot say the exposure preceded the outcome — the link may run the other way (reverse causation).
Cross-sectional designs are quick, cheap and the workhorse for prevalence surveys and diagnostic-accuracy studies (sensitivity, specificity, PPV/NPV). But the missing time axis means they cannot establish temporality, so they are an unsafe basis for claiming causation.
- Single time-point → gives
prevalence, not incidence - Good for prevalence &
diagnostic accuracy - No temporality →
causation unsafe
National prevalence surveys — a health survey samples a population once and asks, for example, what proportion of adults currently have hypertension, asthma or diabetes. That single snapshot is exactly how a country tracks the burden of a condition and plans services — and the same design underpins ED diagnostic-accuracy studies, where a test and the reference standard are applied to the same patients at one visit. It can tell you how common a condition is and whether a test performs well — it cannot tell you what causes the condition, because nothing was followed over time.
- Reverse causation — assuming exposure caused the outcome when the outcome may have changed the exposure.
- Reporting incidence from a cross-sectional study — it yields prevalence only (no new cases counted over time).
- Neyman (prevalence–incidence) bias — rapidly fatal or quickly-resolving cases are under-represented in a one-off snapshot.
Quick check
A cross-sectional study finds factor A is associated with disease B. Is the relationship causal?
Answer: No — exposure and outcome were measured at the same time, so temporality is unknown. You cannot tell whether A caused B, B caused A (reverse causation), or a confounder drives both.
Ready to build your plan? EMF Premium gives you all 40,000+ questions, 20 mocks and 1,215 OSCE stations from £29/month — or a one-off 3- or 6-month pass.