A Population Snapshot: How to Design a Valid Prevalence Study
Design cross-sectional studies with valid sampling, prevalence estimation, prevalence ratios and appropriate limits on causal interpretation.
What is a cross-sectional study?
A cross-sectional study measures exposures and outcomes in a defined population and time window. It is useful for prevalence estimation, service planning and exploratory association analysis. Because temporal ordering is often unknown, causal conclusions are limited.
Sampling comes first
Valid prevalence requires a suitable sampling frame, not only a large sample. Convenience samples, clinic-only recruitment and non-response can produce serious selection bias. Survey weights, clustering and stratification must be reflected in analysis when complex sampling is used.
Prevalence and association
Report point or period prevalence with a confidence interval and a clearly defined denominator. For common outcomes, odds ratios can overstate prevalence ratios. Robust Poisson regression with a log link is a practical option for estimating prevalence ratios; prevalence differences can also be useful.
Reverse causation
An association between physical activity and depression measured at the same visit cannot establish which came first. Longitudinal or experimental designs are required when temporality is central.
Common mistakes
Generalising a convenience sample to the national population.
Ignoring non-response and missing data.
Interpreting odds ratios as risk ratios.
Making causal claims from simultaneous measurements.
Ignoring the complex survey design.
References
Accessed 20 June 2026.