What is a cross-sectional study?

According to STROBE, in a cross-sectional study everyone in the sample is examined at one and the same moment. Setia adds that, unlike cohort and case-control designs, participants are chosen only by the inclusion and exclusion criteria, not by exposure or outcome, and that the design serves population surveys and prevalence estimates in clinic-based samples.

STROBE notes that a subset of cross-sectional studies is analytical: they can compare disease prevalence between exposed and unexposed groups, or compare the odds of exposure between people with and without the disease. The same paper names time order as the difficulty, since it is hard to establish that exposure preceded disease, although the order is sometimes clear: if the exposure is congenital or genetic, for example, the sequence can be trusted even when both are measured at once.

Setia says that a single measurement makes causal conclusions difficult. In Setia’s example, people who are overweight may already have increased their exercise or changed their diet, so a survey can show them exercising more and eating more salads, which misreads the direction of the association. Setia also points out that prevalence depends on how long people survive with the condition as well as on new cases, so cross-sectional surveys alone may not explain disease trends. Setia calls these studies relatively fast and inexpensive.