Observational Studies

#Follow the design, not just its label

Observational studies examine exposures and outcomes without the researcher assigning the exposure being studied. STROBE covers three common designs: cohort, case-control and cross-sectional studies. The design describes how the comparison was assembled, not whether its conclusion is correct.

In a cohort study, people are followed over time and outcomes are compared across exposure groups. Read when exposure information was collected, when follow-up began and how outcomes were recorded. STROBE warns that words such as "prospective" and "retrospective" have different uses; an account of the actual timing is more informative.

Evidence: STROBE: observational reporting explanation and elaboration

#Cases, controls and snapshots

A case-control study compares exposures among people with an outcome and controls drawn from the population that produced those cases. The source of the controls matters: a convenient but unsuitable comparison group can change the association being estimated.

A cross-sectional study assesses a sample at a point in time, often to describe how common an exposure or condition is. It may be hard to establish whether exposure preceded disease. This is a limitation to investigate, not a rule that timing can never be known; STROBE notes that some exposures clearly predate the outcome.

Evidence: STROBE: observational reporting explanation and elaboration

#An association needs alternative explanations

A difference between exposed and unexposed groups may reflect the exposure, other differences between the groups, or how participants and measurements were selected. In non-randomised treatment research, illness severity can influence both which treatment is received and the later outcome. Cochrane calls this confounding.

Statistical adjustment can address measured confounders, but unmeasured factors, measurement error and an unsuitable model can leave residual confounding. An adjusted association is not automatically a causal effect. Read which factors were considered and why, rather than treating "adjusted" as a guarantee.

Evidence: Cochrane Handbook, Chapter 25: bias in non-randomized intervention studies / STROBE: observational reporting explanation and elaboration

#Transparent reporting helps you judge the study

Look for eligibility rules, the source population, exposure and outcome definitions, follow-up, missing information and the analysis used. These details help explain what the comparison can and cannot answer. For intervention studies, Cochrane separates bias in a result from whether that result applies to other populations.

STROBE is reporting guidance, not a prescription for conducting research or a tool that certifies study quality. Use it to find information needed for appraisal. This page introduces reading questions; it does not perform a formal risk-of-bias assessment or decide care for an individual.

Evidence: STROBE: observational reporting explanation and elaboration / STROBE: aims and limits of reporting guidance / Cochrane Handbook, Chapter 25: bias in non-randomized intervention studies

Source note

The sections above were checked against the linked sources. No clinical review has been performed. This is general research education, not a clinical guideline.