Research Methodology and Biostatistics Series VI - Observational Studies

Studies on antecedent-outcome relationships in naturally occurring events are called observational studies. Hazardous ‘interventions’, such as smoking and anaemia, are particularly suitable for observational studies because these cannot be intentionally introduced but occur naturally In a prospective observational study, subjects with specific antecedents or exposure are followed up for the occurrence of the outcome. This design yields more valid results because the subjects are monitored over time. A retrospective study starts with the outcome—for example, individuals with and without a disease—and investigates previously occurring events that might have led to the disease in some and not in others. This type of study can be completed more quickly but may involve biased assessment of the antecedents. A crosssectional study requires a carefully chosen, adequately representative sample and gives valid results when both antecedents and outcomes are properly assessed at the same time. The merits and demerits of these designs are discussed in this article. This article also describes some other observational study designs such as case-control, nested case-control, and retrospective cohort studies, which are sometimes misunderstood.