# Observational study

An observational study measures what people already do instead of assigning them to do it. Researchers record exposure and outcome and never intervene. Three designs are common: the cohort study, which follows exposed and unexposed groups forward in time; the case-control study, which starts from people who already have the outcome and looks backwards; and the cross-sectional study, which measures both at one moment. Reporting follows the STROBE statement. Because assignment is not random, the groups differ in ways that also affect the outcome. Statistical adjustment handles the confounders that were measured and nothing else. Hormone therapy is the standard cautionary case. Observational data suggested a large drop in coronary heart disease, yet the Women's Health Initiative trial in 16,608 women found a hazard ratio of 1.29 and stopped early. Later reanalysis showed that most of the gap came from comparing women who started therapy at different points after menopause, not from an unmeasurable bias. Target trial emulation formalises that repair: describe the randomised trial you would have run, then build the analysis to match it. Observational work stays indispensable in longevity and nutrition, where randomising decades of diet is impossible, but a bare association is a hypothesis rather than a finding.

## Sources

- von Elm E, Altman DG, Egger M, et al.. (2007). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. The Lancet. https://doi.org/10.1016/S0140-6736(07)61602-X
- Rossouw JE, Anderson GL, Prentice RL, et al.. (2002). Risks and Benefits of Estrogen Plus Progestin in Healthy Postmenopausal Women: Principal Results From the Women's Health Initiative Randomized Controlled Trial. JAMA. https://doi.org/10.1001/jama.288.3.321
- Hernán MA, Alonso A, Logan R, et al.. (2008). Observational Studies Analyzed Like Randomized Experiments: An Application to Postmenopausal Hormone Therapy and Coronary Heart Disease. Epidemiology. https://doi.org/10.1097/EDE.0b013e3181875e61
- Hernán MA, Robins JM. (2016). Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. American Journal of Epidemiology. https://doi.org/10.1093/aje/kwv254

---

_Canonical: https://longevity-germany.com/en/glossary/observational-study · Part of Longevity Cities · Updated 2026-08-25_