Observational studies occupy a prominent place in clinical and public health research. Cohort, case-control and other observational studies regularly contribute key aetiological, clinical and public health knowledge, and most research papers published in medical journals describe observational studies.1 Sexually transmitted infections are no exception. For example, in the early 1980s, before the human immunodeficiency virus (HIV) was discovered, a case-control study identified a higher number of sexual partners as a risk factor for Kaposi sarcoma in men who have sex with men.2 Later, a cohort study of female sex workers in Kinshasa showed that gonorrhoea and genital Chlamydia infection were strongly associated with HIV-1 transmission.3 A possible protective effect of male circumcision on the risk of HIV infection was first suggested in 1986, based on biological reasoning.4 Over subsequent years, the association between male circumcision and HIV infection was examined in a large number of observational studies, including ecological, cross-sectional, case-control and cohort studies, which taken together showed a substantial reduction in risk, similar to that observed in the recent randomised trials.5

The reputation of observational epidemiology is nevertheless mixed: it is not uncommon for observational research to produce contradictory, or even spurious, findings.6 7 Conflicting results from epidemiological studies, often of exposures related to diet or other life style factors, provide a constant source of anxiety for the public.8 The credibility of observational study (and any other research) depends on its …