abstract
- The study of causal relationships is important when addressing questions of efficacy of treatment interventions and etiology of disease. The evaluation of a cause-and-effect relationship between exposure to a putative causal factor and outcome can be undertaken using a variety of study designs including randomized controlled trial and cohort and case control studies. Study participants should be selected in a manner that minimizes bias and confounding and is representative of the target population. Confounding can be controlled by using several strategies including restriction, randomization, stratification, matching, and multivariable analyses. The degree of association is then summarized by the relative risk for prospective studies and the odds ratio for retrospective studies. The precision of these estimates should be indicated by providing their confidence intervals. Important indicators of causation are correct temporal and dose-response relationships between exposure and outcome, a large magnitude in the strength of association, and consistency and specificity of association. Biological and epidemiological sensibility and analogy to other well-established relationships provide additional support for a causal hypothesis.