Arriving at Correct Conclusions

Abstract

Declaring that a causal and not solely a correlative relation exists between a risk factor and a disease creates significant implications for patients and physicians. No matter the forum, when investigators or clinicians make such a claim, it is essential to explain how this determination was made so that appropriate recommendations are made in all areas of our professional practice. When we review the medical literature it is similarly crucial to understand this distinction between causality and association. The Bradford Hill criteria of strength of association, consistency, temporality, biological gradient, biological plausibility, coherence, experimental evidence, and analogy can be used to help establish causality. It is also important to understand the distinction between clinical and statistical significance to complete our appraisal of the implications of a clinical study. Statistically significant results, although not the result of chance, may be clinically insignificant. Statistically insignificant results, conversely, may not exclude the possibility of a clinically important relation. This article reviews the concepts of causality and association and clinical versus statistical significance and provides examples from the literature.