Abstract:
|
The strength of evidence provided by epidemiological and observational studies is inherently limited by the potential for unmeasured confounding. Thus, we would expect every observational study to include a quantitative sensitivity to unmeasured confounding analysis. However, we reviewed 90 recent studies with statistically significant findings, published in top teir journals. We found 41 mentioned the issue of unmeasured confounding as a limitation, but only 4 included a quantitative sensitivity analysis. This disparity led us to develop a universal figure that can be referenced by researchers wanting to include a simple rule out sensitivity analysis. This figure takes three quantities into account: the bound of the confidence interval for the observed exposure's effect closest to the null, the strength of the association between an unmeasured binary confounder and the outcome, and the differential prevalence between the exposed and unexposed populations needed to nullify the statistically significant effect. These figures can be utilized both by researchers and by readers wishing to understand the sensitivity of studies that failed to include an informative sensitivity analysis.
|