There may be residual confounding, because women who undergo surgery may have differed from women in the control cohort with respect to other factors not accounted for in the analyses. Also, there is a possibility of chance findings, since we investigated multiple outcomes.