Testing a tool for assessing the risk of bias for nonrandomized studies showed moderate reliability and promising validity.