Actual versus actuarial analysis for cardiac valve complications: the problem of competing risks

Methods for analyzing rates of events such as heart valve failure following surgery are important for comparing different techniques and devices; howe ver, in patients undergoing major surgery, other risks such as mortality co mpete with the risk of heart valve failure to determine each patient's fina l outcome. When multiple, mutually exclusive endpoints are possible, a situ ation known to statisticians as a competing risks problem arises. No single statistical technique that is currently available provides an entirely sat isfactory solution to this problem. We argue that in order for valve failur e incidences to be useful clinically, the overall patient outcome milieu fr om which these failures arise must be considered. In this article, we revie w recent work in the area of competing-risks analysis as it pertains to hea rt valve surgery outcome.