In order to validate a computerized screening method, this study determined whether ICD-9 codes used to identify complications are coded accurately and if the screening method appropriately distinguished a preexisting condition from one that developed during hospitalization. Investigators compared coding from Medicare claims with those independently reviewed for more than 1200 medical records in two states. Discussion includes full presentation of the findings for each diagnosis code as well as the estimated predictive values of the computer screening method in correctly identifying hospital complications. While the authors suggest that their method has validity for capturing most surgical complications, the screening tool fails to adequately identify enough events. They propose the addition of a timing indicator to Medicare claims to improve detection of hospital complications and validate this important quality measurement tool.