Two individuals trained and supervised by the study's senior
investigators coded the records. The investigators and coders held consensus meetings biweekly to review the coding process and to resolve ambiguities. Coders reviewed data for consistency when coding rules were revised.