Childbirth research data: medical records or women's reports?

Medical records and women's reports were compared as sources of data for childbirth research. Three weeks after they had given birth in 1982 at five teaching hospitals in Sydney, Australia, 397 low-risk primiparous women in a random sample were interviewed about their birth experiences. The women's reports were compared with data from their medical records. Error sources in data collection were identified at four points: from the actual event to hospital recording, in the abstraction of data from medical records, in women's memory of the actual events, and in women's reporting of their encoded information. Both corrected data sources were accurate for most major variables. It is concluded that both data sources are subject to variation from the actual events they represent and that the assumption that medical records are always more accurate and acceptable is not supported.