Reliability of data on medical conditions, menstrual and reproductive history provided by hospital controls

To assess the reliability of data on medical conditions, and menstrual or reproductive history, a sample of 294 controls interviewed in hospital between 1989 and 1992 for an Italian case-control study on digestive tract neoplasms was re-interviewed at home during 1993. A high agreement between responses at the two interviews (κ ≥ 0.85) was observed for most medical conditions, including diabetes, cholelitiasis, hepatitis, duodenal ulcer, and, among female conditions, uterine fibromas, benign breast disease, hysterectomy and monolateral ovariectomy. For gastric ulcer and parotitis the reliability was less satisfactory (κ = 0.35 and 0.20, respectively). The agreement was high (κ > 0.80) also for age at menarche, menopausal status, type and age at menopause, number of children, age at first pregnancy, age at first and last birth, and spontaneous abortions. The agreement was lower for questions on menstrual pattern (κ = 0.68) and induced abortions (κ = 0.62). Thus, this study indicates that information on personal medical conditions, and menstrual or reproductive history, provided by hospital controls through an interviewer-administered questionnaire is satisfactory for the purposes of epidemiological inference, and that the interview setting does not substantially influence the recall of this information.