Problems and proposals for interview data in epidemiological research.

We assessed the reliability of epidemiological data obtained by interview of 120 patients in a case-control study. The collected data, which were obtained by interview on two separate occasions, included such clinical and pharmaceutical features as history of lactation, hysterectomy, diabetes, type of menopause, and whether a woman had ever used exogenous oestrogens. Although we found generally high rates of agreement between interviews, errors in collecting and classifying data did occur, and were especially common for complicated clinical events, such as whether an oophorectomy accompanied the surgical removal of the uterus. Patients were also likely to disagree with previous responses when asked to recall a drug exposure occurring many years before. We identified seven sources of this variability, five in collecting the data, and two in coding. As a result of these findings, strategies are proposed for improving the quality of interview data obtained in epidemiological research.