The repeatability of self-reported exposure after miscarriage

Background. The Avon Longitudinal Study of Pregnancy and Childhood is a prospective study of women who ware resident in Avon and who were expected to deliver a baby between April 1991 and December 1992. Methods. The study provided an opportunity to test the repeatability of responses from 220 women who experienced a miscarriage and who reported exposure to occupational substances and common household products and appliances in two questionnaires. The first questionnaire was completed in the early part of the pregnancy and the second after the miscarriage. Women were asked to score their frequency of exposure on a five-point scale from 'daily' to 'never'. Their responses were analysed to assess the degree of agreement between replies to identical questions in the two questionnaires using the kappa statistic. A new frequency variable was created which compared the replies for the two questionnaires; this was analysed for all exposures by cross-tabulation with possible explanatory variables (age of mother, social class, history of miscarriage and the time lag between questionnaires). Results. In general there was good agreement in the reported exposures to 48 substances and products. The results showed a small and consistent pattern of reporting exposures less frequently in the second questionnaire, i.e. after miscarriage. This was not explained by the analysis of possible confounding variables. Given the literature, the authors had expected to find a shift in the opposite direction. Conclusion. The study reinforces the need to be cautious when using the results from single surveys of retrospective self-reported exposure.