Long-term recall of time-to-pregnancy

Abstract

To validate two versions of a short self-completion questionnaire on time-to-pregnancy. Information from the questionnaire was compared with concurrently collected data from the same individuals. Questionnaires were sent to 1,647 women who continue to be followed up by the Oxford Family Planning Association Contraceptive Study. Replies were received from 1,498, a response rate of 91.0%. Successful matching was achieved with 1,392 pregnancies that met the study criteria and that had values of time-to-pregnancy in both data sources. Median recall time was 14 years (interquartile range, 11 to 16 years). At the group level, the frequency distributions of time-to-pregnancy from the two sources are presented as cumulative percentages. At the individual level, the distribution of discrepancies between the sources is tabulated separately for each value of time-to-pregnancy, and accuracy of detection of clinical subfertility is presented (sensitivity and specificity). At the group level, remarkably good agreement was found between the two sources of information. Digit preference was present to a limited degree. There were no important differences between the two questionnaire versions. At the individual level, some misclassification was evident. For the detection of clinical infertility, sensitivity was 79.9% and specificity was 94.9%. Short, self-completion questionnaires are remarkably accurate for assessing time-to-pregnancy at a group level. Individual-level misclassification is frequent, but detection of clinical subfertility is fairly accurate.