Article Text
Abstract
STUDY OBJECTIVE--To establish the degree of validity of data on time to pregnancy, derived retrospectively using a short questionnaire. DESIGN--Information from the questionnaire was compared with data that had been collected concurrently from the same individuals. SETTING AND PARTICIPANTS--Questionnaires were mailed to 1647 women who continue to be followed up by the Oxford Family Planning Association contraceptive study, and a further 424 were approached for personal interview. Response rates were 91% and 79% respectively. MAIN RESULTS--Matching was successful in 91% of pregnancies. Median recall time was 14 years (interquartile range, 11-16 years). At the group level, remarkably good agreement was found between the two sources of information, presented as cumulative percentage distributions of live births. The findings were at least as good with longer recall (> 14 years) as with shorter recall. Digit performance was present to a limited degree. At the individual level, some misclassification was evident, which has implications for statistical power. For detection of clinical infertility (no conception within 12 months), the sensitivity was in the range 67%-91%, and the specificity was 92%-96%. Variations with format, duration of recall, age at delivery, year of birth, parity, social class, smoking habit, last contraceptive method, and outcome (live birth or not) were generally small, and were not statistically significant. CONCLUSIONS--Time to pregnancy is a sensitive way of assessing reproductive function in either sex. Valid data at a group level can be derived retrospectively, with a long duration of recall, using a short questionnaire.