Five-year reliability of key demographic and life history data collected from self-respondents and next-of-kin.

The ability of study participants or their next-of-kin to provide basic demographic and life history data is of critical importance in epidemiologic studies and government surveys. Most interview studies to date have focused on the reliability of these variables either as self-reported over short periods of time (<2 years) or as reported by the next-of-kin concurrently with the study participant. In a unique reinterview study, the authors examine the concordance of responses 5 years after the initial interview among 196 surviving study participants and 107 next-of-kin respondents of participants who died after the first interview. The reliability of demographic, anthropometric, reproductive, and residential history questions was high overall, with most κ and Spearman rank-correlation coefficients being above 0.80 for both self-respondents and next-of-kin at reinterview in 1985. In particular, almost perfect agreement was observed for year of birth, religion raised, number of children, and age at first birth, whereas agreement was lowest for childhood residential history. Contrary to expectation, the next-of-kin responses were generally as reliable as those of the original participants themselves 5 years after the initial study, providing further support for the usefulness of surrogate respondents in collecting demographic and life history information in epidemiologic investigations when no other source is available.