Abstract
BACKGROUND: Menarche timing may affect female health. While previous studies evaluated self-reported age at menarche reproducibility, they did not assess types of respondents. This study compared the reproducibility of age at menarche among self-responders and proxy respondents and assessed proxy-respondent reproducibility by relationship and survey age. METHODS: Data on age at menarche reported in both the 1969 and 1978 mail questionnaires among 9,043 females from the Life Span Study cohort of atomic bomb survivors were analyzed. The reproducibility of menarcheal age was assessed by the type of respondents, by proxy's relationship to participant, and by age at the 1969 survey using Bland-Altman's method and the intraclass correlation coefficient (ICC). RESULTS: Reproducibility was moderate (95% limits of agreement, -2.3 to 2.4 years; ICC 0.72; 95% confidence interval, 0.71-0.73). Both self-respondents (N = 6,664) and the total study population (N = 9,043) maintained moderate reproducibility even at older ages. Groups with proxy reports showed lower reproducibility than self-respondents, with spouse proxy reports showing highest reproducibility and parent proxy reports showing lowest reproducibility among proxy reports, although the comparisons are based on different survey ages in 1969. CONCLUSION: This is the first study to evaluate menarcheal-age reproducibility between self- and proxy-reports using appropriate measures. Mail questionnaires at 9-year interval showed moderate reproducibility across all ages, including elderly self-respondents. Reproducibility varied by the respondent-target relationship, with spouse proxies showing highest and parent proxies showing lowest reproducibility among proxy reports. Additional data are required to establish appropriate methods for handling specific proxy responses.