Background: Previous studies of the reproducibility of self-reported age at menarche have been limited because of small study samples, short follow-up and the limited age span of the women included.
Methods: The present study assessed the reproducibility of age at menarche in 6731 women with a wide variation of age when giving the information about age at menarche. The women reported age at menarche in a self-administered questionnaire, both in 1986-1987 and 1994-1995. They were all residents of Tromsø, Norway, and aged 25-73 in 1994-1995. In order to investigate the agreement between self-reported age at menarche at the two points in time, Pearson's correlation coefficient was applied to assess the linear correlation between the reported menarcheal age at the two occasions. Analyses were stratified for age. A Bland-Altman plot was produced and limits of agreement computed.
Results: We found a high correlation and a strong agreement between self-reported age at menarche in 1986-1987 and 1994-1995. The overall Pearson's correlation coefficient was 0.84 and was not attenuated by increasing age of the women. The Bland-Altman plot showed a strong agreement in self-reported age at menarche. The mean difference between self-reported age at menarche was 0.01 years with limits of agreement -1.52 to 1.54.
Conclusion: We found high reproducibility of self-reported age at menarche. The mean menarcheal age in the two surveys was identical (13.2 years) with 95% of the women reporting the same age at menarche or with a difference of 1 year. Only 0.7% of the women reported age at menarche with a difference of more than 2 years in 1986-1987 and 1994-1995.
Keywords: Correlation; Menarche; Menstruation; Reliability; Reproducibility.