[Reliability of a questionnaire for women's reproductive history]

This study was performed to evaluate the reproducibility of a questionnaire concerned with reproductive history and to ascertain which characteristics of the subjects (age, the visit-revisit intervals, education and chronic disease) are associated with good reliability in the Korean Multi-Center Cancer Cohort (KMCC) study.

A total of 19,688 participants were enrolled between 1993 and 2004. Among them, we selected 386 participants who were aged 40 or more and who re-visited within 8 years after the first visit. Reliability was measured by the percent agreement according to error range for the continuous variables and the percent agreement and kappa statistics for the categorical variables.

The pregnancy histories were reliable (kappa=0.67) and the reasons for being menopausal among the postmenopausal women were also reliable (kappa=0.92). The percent agreement of the breast-feeding history was high (96.1%), although the kappa statistic was low. For the continuous variables, when the error range of one variable was considered to be reliable, the percent agreement of the age at menarche and the age at the first full term pregnancy was good (69.4% and 83.6%), whereas that of the age at menopause was low (51.5%). The factors associated with high reliability were a younger age, the presence of chronic disease and a short visit-revisit time interval.

The agreements for parity, the reasons for menopause, and the breastfeeding history in the reproductive history questionnaire used in the KMCC were relatively good. The questionnaire for the menarche age and the menopausal age might have lower reliability due to the difference between Korean age and American age. To obtain reliable information, more attention should be given to the items in questionnaire-based surveys, and especially for surveying old-aged women.