Reproducibility of a reproductive questionnaire: Results from a case- control study

Background - Questionnaires on reproductive/hormonal history are widely used in epidemiological studies, but their reproducibility has not been tested. Methods - Three-hundred and ninety-two newly diagnosed breast cancer cases and 400 controls were interviewed about their reproductive and hormonal history. A random sample of 200 subjects was selected from this cohort and re-interviewed over the telephone one year later. In addition, a study was conducted to test the internal validity of the information obtained on mother's age at first full-term pregnancy. Results - For all categorical variables, we observed a high level of agreement between the first and the second interview, both among cases and controls. The variables that showed a lower reproducibility were: age at first use of estrogen replacement therapy (intraclass correlation coefficient 0.74), length of the menstrual period (0.61) and of the menstrual cycle (0.72). In the internal validity study, the correlation of mother's age at first full-term pregnancy was 0.88 for the total sample; 0.81 for the cases and 0.87 for the controls. Conclusion - Our results show that most of the variables have a very high degree of reproducibility, and that the reproducibility is generally higher among cases than among controls.