Agreement of self-reported medical history: Comparison of an in-person interview with a self-administered questionnaire

Purpose: To compare history of 22 different diseases reported during an in-person interview with that reported on a mailed self-administered questionnaire. Methods: 7841 participants of the European Prospective Investigation into Cancer (EPIC)-Potsdam study. The interview at baseline and the questionnaire at follow-up approximately 2 years later included identical questions about whether the participant had ever had a physician diagnosis of each disease. Incident diagnoses occurring in the interval between the interview and questionnaire were excluded from the analysis. Results: Agreement between self-report from the interview and from the questionnaire was highest (κ = 0.83-0.88) for myocardial infarction, cancer and diabetes mellitus; it was lower (κ = 0.68-0.77) for gout, hypertension, hay fever, asthma, osteoporosis, ulcer of the duodenum, thyroid disease, stroke, and kidney stones, and was lowest (κ = 0.39-0.59) for chronic gastritis, ulcer of the stomach, cerebral ischemia, benign tumor, inflammatory bowel disease, angina pectoris, hyperlipidemia, rheumatism, colon polyps and skin disease. The poor agreement for less severe or more transient diseases was primarily a result of disease frequently being reported at the interview but not on the questionnaire. Conclusion: Self-administered questionnaires do not generate same information particularly for less severe or transient diseases as personal interviews. For these diseases, self-administered questionnaires are not recommended. Pilot studies that test validity will be necessary.