BACKGROUND: The self-report of medical history and medication use is a common feature of epidemiological research. METHODS: In a unique re-interview study, we evaluated the concordance of medical conditions and past medication use reported in two similar interviews 5 years apart. RESULTS: In 196 re-interviews with the subjects themselves, and in 107 with next-of-kin of subjects who died after the first interview, agreement was good or excellent (kappa > or =0.40) for 90% (9/10) of the conditions asked about in the personal medical history for both next-of-kin and self-respondents. Agreement was excellent (kappa >0.75) for two conditions, high blood pressure and hysterectomy, among self-respondents. Self- and surrogate respondents also showed similar reproducibility for prescription medications, but next-of-kin respondents tended to have poor agreement (kappa <0.40) for over-the-counter (OTC) medications such as antacids, antihistamines, and analgesics. Next-of-kin also less reliably reported a family history of cancer. When analyses were stratified by type of surrogate respondent, concordance between the two interviews was generally higher for spouses than for other surrogate respondents. CONCLUSIONS: This research demonstrates that personal medical history and prescription medication use may be as reliably reported by next-of-kin as self-respondents, but suggests that additional information may be needed to validate measures of OTC medication use and family history of cancer for next-of-kin respondents, possibly through the review of hospital records.
BACKGROUND: The self-report of medical history and medication use is a common feature of epidemiological research. METHODS: In a unique re-interview study, we evaluated the concordance of medical conditions and past medication use reported in two similar interviews 5 years apart. RESULTS: In 196 re-interviews with the subjects themselves, and in 107 with next-of-kin of subjects who died after the first interview, agreement was good or excellent (kappa > or =0.40) for 90% (9/10) of the conditions asked about in the personal medical history for both next-of-kin and self-respondents. Agreement was excellent (kappa >0.75) for two conditions, high blood pressure and hysterectomy, among self-respondents. Self- and surrogate respondents also showed similar reproducibility for prescription medications, but next-of-kin respondents tended to have poor agreement (kappa <0.40) for over-the-counter (OTC) medications such as antacids, antihistamines, and analgesics. Next-of-kin also less reliably reported a family history of cancer. When analyses were stratified by type of surrogate respondent, concordance between the two interviews was generally higher for spouses than for other surrogate respondents. CONCLUSIONS: This research demonstrates that personal medical history and prescription medication use may be as reliably reported by next-of-kin as self-respondents, but suggests that additional information may be needed to validate measures of OTC medication use and family history of cancer for next-of-kin respondents, possibly through the review of hospital records.