Objectives: To assess the agreement between proxy informants' reports of history of surgery and childbirth and older index subjects' own recall.
Design: Interrater reliability study.
Setting: An outpatient family medicine clinic and a provincial electoral district in Montreal, Canada.
Participants: Eighty-two subjects aged 65 years and older without cognitive impairment, identified from clinic and community settings, and each index subject's proxy respondent.
Measurements: Identical questionnaires were administered to index subjects and proxies.
Results: Proxies failed to report 39% of non-childbirth surgeries reported by index subjects, but failed to report only 10% of childbirths. Female proxies were significantly less likely than male proxies to underreport non-childbirth surgeries after controlling for age of index subject and interval since surgery. Longer interval since surgery was significantly associated with greater underreporting, whereas age of the index subject and relationship between proxy and index subject were not. Agreement between proxies and index subjects on date of surgery was much higher for childbirths than for non-childbirth surgeries.
Conclusions: Our findings suggest that proxy respondents can provide reliable information on older women's history of childbirth but that use of proxy respondents for history of non-childbirth surgeries may result in substantial underreporting.