The purpose of this study is to examine the validity of, and factors associated with, the accuracy of self-report (participant-report and proxy-report) for fractures.
Study participants were from the Women's Health Initiative Clinical Trial and Observational Study cohorts. All women were postmenopausal; populations included American Indian, Asian/Pacific Islander, black, Hispanic, and non-Hispanic white. The average length of follow-up was 4.3 years. Self-reported fractures were adjudicated by reviewing medical records. The first adjudicated self-report of fractures for each participant was included in the analysis (n = 6,652).
We found substantial variations in validity of self-report by the fracture site. Agreements between self-reports for single-site fractures and medical records were high for hip (78%) and forearm/wrist (81%) but relatively lower for clinical spine fractures (51%). The average confirmation rate for all single-site fractures was 71%. Misidentification of fracture sites by participants or proxy-reporters seemed to be a cause of unconfirmed self-reports. Higher confirmation rates were observed in participant-reports than in proxy-reports. Results of the multivariate analysis indicated that multiple factors, such as ethnicity, a history of osteoporosis or fractures, body mass index, years since menopause, smoking status, and number of falls in the past year were significantly (P < 0.05) related to the validity of self-report.
The validity of self-reports for fracture varies by fracture sites and many other factors. The assessed validity in this study is likely conservative because some of the unconfirmed self-reports may be due to poor medical record systems. The validity of self-reports for hip and forearm/wrist fractures is high in this study, supporting their use in epidemiological studies among postmenopausal women.
Self-reports for fracture are often used in epidemiologic studies. However, the accuracy of these reports varies by skeletal sites and other factors, suggesting careful evaluation and selective use of self-reported fracture outcomes in future research.
From the 1University of Arizona, Tucson, AZ; 2Fred Hutchinson Cancer Research Center, Seattle, WA; 3University of Pittsburgh, Pittsburgh, PA; 4University of Alabama at Birmingham, Birmingham, AL; 5University of Florida, Gainesville, FL; 6UCLA Medical Center, Los Angeles, CA; and 7Ohio State University, Columbus, OH.
Received June 2, 2003; revised and accepted August 13, 2003.
The research upon which this article is based was performed pursuant to Contract No: N-01-WH-3-2115 with the National Institutes of Health, Department of Health and Human Services. Zhao Chen, MPH, PhD, is supported by a career development award (1 K01 AR02060) from the National Institutes of Health, Department of Health and Human Services.
Address correspondence to: Zhao Chen, MPH, PhD, Epidemiology and Biostatistics, College of Public Health, University of Arizona, 2501 E. Lee St., Tucson, AZ 85716. E-mail: email@example.com.