Validity of self-reported osteoporosis in mid-age and older women

Abstract

The validity of self-reported osteoporosis is often questioned, but validation studies are lacking. We validated self-reported prevalence and incidence of osteoporosis against self-reported and administrative data on medications. The concurrent validity was moderate to good for self-reported prevalent osteoporosis, but only poor to moderate for self-reported incident osteoporosis in mid-age and older women, respectively. Construct validity was acceptable for self-reported prevalent but not for incident osteoporosis.

Introduction: The validity of self-reported osteoporosis is often questioned, but validation studies are lacking. The aim was to examine the validity of self-reported prevalence and incidence of osteoporosis against self-reported and administrative data on medications.

Methods: Data were from mid-age (56-61 years in 2007) and older (79-84 years in 2005) participants in the Australian Longitudinal Study on Women's Health. Self-reported diagnosis was compared with medication information from (1) self-report (n(mid) = 10,509 and n(old) = 7,072), and (2) pharmaceutical prescription reimbursement claims (n(mid) = 6,632 and n(old) = 4,668). Concurrent validity of self-report was examined by calculating agreement, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). Construct validity was tested by examining associations of self-reported diagnosis with osteoporosis-related characteristics (fracture, weight, bodily pain, back pain, and physical functioning).

Results: Agreement, sensitivity and PPV of self-reported prevalent diagnosis were higher when compared with medication claims (mid-age women: kappa = 0.51, 95% confidence interval [CI] = 0.46-0.56; older women: kappa = 0.65, 95% CI = 0.63-0.68) than with self-reported medication (mid-age women: kappa = 0.41, 95% CI = 0.37-0.45; older women: kappa = 0.57, 95% CI = 0.55-0.59). Sensitivity, PPV and agreement were lower for self-reported incident diagnosis (mid-age women: kappa = 0.39, 95% CI = 0.32-0.47; older women: kappa = 0.55, 95% CI = 0.51-0.61). Statistically significant associations between self-reported diagnosis and at least four of five characteristics were found for prevalent diagnosis in both age groups and for incident diagnosis in older women.

Conclusions: The concurrent validity was moderate to good for self-reported prevalent osteoporosis, but only poor to moderate for self-reported incident osteoporosis in mid-age and older women, respectively. Construct validity was acceptable for self-reported prevalent but not for incident osteoporosis.