A clinical tool to identify older women with back pain at high risk of osteoporotic vertebral fractures (Vfrac): a population-based cohort study with exploratory economic evaluation

Abstract

Background: Osteoporotic vertebral fractures (OVFs) identify people at high risk of future fractures, but despite this, less than a third come to clinical attention. The objective of this study was to develop a clinical tool to aid healthcare professionals decide which older women with back pain should have a spinal radiograph. Methods: A population-based cohort of 1635 women aged 65+ years with self-reported back pain in the previous four months were recruited from primary care. Exposure data were collected through self-completion questionnaires and physical examination including descriptions of back pain and traditional risk factors for osteoporosis. Outcome was the presence/absence of OVFs on spinal radiographs. Logistic regression models identified independent predictors of OVFs, with the Area Under the (Receiver Operating) Curve (AUC) calculated for the final model, and a cut-point identified. Results: Mean age was 73.9 years and 209 (12.8%) had OVFs. The final Vfrac model comprised 15 predictors of OVF, with an AUC of 0.802 (95%CI 0.764-0.840). Sensitivity was 72.4% and specificity 72.9%. Vfrac identified 93% of those with >1 OVF and two-thirds of those with one OVF. Performance was enhanced by inclusion of self-reported back pain descriptors, removal of which reduced AUC to 0.742 (95%CI 0.696-0.788) and sensitivity to 66.5%. Health economic modelling to support a future trial was favourable. Conclusions: The Vfrac clinical tool appears valid and is improved by the addition of self-reported back pain symptoms. The tool now requires testing to establish real-world clinical and cost-effectiveness.