Vfrac – a simple clinical tool that identifies older women with back pain at high risk of osteoporotic vertebral fractures

Conference:

SAPC ASM 2021 - virtual

Talk Code:

4E.2

Presenter:

Tarnjit K Khera

Co-authors:

Linda P Hunt, Sarah Davis, Rachael Gooberman-Hill, Zoe Paskins, Tim J Peters, Jon H Tobias, Emma M Clark

Author institutions:

University of Bristol, Keele University, University of Sheffield

Problem

Osteoporosis and associated fragility fractures are one of the most common musculoskeletal conditions in older people. It is estimated that approximately three million people in the UK have osteoporosis. Osteoporotic vertebral fractures (OVFs) identify people at one of the highest risks of future fractures, but despite this, less than a third of patients with fractures come to clinical attention. Improving understanding about which clinical signs should be used to trigger referral for diagnostic spinal radiograph has potential to increase identification of OVF so that treatments to prevent further fracture can be instigated. Building on evidence that symptoms differ between adults with back pain and OVFs and those with back pain but no OVFs, the study aimed to develop a simple clinical tool to help clinicians to decide which older women with back pain should have a spinal radiograph.

Approach

1635 women aged 65+ with backpain in the previous four months were recruited from primary care in two regions of the United Kingdom (NRES 18/WS/0061; ISRCTN16550671). Data were collected through self-completion questionnaires, simple physical examination, spinal radiographs and GP records. Exposure data included descriptions of back pain, traditional risk factors for osteoporosis, basic anthropometry and reported height loss. The outcome was the presence/absence of OVF identified using the Algorithm-Based Qualitative method. Severity of the fracture was categorised using the Genant semi-quantitative (SQ) method. A series of logistic regression models identified independent predictors of OVFs. Model validation included calibration-in-the-large (CITL), calibration slope and heuristic shrinkage (Van Houwelingen). 500 bootstrapped samples were obtained and used to estimate shrinkage and adjust the calibration slope and AUC optimism. The proposed final cut-off to identify which older women with back pain should have a spinal radiograph because of high risk of fracture was based on a maximised sum of sensitivity and specificity.

Findings

Mean age was 73.9 years (range 65.4 to 96.8), and 209 (12.8%) had OVF. The AUC of the final model (including 15 independent predictors) was 0.802 (0.764-0.840), sensitivity was 72.4% and specificity was 72.9%. Of those recommended for spinal radiographs based on Vfrac, 27.1% had an OVF. Vfrac identified 62.0% of those with one OVF and 92.7% of those with >1 OVF. It identified 92.3% of those with severe OVF and 63.1% of those with mild/moderate OVF. The Vfrac tool provides a targeted method to identify people with OVF and hence at high risk of future fragility fracture. Next steps include testing of the tool’s real-world clinical and cost-effectiveness.

Consequences

Vfrac will be a web-based online tool that can be supported through NHS IT systems.

Submitted by:

Tarnjit Khera

Funding acknowledgement:

The Vfrac Study is funded by a Clinical Studies grant from Versus Arthritis (grant reference 21507). Vfrac also acknowledges support of the NIHR Portfolio through the Clinical Research Network.