Red flags to screen for vertebral fracture in people presenting with low back pain

Key points

1. 01Only a few of the 130 red flags tested across 14 studies were informative; most had positive and negative likelihood ratios close to 1
2. 02In primary care, trauma, older age (above 70), and corticosteroid use showed informative positive likelihood ratios for unspecified or osteoporotic vertebral fracture
3. 03Combining red flags improved diagnostic value, for example older age plus female gender raised the positive likelihood ratio from 9.39 to 16.17 in primary care
4. 04Contusion or abrasion in a tertiary emergency setting had the strongest single-test positive LR of 31.09 (95% CI 18.25 to 52.96)
5. 05Meta-analysis was not possible due to study heterogeneity; all findings come from single studies and should be interpreted with caution

How it was conducted

Design

Cochrane systematic review of diagnostic test accuracy studies; update of Williams 2013

Included studies

14 studies (8 from previous review, 6 new); 4 prospective cohorts, 4 retrospective chart reviews; 6 primary care, 5 secondary care, 3 tertiary care

Participants

Total participants across studies ranged from 101 to 9940 per study; combined across all settings approximately 19,759

Index tests

130 red flag tests covering history and physical examination; 18 presented as combinations

Reference standards

X-ray, MRI, CT, SPECT, or clinical follow-up imaging depending on fracture type and study

Quality assessment

QUADAS-2 tool; no study had low risk of bias across all domains; risk of bias often rated unclear

What they found

• Trauma in primary care (unspecified fracture): sensitivity range 0.21-0.65, specificity range 0.90-0.98, +LR range 1.93-12.85
• Trauma for osteoporotic fracture in primary care (Enthoven 2016, adults over 55): sensitivity 0.21 (95% CI 0.09 to 0.39), specificity 0.97 (95% CI 0.95 to 0.98), +LR 6.42 (95% CI 2.94 to 14.02)
• Older age above 70 years in primary care (unspecified fracture, Henschke 2009, n=1172): sensitivity 0.50 (95% CI 0.16 to 0.84), specificity 0.96 (95% CI 0.94 to 0.97), +LR 11.19 (95% CI 5.33 to 23.51)
• Corticosteroid use in primary care (unspecified fracture): sensitivity range 0.00-0.25, specificity range 0.99-0.99, +LR range 3.97 (95% CI 0.20 to 79.15) to 48.50 (95% CI 11.46 to 204.98)
• Combination of 2 of 4 red flags (Henschke) in primary care (unspecified fracture, n=1172): sensitivity 0.63 (95% CI 0.24 to 0.91), specificity 0.96 (95% CI 0.95 to 0.97), +LR 15.48 (95% CI 8.45 to 28.36)
• Female and age above 74 years in primary care (unspecified fracture): +LR range 4.36-16.17
• Trauma in secondary care (unspecified fracture, Premkumar 2018, n=9940): sensitivity 0.25 (95% CI 0.21 to 0.29), specificity 0.89 (95% CI 0.88 to 0.89), +LR 2.18 (95% CI 1.86 to 2.54)
• Older age above 74 years in secondary care (osteoporotic fracture, Kilic 2021): sensitivity 0.53 (95% CI 0.41 to 0.64), specificity 0.79 (95% CI 0.67 to 0.88), +LR 2.51 (95% CI 1.48 to 4.27)
• Older age above 70 plus trauma in secondary care (unspecified fracture, Premkumar 2018): +LR 4.35 (95% CI 2.92 to 6.48)
• 4 of 5 features positive (Roman combination) in secondary care (osteoporotic fracture, n=1448): sensitivity 0.37 (95% CI 0.22 to 0.54), specificity 0.96 (95% CI 0.95 to 0.97), +LR 9.62 (95% CI 5.88 to 15.73)
• Contusion or abrasion in tertiary care (compression fracture, Patrick 1983, n=552): sensitivity 0.85 (95% CI 0.70 to 0.94), specificity 0.97 (95% CI 0.95 to 0.98), +LR 31.09 (95% CI 18.25 to 52.96), -LR 0.15 (95% CI 0.07 to 0.32)

Limitations

• Meta-analysis was not possible due to large heterogeneity in index tests, healthcare settings, and fracture type definitions across studies
• Most informative findings come from single studies only, making replication and confidence in estimates impossible
• Risk of bias was unclear across most domains in the majority of studies, and no study had low risk of bias in all four QUADAS-2 domains
• Definitions of red flags, such as corticosteroid dose and duration, varied widely across studies, likely explaining inconsistent results

Why it matters

For patients

If you have low back pain and are older, have had a fall or trauma, or take corticosteroid medicines, your clinician may use those findings together to decide whether imaging is needed, but no single symptom reliably rules a fracture in or out.

For clinicians

Only a handful of red flags, particularly trauma, older age (above 70), corticosteroid use, and contusion or abrasion, show informative positive likelihood ratios; combining these flags improves yield, and guideline-recommended flags such as female sex and osteoporosis diagnosis remain poorly supported by current evidence.

For readers

This Cochrane review highlights a significant evidence gap: with only 14 mostly heterogeneous studies and no meta-analysis possible, current recommendations for vertebral fracture screening rest on very limited and inconsistent primary data.