Diagnostic accuracy of the Ottawa ankle rule to exclude fractures in acute ankle injury

Key points

1. 01Pooled sensitivity was 0.91 (95% CI 0.89-0.92), meaning the tool correctly identifies most true fractures
2. 02Pooled specificity was only 0.25 (95% CI 0.24-0.26), so many people without fractures still test positive
3. 03A negative result reduces the odds of fracture by a factor of 1.47 (negative LR 0.15)
4. 04A positive result increases the odds of fracture only 1.5 times (positive LR 1.47), limiting its confirmatory value
5. 05The rules perform best as a rule-out tool and should always be used alongside clinical reasoning

How it was conducted

Design

Systematic review and meta-analysis (PRISMA-P compliant)

Search dates

Database inception (1992) to December 2020, across SPORTDISCUS, COCHRANE, MEDLINE, EMBASE, EMCARE, SCOPUS

Participants

8,560 adult patients (aged 18 and over) from 15 included studies across 13 countries

Included studies

15 studies: 9 prospective, 3 retrospective, 1 with both designs, 1 RCT, 1 non-randomised controlled trial

Risk of bias tool

QUADAS-2 applied independently by two reviewers

Statistical analysis

Bivariate random-effects meta-analysis; pooled sensitivity, specificity, likelihood ratios, diagnostic odds ratio, and SROC curve

What they found

• Pooled sensitivity: 0.91 (95% CI 0.89 to 0.92)
• Pooled specificity: 0.25 (95% CI 0.24 to 0.26)
• Positive likelihood ratio: 1.47 (95% CI 1.11 to 1.93)
• Negative likelihood ratio: 0.15 (95% CI 0.72 to 0.29)
• Diagnostic odds ratio: 10.95 (95% CI 5.14 to 23.35)
• Individual study sensitivity ranged from 59% to 100%; specificity ranged from 2% to 69%
• High between-study heterogeneity observed (I2 for sensitivity and specificity both p < 0.01)
• Sensitivity in low-risk-of-bias studies only: 96.4% (83.7% to 99.3%); specificity: 31.9% (8.3% to 70.7%)
• Spearman correlation coefficient was 0.31, confirming no threshold effect
• Area under SROC curve calculated but exact value partially obscured in source text

Limitations

• High between-study heterogeneity (I2 ranging from 81.7% to 99.7%) limits confidence in pooled estimates
• Only adult populations included (age 18+), so findings do not apply to children; some studies included patients over 80, which may have introduced variability
• Non-English articles and conference abstracts were excluded, introducing possible publication bias
• Eight studies did not report blinding of the assessor applying the reference standard, raising risk of bias

Why it matters

For patients

If a clinician applies the Ottawa Ankle Rules and the result is negative, it is very unlikely you have a fracture and you may safely avoid an X-ray.

For clinicians

The OAR is a reliable rule-out tool for ankle fractures in adults when negative, but positive findings must not be used alone to order imaging; clinical reasoning remains essential to avoid excessive X-ray referrals.

For readers

This meta-analysis confirms the OAR has high sensitivity but low specificity in adults, supporting its use to reduce unnecessary radiography while acknowledging its limitations as a standalone rule-in tool.