Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Predicting the need for knee radiography in the emergency department: Ottawa or Pittsburgh rule?

Three Part Question

In [adult patients presenting to the emergency department with traumatic knee pain] does the [Ottawa knee rule or the Pittsburgh knee rule] have [greater sensitivity and specificity for knee fractures]?

Clinical Scenario

A 30-year-old man presents to the emergency department (ED) after twisting his knee. You suspect a soft tissue injury and are aware that the Ottawa knee rule could be used to help determine whether radiography is necessary. A colleague suggests that you should use the Pittsburgh rule instead. You wonder which rule has greater sensitivity (thus missing fewer fractures) and greater specificity (thus reducing the need for unnecessary radiography).

Search Strategy

We searched Medline 1948 – Week 3 October 2011 and Embase 1980 – 2011 Week 44 using the Ovid interface. We also searched the Cochrane database of systematic reviews on 7 November 2011.
Medline and Embase:
(exp Knee/ or exp Knee Joint/ or knee$.af.) and ((Ottawa and Pittsburgh) or decision rule$).af. limit to humans and English language

Cochrane database: ‘Ottawa knee rule’ (Title, Abstract or Keywords) OR ‘Pittsburgh knee rule’(Title, Abstract or Keywords)

Search Outcome

Search outcomes:

We identified 31 papers in Medline, 35 in Embase and 5 in the Cochrane database. We identified two papers that addressed the three-part question.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Seaberg et al,
United States
A convenience sample of 934 patients with acute knee injuries across 3 hospitals. Data were collected that enabled retrospective assessment of both the Pittsburgh and Ottawa knee rules. Patients underwent radiography at the discretion of the treating physician.Prospective validation and comparison study with cohort designComparison of sensitivityNo statistically significant difference. Pittsburgh: 99%, Ottawa 97%. Difference 2% (95% CI -2 - 7%)Clinicians only partially blinded. Children included in Pittsburgh rule but not Ottawa. Convenience sampling used could lead to selection bias. (Pittsburgh rule definition of walking used for all (stricter than Ottawa); would only have increased Ottawa specificity to 37%) Verification bias: radiography was requested at the discretion of the treating physician. Only around 750 patients (of 934) underwent radiography.
Comparison of specificityPittsburgh rule significantly more specific. Pittsburgh: 60%, Ottawa 27%. Difference 33% (95% CI 28-38%)
Comparison of misclassification ratesPittsburgh rule had lower misclassification rate. Pittsburgh: 36%; Ottawa 65%. Difference 29% (95% CI 25 - 35%). 1 (of 91) missed fractures with Pittsburgh rule; 3 (of 87) missed fractures with Ottawa rule
Richman et al,
351 patients assessed with standardised history and examination criteria for both rules in 2 hospitals. Radiography was undertaken at the discretion of the treating physician with a structured 3-week follow up for those who did not undergo radiography. 281 (87%) patients underwent radiographyProspective diagnostic cohort studyComparison of sensitivityOttawa 84.6% (95% CI 65.1 - 95.1%); Pittsburgh 84.6% (95% CI 65.1 - 95.6%). No statistical comparison but each rule missed 4 (of 26) fractures (Ottawa missed 2 tibial plateau fractures and 2 patella fractures; Pittsburgh missed 4 patella fractures)Not designed as a comparison study. Clinicians only partially blinded. Some patients lost to follow up. No power calculation performed. Clinician’s compliance with standardised tool not assessed. Not all patients underwent radiography (verification bias)
Comparison of negative predictive valueOttawa 97.6% (93.8 - 99.3%); Pittsburgh 97.5% (95% CI 93.8 - 99.3%)
Comparison of speicficityOttawa 49.8% (95% CI 44.3 - 55.4%); Pittsburgh 48.9% (95% CI 43.4 - 54.4%)
Comparison of positive predictive valueOttawa 11.9% (95% CI 7.7 - 17.5%); Pittsburgh 11.7% (7.5 - 17.2%)
Projected reduction in radiographyWith use of the decision rules, radiography could have been reduced by 98 (34%) with the Ottawa rule and 95 (33%) with the Pittsburgh rule
Physician estimate of probability of fractureUsing unstructured assessment, if physicians estimated the probability of fracture as 0%, 3/200 (1.5%) of patients had a fracture. As there were 26 fractures, the sensitivity was 88.5%.


The Ottawa rule recommends radiography if any of 5 features are present: (1) Age ≥55 years; (2) Isolated tenderness of the patella; (3) Tenderness at the fibular head; (4) Inability to flex the knee to 90 degrees; (5) Inability to bear weight both immediately and in the ED (limping is acceptable). The Pittsburgh rule recommends radiography if the mechanism of injury is blunt trauma or a fall and either (1) age is <12 years or >50 years, or (2) the patient cannot complete four weight-bearing steps in the ED.

We identified two cohort studies (see table 1) that have directly compared the diagnostic accuracy of these two clinical decision rules. The studies demonstrate that the rules have similar sensitivity. Both rules would miss some fractures when used alone. In the larger study, the Pittsburgh rule had a sensitivity of 99% with significantly superior specificity compared to the Ottawa rule (60% vs 27%). Thus, while the Ottawa knee rule has been more extensively investigated in both cohort studies and randomised controlled trials of clinical implementation (Bachmann et al 2004, Stiell et al 1997), the evidence presented here suggests that the Pittsburgh rule may be a viable alternative and could lead to less use of radiography without lowering sensitivity. Randomised controlled trials are now necessary to definitively answer this question.

Clinical Bottom Line

The Ottawa and Pittsburgh rules appear to have similar sensitivity, although the Pittsburgh rule may have higher specificity and could therefore lead to less unnecessary radiography. Neither rule has perfect sensitivity so clinical judgement should still be exercised. Randomised controlled trials are necessary to definitively answer this question.


  1. Seaberg DC, Yealy DM, Lukens T et al. Multicenter comparison of two clinical decision rules for the use of radiography in acute, high-risk knee injuries. Annals of Emergency Medicine 1998; 32(1): 8 - 13.
  2. Richman PB, McCuskey CF, Nahed A et al. Performance of two clinical decision rules for knee radiography. Journal of Emergency Medicine 1997; 15(4): 459 - 463.
  3. Bachmann L, Haberzeth S, Steurer J, et al. The accuracy of the Ottawa knee rule out fractures. Ann Intern Med 2004;140:121e4.
  4. Stiell IG, Wells GA, Hoag RH, et al. Implementation of the Ottawa knee rule for the use of radiography in acute knee injuries. J Am Med Assoc 1997;278:2075e9.