Trauma · Clinical decision rule
Ottawa Knee Rules
The Ottawa Knee Rules decide who needs a knee radiograph after an acute blunt injury — designed to catch every clinically significant fracture while sparing routine imaging of every sprain. Order a knee series if ANY ONE of five criteria is positive: age 55 years or older; isolated tenderness of the patella; tenderness at the head of the fibula; inability to flex the knee to 90°; or inability to bear weight for four steps both immediately after the injury and in the department. Sensitivity for fracture is ~98–100% but specificity is deliberately low (~19–49%) — the rule over-images a little so nothing significant is missed, and applying it cuts knee radiographs by ~20–30%. It is for the acute, isolated, blunt knee injury in an alert patient (valid in children, with the convention of using it from about age 5); it does NOT apply to pregnancy, obvious deformity, penetrating or multi-trauma, or re-assessment of an already-imaged knee — and a negative rule does not exclude internal derangement (ACL/meniscal/osteochondral).
5 criteriaAny ONE positive → radiograph
~98–100% sens.For significant knee fracture
20–30%Fewer knee radiographs
4 stepsWeight-bear, both times
Bottom lineImage if ANY of: age ≥55, isolated patellar tenderness, fibular head tenderness, can't flex to 90°, or can't weight-bear 4 steps (both immediately and in the ED). One positive criterion is enough. A negative rule safely avoids imaging in the cooperative patient — but document each criterion and the weight-bearing attempt, and remember it is a fracture rule, not a soft-tissue rule.
Traps that cost marks
- The weight-bearing criterion must be positive at BOTH time points — immediately after the injury AND in the department. A limp counts as weight-bearing; a patient who limps into the ED but could not walk at the scene is still positive. This is the most mis-applied criterion.
- "Isolated" patellar tenderness means the patella is the ONLY tender bony point. Joint-line tenderness does NOT satisfy the rule (it is a soft-tissue sign, not a criterion).
- One positive criterion is enough — do not wait for several before imaging.
- It is a fracture rule, not a soft-tissue rule. A negative rule does NOT exclude an ACL/meniscal/osteochondral injury — a tense haemarthrosis in a young athlete is an ACL/osteochondral injury until proven otherwise, regardless of normal films.
- Know when it does NOT apply — pregnancy, obvious deformity, penetrating injury, polytrauma/distracting injury, or re-assessment of an already-imaged knee.
The five Ottawa Knee Rule criteria
The rule is applied in an alert patient (and, with the published paediatric validation, in children — by convention from about age 5) with an acute blunt knee injury. A knee radiograph series is indicated if any one of the following is positive:
1
- Criterion
- Age 55 years or older
- How to test
- Ask the patient (or check the record)
- Positive when
- Age 55 or more at the time of injury
2
- Criterion
- Tenderness at the head of the fibula
- How to test
- Palpate the fibular head and proximal fibula
- Positive when
- Reproducible tenderness at the lateral prominence
3
- Criterion
- Isolated tenderness of the patella
- How to test
- Palpate the patella and the rest of the knee
- Positive when
- Tenderness confined to the patella with no other bony tenderness
4
- Criterion
- Inability to flex the knee to 90°
- How to test
- Ask the patient to bend the knee
- Positive when
- Cannot reach 90° of flexion (pain or mechanical block)
5
- Criterion
- Inability to bear weight for four steps
- How to test
- Ask about immediate weight-bearing AND observe in the ED
- Positive when
- Cannot take four weight-bearing steps (limping counts) both immediately and in the ED
Mnemonic
AF-IPPThe five criteria
A
Age 55+
55 years or older at the time of injury
F
Fibular head tenderness
Reproducible tenderness on palpating the fibular head
I
Isolated patellar tenderness
The patella is the only tender bony point
P
Passive flexion to 90
Cannot flex the knee to 90° (pain or block)
P
Partial weight-bearing ×4
Cannot take four steps, both immediately and in the ED
Hook:AF-IPP — Age 55+, Fibular head, Isolated patella, Passive-flexion failure, or Partial weight-bearing failure = image. One positive criterion is enough.
The points most often mis-applied
'Isolated' patellar tenderness means the patella is the ONLY tender bony point — actively palpate it so a vertical/osteochondral fracture is not missed. Weight-bearing is four steps (two onto each foot) at BOTH time points — immediately after the injury AND in the department; a limp counts as weight-bearing, and a patient who limps in but could not walk at home is still positive. One positive criterion is enough — do not wait for several.
How to apply the rule in practice
Applying the Ottawa Knee Rules
- 1Confirm the rule appliesAcute, isolated, blunt knee injury in an alert patient. NOT for pregnancy, obvious deformity, penetrating/multi-trauma, or re-assessment of an already-imaged knee.
- 2Check all five criteriaAge ≥55, fibular head tenderness, isolated patellar tenderness, flexion to 90°, and the four-step weight-bearing test (ask about the scene AND observe in the ED).
- 3Any positive → radiographStandard knee series (AP, lateral, skyline; add a tunnel/notch view if a patellar or osteochondral fracture is suspected). Record the positive criterion.
- 4All negative → clinical careDiagnose a soft-tissue injury; RICE, analgesia, early mobilisation, and clear return precautions. Document the negative criteria and the weight-bearing attempt.
Bottom lineThe rule is a floor, not a ceiling: a negative rule lets you safely avoid imaging, but if clinical suspicion is high (severe pain, tense effusion, mechanism), image anyway and document why.
Caution
A negative Ottawa Knee Rule means a clinically significant fracture is very unlikely, but it does NOT exclude internal derangement (ACL, meniscal, osteochondral) or a non-displaced patellar/tibial-plateau fracture. Give clear return precautions: increasing pain, inability to weight-bear beyond 48 hours, a new effusion, or a locked knee all warrant re-assessment and often MRI. Severe pain disproportionate to the mechanism in an older patient should trigger imaging even if the rule is negative.
The tense haemarthrosis and the radiographic clues a 'normal' film can hide
Because the rule is a fracture tool, the examiner pushes on what it misses — the acute tense haemarthrosis and the subtle plain-film signs of occult injury:
- Differential of an acute traumatic haemarthrosis — the commonest cause is an ACL rupture (around 70–75%); then a peripheral meniscal tear, a patellar dislocation (MPFL/medial retinacular tear, often self-reduced), an osteochondral fracture, and a PCL injury or intra-articular fracture. A tense haemarthrosis in a young athlete is one of these until proven otherwise, regardless of normal films.
- Lipohaemarthrosis (the fat-fluid level / "FBI" sign) — on a horizontal-beam (cross-table) lateral, a fat-fluid level from marrow fat escaping into the joint is pathognomonic of an intra-articular fracture even when no fracture line is visible — a reason to get the horizontal-beam lateral and look hard (often an occult tibial plateau fracture).
- The avulsion clues — a Segond fracture (a small avulsion off the lateral tibial rim, the anterolateral ligament/lateral capsule) is strongly associated with an ACL rupture; a reverse Segond (medial rim) points to a PCL/medial-meniscus injury; and an arcuate sign (avulsion of the fibular head/styloid) signals a posterolateral corner injury. Spotting any of these on a "negative-for-obvious-fracture" film changes the whole work-up toward MRI and ligament assessment.
Limitations, pitfalls & modern context
- Specificity is intentionally low (~19–49%). The rule over-images a little so no significant fracture is missed — use it to decide whether plain films are needed, not to "clear" a knee for sport.
- Invalid in specific populations: pregnancy, obvious deformity/clinical fracture, penetrating trauma, polytrauma or a distracting injury, isolated superficial skin injury, or re-assessment of an already-imaged knee.
- It is a fracture rule, not a soft-tissue rule — it will not detect an isolated ACL rupture, meniscal tear, or patellar dislocation; these need examination and frequently MRI.
- Paediatric use: validated in children aged 2–16 with 100% sensitivity, though most guidelines use about age 5 as the practical lower limit; apply with caution in pre-school children.
- Inter-observer agreement is good (kappa ~0.77–0.91) but the age and weight-bearing items depend on the history given, and isolated patellar tenderness on careful palpation — document what you asked and found.
- CT/MRI are more sensitive for occult tibial-plateau, osteochondral and Segond-type injuries — a negative rule with persistent inability to weight-bear beyond 48 hours, an effusion, or mechanical symptoms is a legitimate indication for cross-sectional imaging.
- The cost-saving holds across health systems — the rule consistently reduces knee radiography by ~20–30% without an increase in missed fractures.
The competing rule — the Pittsburgh Knee Rules
The examiner often asks whether you know an alternative to the Ottawa Knee Rules; the main one is the Pittsburgh Knee Rules, which apply specifically to a knee injury from blunt trauma or a fall:
- Radiograph if the mechanism is blunt trauma/a fall AND either age younger than 12 or older than 50 years, OR an inability to walk four weight-bearing steps in the emergency department. (Exclusions: injury more than ~6 days old, prior surgery/fracture on that knee, or re-assessment.)
- How they compare — both rules have very high (near-100%) sensitivity for fracture, but the Pittsburgh rule is generally reported to have higher specificity (so it sends fewer uninjured knees for radiographs). The trade-offs are that Pittsburgh requires a qualifying mechanism and uses an age cut-off at both extremes, whereas Ottawa adds the patellar/fibular-head tenderness and flexion criteria.
- Bottom line — Ottawa is the more widely taught and used; Pittsburgh is the credible, more-specific alternative worth naming. Either way the same safety nets apply: acute injury only, and a negative rule excludes a fracture, not internal derangement.
Mnemonic
4 PsWhen NOT to use it
P
Pregnant
Avoid ionising radiation where possible — image on clinical judgement
P
Penetrating injury
Open wounds follow a different pathway; the rule does not apply
P
Polytrauma / distracting injury
ATLS governs; the rule does not apply to the multiply injured
P
Pre-imaged / re-assessment
Already imaged this episode — the rule is inappropriate
Hook:4 Ps — do NOT apply the Ottawa Knee Rules if the patient is Pregnant, has a Penetrating injury, has Polytrauma/distracting injury, or is being Pre-imaged/re-assessed.
Viva practice
The points that win marks
- Five criteria (AF-IPP): Age ≥55, Fibular head tenderness, Isolated patellar tenderness, can't Passively flex to 90°, can't Partially weight-bear 4 steps — ANY ONE positive = image.
- Weight-bearing = 4 steps at BOTH time points (scene + ED); limp counts as weight-bearing.
- ~98–100% sensitivity, low specificity by design; ~20–30% fewer radiographs.
- Do NOT apply in pregnancy, deformity, penetrating/multi-trauma, or re-imaging.
- Fracture rule, not soft-tissue rule — a negative rule doesn't exclude ACL/meniscal/osteochondral injury.
- Paediatric: validated from age 2, used from ~age 5 by convention.
Exam viva
Practise clinical reasoning and management decisions out loud
Viva scenarioStandard
Clinical prompt
“A 34-year-old man twists his knee playing football. He is tender only over the medial joint line, can flex to 110 degrees, has walked into the department limping, and is 32 years old. He cannot recall whether he could bear weight immediately after the injury. Examination shows no deformity, no effusion, and a stable knee. Does he need radiographs?”
Viva scenarioAdvanced
Clinical prompt
“A 17-year-old girl is brought in after a fall during netball. She has an obvious tense effusion, a guarded knee, and cannot bear weight. She is tender over the lateral femoral condyle. How do you apply the Ottawa Knee Rules and what is your initial work-up?”
Exam cheat sheet
Exam day cheat sheet
Ottawa Knee Rules — one-screen revision
The five criteria
- Age 55 years or older
- Tenderness at the head of the fibula
- Isolated tenderness of the patella
- Inability to flex the knee to 90°
- Inability to bear weight for four steps, both immediately after injury and in the ED
Apply and act
- Any one criterion positive = standard knee series (AP, lateral, skyline; tunnel if indicated)
- All five negative = clinically clear: RICE, analgesia, early mobilisation, return precautions
- Negative rule but high suspicion = image anyway and document the reason
Caveats & modern context
- Sensitivity ~98–100%, specificity ~19–49% — designed to over-image a little
- Do NOT use in pregnancy, obvious deformity, penetrating injury, multi-trauma, or re-assessment
- Reduces knee radiography by ~20–30%; consistent across health systems and seniority
- A negative rule does not exclude internal derangement — persistent locking/effusion/non-weight-bearing beyond 48 h warrants MRI
Evidence
Evidence
Derivation of a decision rule for the use of radiography in acute knee injuries
Stiell IG, Greenberg GH, Wells GA, et al. • Ann Emerg Med (1995)
Key Findings:
- Prospective derivation in 1047 adults; from 23 candidate findings, five variables predicted the 68 knee fractures: age ≥55, fibular head tenderness, isolated patellar tenderness, inability to flex to 90°, and inability to bear weight 4 steps.
- The derived rule had 100% sensitivity (95% CI 0.95–1.0) and 54% specificity for clinically significant fracture.
- Applying it would have reduced knee radiography by ~28% (from 68.6% to 49.4%).
Clinical implication: Established the five-criterion rule that subsequent validation confirmed and most EDs now use as a default decision tool.
Limitation: Single-centre derivation; required prospective validation before adoption.
Source: Ann Emerg Med 1995;26(4):405-13
Evidence
Prospective validation of a decision rule for the use of radiography in acute knee injuries
Stiell IG, Greenberg GH, Wells GA, et al. • JAMA (1996)
Key Findings:
- Prospective validation in 1096 adults; the rule was 100% sensitive (95% CI 0.94–1.0) for the 63 clinically important fractures.
- The probability of fracture when the rule was negative was 0% (95% CI 0–0.4%); physicians interpreted it reliably (kappa 0.77).
- Potential radiography reduction ~28%; attempts to refine the rule improved specificity only at an unacceptable cost to sensitivity.
Clinical implication: Confirmed the derivation performance and supported the rule as a frontline ED screening tool.
Limitation: Few high-energy mechanisms; not yet extended to children.
Source: JAMA 1996;275(8):611-5
Evidence
Implementation of the Ottawa Knee Rule for the use of radiography in acute knee injuries
Stiell IG, Wells GA, Hoag RH, et al. • JAMA (1997)
Key Findings:
- Before/after controlled trial across 4 EDs (3907 patients): implementation produced a 26.4% relative reduction in knee radiography in the intervention hospitals versus 1.3% in controls, with no missed fractures.
- The rule was 100% sensitive for the 58 fractures and reliably interpreted (kappa 0.91).
- Patients discharged without radiography spent less time in the ED and incurred lower charges.
Clinical implication: Confirmed the rule is safe, reproducible and resource-saving when implemented prospectively across diverse EDs.
Limitation: Before/after design; small numbers lost to follow-up (none with a reported missed fracture).
Source: JAMA 1997;278(23):2075-9
Evidence attribution
According to PubMed, the rule was derived by Stiell et al. 1995 (DOI 10.1016/s0196-0644(95)70106-0; 100% sensitivity, ~28% reduction), prospectively validated by Stiell et al. 1996 (PMID 8594242; 0% fracture probability when negative) and implemented by Stiell et al. 1997 (PMID 9403421; 26.4% reduction, kappa 0.91). The pooled accuracy (sensitivity 98.5%, specificity 48.6%, negative LR 0.05) comes from Bachmann et al. 2004 (DOI 10.7326/0003-4819-140-5-200403020-00013), and the paediatric validation (100% sensitive, 31.2% reduction) from Bulloch et al. 2003 (DOI 10.1067/mem.2003.196).