Pulmonary embolism: a clinical prediction rule is useful in diagnosis.

Clinical bottom line (level 1a)

  1. A clinical prediction rule can help diagnose and exclude a pulmonary embolism.
  2. Further testing is required in all cases.
Wells et al: Annals of Internal Medicine 1998; 129 (12): 997-1005
Expires September 2003

The study

Setting: five university hospitals, Canada

1239 patients (aged ?, ?% male) suspected pulmonary embolism and symptoms for < 30 days

Excluded if
  • suspected upper-extremity DVT as the source of the PE
  • symptoms of PE for > three days before presentation
  • use of anticoagulants for > 72 hours
  • expected survival < 3 months
  • contraindication to contrast media
  • pregnancy
  • follow-up impossible due to geographic inaccessibility
  • <18 years old



    • Independent blinded reference standard, applied in all patients from a consecutive appropriate spectrum.
    Reference standard:
    • The following methods were used
      • ventilation-perfusion scan: classified as normal (no perfusion defects); high probability (one or more segmental or greater perfusion defects with normal ventilation, or two or more large subsegmental defects (>75% of a segment) with normal ventilation); or non-high probability
      • bilateral compression ultrasonography of lower limbs
      • follow-up for 3 months
      • if clinical rule and v/q scan discordant: venography followed by pulmonary angiography if negative
      Positive for PE if:
      • abnormal pulmonary angiogram, ultrasound or venogram
      • high-probability v/q scan plus moderate or high pretest probability
      • venous thromboembolic event during 3 month follow-up
      Patients with negative results had anticoagulation withheld.
    Diagnostic test: prospective validation of a clinical prediction rule. All patients had a history and physical examination, chest X-ray, oxygen saturations and ECG.
    • signs and symptoms- atypical for PE (respiratory or cardiac symptoms not meeting criteria for 'typical'): alternative diagnosis- likely or more likely: risk factors- no: probability of PE- low
    • signs and symptoms- atypical for PE (respiratory or cardiac symptoms not meeting criteria for 'typical'): alternative diagnosis- less or more likely: risk factors- yes: probability of PE- low
    • signs and symptoms- atypical for PE (respiratory or cardiac symptoms not meeting criteria for 'typical'): alternative diagnosis- less likely: risk factors- no: probability of PE- low
    • signs and symptoms- atypical for PE (respiratory or cardiac symptoms not meeting criteria for 'typical'): alternative diagnosis- less likely: risk factors- yes: probability of PE- moderate
    • signs and symptoms- typical for PE(2 or more respiratory points and heart rate > 90 beats/min, leg symptoms, low-grade fever, or results of CXR compatible with PE): alternative diagnosis- likely or more likely: risk factors- no: probability of PE- low
    • signs and symptoms- typical for PE(2 or more respiratory points and heart rate > 90 beats/min, leg symptoms, low-grade fever, or results of CXR compatible with PE): alternative diagnosis- likely or more likely: risk factors- yes: probability of PE- moderate
    • signs and symptoms- typical for PE(> 2 respiratory points and heart rate > 90 beats/min, leg symptoms, low-grade fever, or results of CXR compatible with PE): alternative diagnosis- less likely: risk factors- no: probability of PE- moderate
    • signs and symptoms- typical for PE(> 2 respiratory points and heart rate > 90 beats/min, leg symptoms, low-grade fever, or results of CXR compatible with PE): alternative diagnosis- less likely: risk factors- yes: probability of PE- moderate
    • signs and symptoms- severe (patient meets 'typical' definition, but also (1) syncope, (2) blood pressure <90 mmHg with heart rate >100 beats/min, (3) receiving ventilation or needs oxygen flow >40%, or (4) new-onset right heart failure (elevated JVP and new S1, Q3 and T3 or RBBB, or (4) plus (1), (2) or (3) regardless of other signs or symptoms: alternative diagnosis- less or more likely: probability of PE- moderate
    • signs and symptoms- severe: alternative diagnosis- less likely: probability of PE- high

    • Respiratory points:
      • dyspnoea or worsening of chronic dyspnoea
      • pleuritic chest pain
      • arterial oxygen saturation < 92% while breathing room air that corrects with 40% O2
      • haemoptysis
      • pleural rub
    • Risk factors:
      • surgery or fracture of lower extremity and immobilisation of fracture within 12 weeks
      • immobilisation (complete bed-rest) for three or more days in the previous four weeks
      • previous DVT or objectively diagnosed PE
      • strong family history of DVT or PE (two or more family members with objectively-proven events or first-degree relative with hereditary thrombophilia)
      • cancer (treatment on-going, within the past six months or in the palliative stages)
      • post-partum period
      • lower extremity paralysis
      • v/q scan results- normal: PE probability- low, moderate or high: next test- ultrasound scan: positive PE; negative- no PE
      • v/q scan- non-high: PE probability- low or moderate: next test- serial ultrasound scan (day 1, 3, 7, 14): positive- PE; negative- no PE
      • v/q scan result- non-high: PE probability- high: next test- ultrasound scan: positive- PE; negative- venogram: next test- venogram: positive- PE; negative- pulmonary angiography: next test- angiography: positive- PE; negative- no PE
      • v/q scan result- high: PE probability- low: next test- ultrasound scan; positive- PE; negative- venogram: next test- venogram: positive- PE; negative- pulmonary angiography: next test- angiography: positive- PE; negative- no PE
      • v/q scan result- high: PE probability- moderate or high: next test: PE

    The evidence

    pre-test probability of pulmonary embolism: 17%, (95% CI: 15% to 19%)

    diagnostic test pulmonary embolism no pulmonary embolism LR
    (95% CI)    		 post-test probability
    clinical prediction rule- high 80 42 9.2
    (6.5 to 13)     		66%
    moderate 112 291 1.9
    (1.6 to 2.2)     		28%
    low 25 709 0.17
    (0.12 to 0.25)     		3%
    total 210 1029


    diagnostic test low risk patients: pulmonary embolism no pulmonary embolism LR
    (95% CI)    		 post-test probability
    v/q result- high 4 9 22
    (9.0 to 56)     		31%
    v/q result- intermediate 1 68 0.74
    (0.12 to 4.6)     		1%
    v/q result- low 0 113 0.0
    (0.0 to 1.1)     		0%
    v/q result- normal 2 162 0.62
    (0.19 to 2.0)     		1%
    total


    diagnostic test moderate risk patients: pulmonary embolism no pulmonary embolism LR
    (95% CI)    		 post-test probability
    v/q result- high 32 1 87
    (12 to 620)     		97%
    v/q scan result- intermediate 8 33 0.66
    (0.33 to 1.3)     		20%
    v/q result- low 5 49 0.28
    (0.12 to 0.66)     		9%
    v/q result- normal 0 40 0.0
    (0.0 to 0.20)     		0%
    total


    diagnostic test high risk patients: pulmonary embolism no pulmonary embolism LR
    (95% CI)    		 post-test probability
    v/q result- high 21 3 3.8
    (1.3 to 11)     		88%
    v/q result- intermediate 5 3 0.89
    (0.25 to 3.2)     		63%
    v/q result- low 2 6 0.18
    (0.04 to 0.78)     		9%
    v/q result- normal 0 3 0.0
    (0.0 to 0.51)     		0%
    total

    K interobserver 0.85
    • Clinical prediction rule- high
      • number: 102; 8.2% (95% CI: 6.7% to 9.8%)
      • probability of PE: 66% (57% to 74%)
    • Clinical prediction rule- moderate
      • number- 403: 33% (95% CI: 30% to 35%)
      • probability of PE: 28% (23% to 32%)
    • Clinical prediction rule- low
      • number- 743: 60% (95% CI: 57% to 63%)
      • probability of PE: 3.4% (2.1% to 4.7%)
    • 46 patients required venography or angiography (3.7%: 95% CI: 2.7% to 4.8%).
    • 6 /1022 patients (0.6%: 95% CI: 0.1% to 1.1%) considered negative for PE had a venous thromboembolic event on follow-up. No patient was considered to have died from PE during follow-up.
    • 60% of patients had an alternative diagnosis made initially.

    Citation

    1. Wells PS, Ginsberg JS, Anderson DR, et al: use of a clinical model for safe management of patients with suspected pulmonary embolism. Annals of Internal Medicine 1998; 129 (12): 997-1005
    Search Terms: hand search
    Contributor: Chris Ball and Clare Wotton, September 2000
    Reviewer:

    Clinical Question.
    Patient suspected pulmonary embolism
    Intervention or Exposure clinical prediction rule
    Outcome diagnosis