Pneumonia: five clinical features can help diagnose it.

Clinical bottom line (level 1a)

  1. One in seven patients with suspected community-acquired pneumonia have it.
  2. A clinical prediction rule using temperature >37.8 ° C, pulse >100 beats/minute, rales, decreased breath sounds and absence of asthma can usefully rank patients for risk of pneumonia.
Heckerling et al: Annals of Internal Medicine 1990; 113: 664-670
Expires March 2003

The study

Setting: emergency department, 3 university hospitals, USA

1134 patients (aged mean 45 years, 57% female) with fever or respiratory symptoms, who had a chest X-ray

Independent ?blinded reference standard, applied in all patients from a consecutive appropriate spectrum.
Reference standard:
  • definite or probablepulmonary infiltrates on chest roentgenogram
Diagnostic test: clinical features
  • A logisitic regression analysis was performed to identify independent predictors of pneumonia.
  • A clinical prediction rule was created using 5 features
    • temperature > 37.8 degrees C: OR 2.69 (95% CI: 1.73 to 4.17)
    • pulse > 100 beats/min: OR 2.35 (95% CI: 1.52 to 3.65)
    • crackles: OR 3.73 (95% CI: 2.43 to 5.72)
    • decreased breath sounds: OR 3.58 (95% CI: 2.33 to 5.50)
    • absence of asthma: OR 3.98 (95% CI: 1.89 to 8.42)
  • This clinical prediction rule was validated in 302 patients in 2 other emergency departments.

The evidence

pre-test probability of community-acquired pneumonia (derivation set): 15%, (95% CI: 12% to 17%)
pre-test probability of community-acquired pneumonia (validation sets): 28%, (95% CI: 23% to 36%)

diagnostic test pneumonia (derivation set) no pneumonia LR
(95% CI)		 post-test probability
score 5 15 5 17
(6.5 to 47) 		75%
score 4 37 30 7.2
(4.6 to 11) 		55%
score 3 42 149 1.6
(1.2 to 2.2) 		22%
score 2 28 232 0.70
(0.50 to 0.99) 		11%
score 1 11 316 0.20
(0.11 to 0.36) 		3%
score 0 1 48 0.12
(0.017 to 0.87) 		2%
total 124 780


diagnostic test pneumonia (validation set) no pneumonia LR
(95% CI)		 post-test probability
score 5 6 0 -
(5.2 to infinity) 		100%
score 4 22 10 5.6
(2.8 to 11) 		69%
score 2 or 3 37 102 0.93
(0.72 to 1.2) 		27%
score 0 or 1 6 70 0.22
(0.10 to 0.48) 		8%
total 71 182

Comments

  1. As with all prediction rules, the nature of the patient population strongly influences the usefulness of the guide
  2. The ROC curves for the validation sets were not significantly different from the derivation set.

Citation

  1. Heckerling PS, Tape TG, Wigton RS, et al: Clinical prediction rule for pulmonary infiltrates. Annals of Internal Medicine 1990; 113: 664-670
Search Terms: pneumonia and diagnosis in Medline
Contributor: Clare Wotton & Chris Ball, November 1999
Reviewer: Mitsuhiro Kamei

Clinical Question.
Patient suspected pneumonia
Intervention or Exposure presence of clinical signs, symptoms or factors
Comparison absence of clincal signs, symptoms or factors
Outcome diagnosis of pneumonia