Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: Clinical assessment is inaccurate for diagnosing community-acquired pneumonia

ACP J Club. 1998 May-June;128:72. doi:10.7326/ACPJC-1998-128-3-072

Source Citation

Metlay JP, Kapoor WN, Fine MJ. Does this patient have community-acquired pneumonia? Diagnosing pneumonia by history and physical examination. JAMA. 1997 Nov 5;278:1440-5.



To determine the accuracy of findings based on history taking and physical examination (without chest radiography) for diagnosing community-acquired pneumonia.

Data sources

English-language studies were identified by using MEDLINE (1966 toOctober 1995) with the search terms {physical examination; medical history taking; professional competence; sensitivity and specificity; reproducibility of results; observer variation; diagnostic tests, routine; decision support techniques; and pneumonia}*. Bibliographies of papers were also reviewed.

Study selection

Studies were selected if they assessed the accuracy or precision of history taking or physical examination for diagnosing community-acquired pneumonia. Studies were excluded if patients were< 16 years of age, had known immunosuppression, or had nosocomial infections or if the studies were case series(< 10 patients) or review articles with no original data.

Data extraction

Only studies assessed to have high methodologic quality (level I evidence) were included in the main analyses (i.e., prospective studies with independent, blind comparisons of clinical findings with a gold standard among > 50 consecutive patients suspected of having community-acquired pneumonia or studies with ≥ 2 independent blinded raters of signs or symptoms in patients suspected of having community-acquired pneumonia). Likelihood ratios for the presence of disease if the finding was positive (+LR) and if the finding was negative (-LR) were calculated.

Main results

4 studies were included in the main analyses. All 4 took place in emergency departments, but they varied in patient selection criteria. No individual item from the history or physical examination was identified whose presence or absence consistently (across studies) would eliminate the need for chest radiography. Asthma (+LR 0.10, -LR 3.8) and dementia (+LR 3.4) were the most influential history items in one study. Absence of vital sign abnormalities (< 30 breaths/min, heart rate < 100 beats/min, and temperature < 37.8 _C) had a -LR of 0.18 in one study. Absence of crackles had a -LR of ≥ 0.62. 4 clinical decision rules were tested independently in a prospective trial and compared with physician judgment in predicting the results of chest radiography. Physician judgment had the best -LR at 0.25, and a rule based on having ≥ 1 vital sign abnormality had the best +LR at 2.6.


The presence or absence of individual and combined signs and symptomsassessed with medical history or physical examination (without chest radiography) are not accurate for diagnosing community-acquired pneumonia.

Sources of funding: No external funding.

For article reprint: Dr. M.J. Fine, Montefiore University Hospital, 8 East Room 824, 200 Lothrop Street, Pittsburgh, PA 15213, USA. FAX 412-692-4892.

*Information supplied by authors.


Metlay and colleagues found little high-quality data on the accuracy of clinical assessment in the diagnosis of community-acquired pneumonia. They used sensible inclusion and exclusion criteria and found 52 studies; only 4 studies had level I evidence, and all were set in emergency departments.

The results of the review are disturbing for anyone with great faith in the accuracy of clinical assessment: No part of the history contributed much to the diagnostic accuracy of the assessment, with the exception, in 1 study, of a history of asthma (+LR 0.10) or dementia (+LR 3.4). These LRs compare unfavorably with that for typical angina, which is approximately 100 (1). The examination was slightly more helpful: The absence of any abnormality in the patient's temperature, pulse, or respiratory rate reduced the odds of pneumonia to a fifth of thatbefore examination. Some chest signs had reasonable LRs but were uncommon and had poor reproducibility.

This review gives clinicians some valuable information about what might or might not contribute to the diagnosis of community-acquired pneumonia, but its strongest message is for clinical researchers. We need high-quality evidence about the role of clinical assessment, especially in a broad range of settings. Additional research questions should address the usefulness of clinicalassessment in terms of health outcomes.Finally, the pivotal role of cost in health service decision making demands high-quality data about the relative cost andutility of diagnostic strategies (2).

Michael J. Hensley, MBBS, PhD
David Arnold, MBBSNewcastle University Medical SchoolNewcastle, New South Wales, Australia


1. Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology: A Basic Science for Clinical Medicine. 2d ed. Boston: Little, Brown; 1991.

2. Muir Gray JA. Evidence-based Healthcare. New York: Churchill Livingstone; 1997.