Early diagnosis of pneumococcal pneumonia.

Abstract

Sir, Streptococcus pneumoniae is the commonest cause of community-acquired pneumonia (CAP) and benzylpenicillin is first choice for antimicrobial treatment of pneumococcal infections, unless resistance is a problem. Frequently, patients with CAP receive broader-spectrum antibiotics and macrolides because diagnosis of pneumococcal pneumonia on clinical grounds is difficult, if not impossible. In addition to its effective action benzylpenicillin has several other advantages. The narrow spectrum of benzylpenicillin selects fewer resistant microorganisms with a lower chance of secondary infection. Clinical efficacy can be assessed within 48 h of therapy and the drug is cheap. If a causative pathogen is not found and there is a good clinical response, benzylpenicillin or an oral narrow-spectrum penicillin can be continued without risk of treatment failure. The widespread use of macrolides induces considerable resistance of bacteria to these drugs. 1 Selective use of benzylpenicillin and macrolides may help to halt this development. We have shown that with the parameters ‘presence of cardiovascular disease’, ‘acute onset of disease’, ‘presence of pleuritic pain’, ‘presence of Gram-positive bacteria in sputum smear’ and ‘the number of leucocytes in peripheral blood’, 80% of patients with CAP can be classified correctly as having pneumococcal pneumonia or pneumonia due to other pathogens. 2 Discriminant analysis yielded the following score: pneumococcal score � ‘cardiovascular disease’ (yes � 1, no � 0) � ‘acute onset’ (yes � 1, no � 0) � ‘pleuritic pain’ (yes � 1, no � 0) � 1.5 � ‘Gram-positive bacteria’ (yes � 1, no � 0) � 0.04 � ‘leucocyte count’ (10 9 /L)‐2.9. Since all of these parameters can be known shortly after hospital admission it should be possible to use this score to help to decide on initial treatment of a patient with CAP if the cut-off point indicating whether pneumococci or other pathogens are the most likely pathogens is known. To determine this cut-off point we calculated the score for 125 patients with CAP who were admitted to the Leiden University Medical Centre or Bronovo Hospital in The Hague. Blood cultures, sputum smears and cultures and serology were evaluated for each patient. Thirty-five patients had pneumococcal pneumonia as indicated by positive blood (19) or sputum cultures (16). The mean pneumococcal score for these patients was 0.267 (95% confidence interval: � 0.069‐0.603). Twenty-four patients had other pathogens: Haemophilus influenzae (10), Legionella pneumophila (3), Chlamydia spp. (3), Escherichia coli (3), streptococci (2), Nocardia asteroides (1), Pseudomonas aeruginosa (1) and Cryptococcus neoformans (1). The mean pneumococcal score for these patients was � 0.791 (95% confidence interval: 1.248 to � 0.334). In 66 cases no pathogens were identified. The receiver‐operator curve (Figure) shows the performance of the pneumococcal score as a test to diagnose pneumococcal pneumonia. With � 0.2 as cut-off point the sensitivity of the test is 0.69 and the specificity 0.79. The prevalence of pneumococci as the cause of CAP varies between studies from 10 to 40%, which is very likely to be an underestimate of the true prevalence. Assuming a prevalence of 50%, use of the pneumococcal score to decide whether to give benzylpenicillin (score greater than or equal to ‐0.2) or another antibiotic (score less than ‐0.2), would result in 45% of patients with CAP being treated with benzylpenicillin. The pneumococcal score offers the clinician a bedside tool that will restrict the use of broad-spectrum antibiotics and macrolides in the treatment of CAP.