Clinical or epidemiologic diagnosis of nosocomial pneumonia: is there any difference?

To evaluate the draft Centers for Disease Control and Prevention definitions of nosocomial pneumonia in comparison with current CDC definitions to increase acceptance by intensive care specialists.

A 12-bed medical intensive care unit in a German teaching hospital.

Clinicians were asked to retrospectively assess all patients of the year 1999 according to 3 different definitions of pneumonia: the current CDC definitions, the draft CDC definitions, and their own clinical criteria for diagnosing and treating. To test the agreement observed between the different sets of criteria, the kappa method described by Fleiss was used; for evaluating the strength of agreement between various categories, the recommendations of Landis and Koch were applied.

One thousand eight hundred forty-two patients were assessed. According to the clinical criteria, 31 patients developed nosocomial pneumonia during their stay in the intensive care unit. The number of pneumonia cases according to the current CDC definitions was 30 and according to the draft definitions, 32. However, agreement with the clinical definitions was intermediate for the current definitions kappa = 0.62 (95% CI, 0.47-0.76) but excellent for the draft definitions kappa = 0.98 (95% CI, 0.95-1.02).

Even though the current definitions are useful, agreement with the clinical assessment was greater for the draft definitions and might lead to improved acceptance of surveillance data by intensive care specialists.