Comparative validation of prognostic rules for community-acquired pneumonia in an elderly population

Abstract

The aim of the study was to validate the prediction rule of M.J. Fine and coworkers for clinical outcome variables and three prognostic rules for the individual outcome of community-acquired pneumonia in an elderly population (rule 1: respiratory frequency > or =30 breaths x min(-1), diastolic blood pressure < or =60 mmHg, blood urea nitrogen >7 mM; rule 2: respiratory frequency > or =30 breaths x min(-1), diastolic blood pressure < or =60 mmHg, mental confusion; and rule 3: systolic blood pressure < or =80 mmHg, cardiac frequency > or =90 beats x min(-1), lactate dehydrogenase activity > or =260 IU x L(-1); death was predicted in the presence of at least two of three parameters). Overall 168 consecutive episodes of community-acquired pneumonia in patients aged > or =65 yrs and hospitalized in a primary care hospital were studied prospectively. Fine's rule was tested for its ability to predict length of hospital stay, requirement for intensive care unit (ICU) admission and death. For the three prognostic rules of individual outcome, performance regarding predicting death was determined. Mortality was 17/168 (10%). Fine's rule accurately predicted length of stay, the requirement for ICU admission and the risk of death from pneumonia as compared to the original derivation and validation cohorts. All three rules achieved moderate-to-high specificity (73%, 88% and 80%, respectively) and high negative predictive values (95%, 94% and 93%, respectively) but had a low sensitivity (65%, 47% and 47%, respectively). Rule 2 most closely reflected the risk of death from pneumonia when Fine's classification was used as reference. Fine's rule proved to give valid estimations regarding clinical outcome variables of community-acquired pneumonia in the elderly. The prognostic rules may be useful in determining individual patients at lower risk of death caused by pneumonia.