Study objectives: This study was undertaken to evaluate the laboratory abnormalities observed in patients with bacterial pneumonia as predictors of the severity of illness.
Design: Retrospective analysis.
Setting: Tertiary care hospital.
Patients and participants: We studied 302 consecutive patients who were admitted to the Long Island Jewish Medical Center from January through December 1993 and treated for bacterial pneumonia. The patients were subdivided into two groups based on their serum phosphorus level either on hospital admission or 4 days before the onset of pneumonia, if this was acquired in-hospital. Hypophosphatemia (group 1) was defined as serum phosphorus level of < or = 2.4 mg/dL and normophosphatemia > 2.4 mg/dL (group 2). Three hundred randomly selected hospitalized patients treated for conditions other than pneumonia comprised the control group (group 3).
Measurements: Groups 1 and 2 were compared with respect to laboratory data, mortality rate, and duration of hospitalization. The laboratory data of patients in group 3 were compared with those treated for bacterial pneumonia (groups 1 and 2). Stepwise multivariate logistic regression analysis was employed to identify the variables that best predicted the onset of pneumonia.
Results: In groups 1 and 2, a greater (p < 0.0001) number of patients (135 of 302 patients with pneumonia, 44.7%) developed hypophosphatemia compared with patients in group 3 (31 of 300 control subjects, 10.3%). Patients with pneumonia (groups 1 and 2) had higher levels (p < 0.01) of bicarbonate compared with control subjects. Moreover, patients with pneumonia demonstrated lower levels (p < 0.01) of calcium, phosphorus, albumin, cholesterol, and alanine aminotransferase compared with control patients (group 3). Among patients with pneumonia, those with hypophosphatemia (group 1) had significantly lower levels (p < 0.05) of potassium, calcium, and albumin compared to those subjects with normophosphatemia (group 2). Furthermore, hypophosphatemic subjects manifested higher levels of glucose (p < 0.01) and creatine phosphokinase (p < 0.05) compared to their normophosphatemic counterparts. In addition, hypophosphatemic patients experienced a longer duration of hospital stay (hypophosphatemia, 24.6 +/- 2.0 days, vs normophosphatemia, 14.1 +/- 1.0, p < 0.001) and higher (p < 0.001) mortality compared to normophosphatemic subjects. The incidence of nosocomial pneumonia was higher (p < 0.0001) in hypophosphatemic patients compared to those with normophosphatemia.
Conclusion: We conclude that hypophosphatemia, hypocalcemia, hypokalemia, and hypoalbuminemia may be predictors of the severity of illness in patients admitted to the hospital with bacterial pneumonia.