Retrospective review of a Clinical Prediction Rule (CPR) on the outcome of hospitalized patients with Community Acquired Pneumonia (CAP) in an urban teaching hospital

Prior studies have shown that a CPR (MJ Fine, et al., N Engl J Med 336:243, 1997) may be a useful indicator of mortality associated with CAP. We performed a retrospective review to determine the value of the CPR with other variables in predicting the outcome of CAP in patients admitted to an urban county teaching hospital (Ben Taub General Hospital). 644 charts from two respiratory disease seasons, (10/94 to 3/95 and 10/95 to 3/96) with a discharge diagnosis of pneumonia were reviewed, and 212 met clinical and radiographic criteria for CAP. Reasons for exclusion were hospitalization within 30 days of the current admission, nosocomial acquisition of pneumonia, HIV infection with suspected PCP, and failure to meet criteria for CAP diagnosis. The demographics of the study population were as follows: mean age 46.8 yrs, 121 (57%) males, and 76% non-white. A bacteriologic diagnosis was established in 34.4%, with S. pneumoniae, accounting for 58% of the recognized pathogens. In hospital mortality increased with increasing CPR score: Class I-II (≤70) 0/95; Class III (71-90) 1/30; Class IV (91-130) 6/49; & Class V (> 130) 20/38. 22 Class I/II and 8 Class III patients had hypoxemia as an indication for hospitalization. Many variables also were found to predict mortality in univariate analyses, but only the following were found to be independently predictive of mortality in a stepwise multivariate logistic regression model: CPR, diagnosis of lung cancer or neurologic disease, need of mechanical ventilation, and a complication of pneumonia. A past diagnosis of chronic lung disease (asthma or COPD) was associated with a lower mortality rate. Readmission in year following admission for CAP was independently associated (P<.05) by stepwise multivariate logistic regression with the following: hospitalization in the preceding 90 d or 1 yr prior to admission for CAP, need for mechanical ventilation, anemia, and underlying lung disease or lung cancer. However, readmission was not independently associated with the CPR score. During the study period, the CPR identified approximately one third of patients admitted with CAP as being potential candidates for outpatient antibiotic therapy. Although socioeconomic factors not examined in this study may identify patients needing inpatient treatment, these results suggest that alternative management strategies of our low-risk CAP patients may result in a reduction of medical costs through elimination of unnecessary hospitalization.