Context: There are up to 1 million patients treated in acute-care hospitals for community-acquired pneumonia (CAP), with an estimated annual cost > 8 billion dollars. A newly validated CAP outcomes prediction rule developed by Fine and colleagues has been advocated as a guide to hospitalization decisions.
Objective: To evaluate the clinical characteristics, costs of care, and resource utilization of patients with low-risk CAP at an urban public hospital serving an indigent population.
Design, setting, and participants: Prospective cohort study from June 1, 1994 to May 31, 1996.
Main outcome measures: Clinical characteristics and costs of care of patients with low-risk CAP and features associated with low-risk CAP in this population.
Results: A total of 522 patients were identified at the time of hospital admission as having CAP; 97 patients (19%) were HIV positive on hospital admission and excluded. Of the remaining 425 patients, 253 patients (60%) were classified as pneumonia severity index (PSI) class I-III (low risk). Of the patients with low-risk CAP, only four patients (1.6%; 95% confidence interval, 0.4 to 4.0%) died during hospitalization. Low-risk CAP was both costly and accounted for significant resource use (35.4% of total CAP costs, and 45% of all CAP bed days). Of the patients with low-risk CAP, there were 138 patients (55%) who could potentially have been treated as outpatients (absence of altered mental status, hypotension, hypoxia on hospital admission, or direct ICU admission). However, 49% of these patients had a history of alcoholism, 20% had a blood alcohol level > 50 mg/dL, and 44% were homeless.
Conclusions: A significant proportion of the patients admitted with CAP to a public hospital had low-risk CAP and accounted for a significant proportion of the CAP bed days and costs. The use of the PSI accurately predicted which patients would be at low risk for death; however, the utility of using the PSI to reduce low-risk CAP hospital admissions would have been of limited benefit. High rates of homelessness, substance abuse, and medical needs not captured in the PSI would preclude many of these patients from unsupervised outpatient treatment.