To describe the natural history of community-acquired pneumonia in the subset of a large cohort of patients at low risk for mortality who were admitted to the hospital.

Prospective observational study of all patients at low risk for mortality (risk classes I and II) who presented to 6 hospitals and 1 emergency department in Edmonton, Alberta, Canada with a diagnosis of possible community-acquired pneumonia from November 15, 2000, to November 14, 2002.

A total of 586/3065 (19.1%) low-risk patients (Fine criteria) were admitted, 48.4% of whom stayed more than 5 days. Multivariate analysis revealed that patients who were admitted were more likely to be female, to have presented at Site B, which serves an inner city population, to have diminished premorbid functional status, to have comorbidities likely to be made worse by pneumonia (chronic obstructive pulmonary disease, asthma, heart disease, inflammatory bowel disease), and to suffer from substance abuse or psychiatric illness. A respiratory rate of ≥28 breaths per minute, and symptoms of shaking chills, shortness of breath, nausea or diarrhea were the remaining factors predicting admission. Nineteen percent of the patients suffered one or more complications, the most serious of which was progression of the pneumonia, resulting in respiratory failure necessitating mechanical ventilation in 2.4% and empyema in 1.4%. Four patients had lung cancer, and 1 had cancer of the vocal cords. Thirty-one percent of those who were admitted were still unable to eat or drink enough to maintain hydration by hospital day 5 or on discharge day.

One in 5 patients at low risk for mortality were admitted to the hospital and half stayed more than 5 days; 19% suffered 1 or more complications. Our data emphasize the need for better rules to guide the admission decision and the importance of physician judgment in this decision.