The management of patients with community acquired pneumonia (CAP) is characterised by considerable variation in admission rates, length of hospital stay,¹ and use of institutional resources² in different settings. The Pneumonia Severity Index (PSI) is a prediction rule for the short term risk of death in patients with CAP,³ improving the efficiency of patient care.⁴ In the year 2000, 86% of patients with CAP presenting at the emergency department of our hospital were admitted. A retrospective analysis of the PSI scores of these patients showed that 37% of them were in low risk classes (1 and 2) based on their PSI results, so their admissions were potentially avoidable.^4,5 We therefore designed a prospective study to assess the safety, feasibility, and efficacy of the PSI score for management decisions in patients with CAP. The study was approved by the local ethics committee and written informed consent was obtained from all patients.

The study was carried out in the 12 month period from 1 November 2001 to 31 October 2002. One hundred and seventeen adult patients diagnosed in the emergency department with CAP participated in the study and were managed using a computer based score with dedicated software (GesPOrEx, Saxos software, Modena, Italy) for PSI calculation and data collection. CAP was defined as the presence of a pulmonary infiltrate on the chest radiograph and symptoms consistent with pneumonia including cough, dyspnoea, and pleuritic chest pain. Patients with severe immunosuppression, those admitted to hospital in the previous 15 days, and patients infected with HIV were excluded. According to published data,³ patients with PSI scores of 90 points or lower are recommended for outpatient treatment while those with higher scores are recommended for hospital admission. The score was used only as a guide to the admission decision and did not supersede clinical judgement. Follow up consisted of two visits, the first within 10 days and the second about 1 month after discharge from hospital. The choice of antibiotic treatment, route of administration, duration of antibiotic treatment, and criteria for discharge were according to local guidelines, mainly based on the recommendations of recently published guidelines.⁶ None of these interventions changed between the two study periods. To compare data before implementation of the protocol we retrospectively identified 116 consecutive patients admitted with CAP in the preceding year.

There were no statistically significant differences in demographic and co-morbidity data between the two groups (table 1). In both groups there was a significant proportion of patients in the lowest risk class; this probably reflects the attitude of patients in our healthcare structure to have frequent access to hospital services, particularly when the “family” doctor is unavailable such as at night or during the weekend. In the group managed after implementation of the protocol, 12 patients (10.3%) were admitted against PSI recommendations: six patients (or their relatives) strongly requested hospital admission, four were admitted for lack of adequate home care support, and two did not provide convincing assurance about compliance with treatment. Three (5.9%) of those admitted died; all were in class V of the PSI and two of the deaths were related to CAP. The implementation of PSI based management reduced the median duration of hospital stay from 9.1 (2.1) days to 7.9 (4.9) days, with a total reduction in bed days from 1070 to 463. Of the 1070 total bed days in the retrospective phase of the study, 348 (32.5%) were attributable to patients admitted with PSI scores in class I or II. All patients treated as outpatients were alive at the 1 month follow up visit and all returned to their usual activities. No adverse clinical outcomes, including admission to hospital or the intensive care unit, mortality or complications were detected. Compared with the historical data in the previous year, the rate of admission for CAP during the 12 month study period showed a 37% reduction (95% CI 26 to 49) which was statistically highly significant (p<0.001). The Italian health system estimates the cost in the use of hospital resources as about 1900 Euros per CAP patient treated as an inpatient. Use of this critical pathway significantly decreased the prevalence of admission, theoretically saving about 110 000 Euros in 1 year.