Feasibility of a computerized community-acquired pneumonia care process model in an outpatient setting

Intermountain Health Care experts implemented a paper-based protocol addressing community-acquired pneumonia (CAP) treatment in 1995. LDS hospital's emergency department (ED) used the paper protocol for less than 10% of their CAP patients. ED physicians did not recognize a benefit and were intolerant of the extra paper work required. The purpose of this research was to use the ED's existing information system to create a portable CAP protocol that was integrated into the ED workflow to save time and effort of the physicians and to assess its clinical feasibility. A computerized algorithm was implemented in Arden Syntax and reviewed for portability. Thirty of the 39 Arden Syntax medical logic modules were directly transportable. The work required by physicians could have been reduced by 76%. Two hundred sixty-three eligible (CAP symptoms, age > 17, immunocompetent) CAP patients were studied and identified by ICD-9 codes 480-486. Physicians ordered 245 lab tests whereas the protocol suggested ordering 152. Physicians ordered 255 chest x-rays whereas the protocol suggested ordering 211. Physicians admitted 181 CAP patients whereas the protocol suggested admission for 149 CAP patients, 134 of the 149 were admitted. Physicians admitted 47 patients that the CAP CPM did not recommend admission for. Twenty patients were randomly selected in a manual chart review. Physicians prescribed cephalosporins for 17 patients; 5 of the 17 were combined with a macrolide; antibiotics were not mentioned for 3 patients. The computerized CAP protocol recommended cephalosporins plus a macrolide for 11 patients, penicillin for 2, a macrolide for 4, and made no reconmnendation for 3. Based on an 80% CAP CPM compliance for 263 CAP patients, the projected cost savings would have been $208,443. Ten percent fewer patients would have been admitted. Forty-four percent would have been treated with a cephalosprin plus a macolide. Sixteen percent would have been treated with a macrolide. Eight percent would have been treated with penicillin. The CAP CPM made no treatment recommendation for 32% of patients; antibiotic therapy would have been the physician's choice. An exportable computerized CAP protocol was successfully created and integrated into the ED workflow and clinical feasibility was assessed.