A practical application of practice-based learning: development of an algorithm for empiric antibiotic coverage in ventilator-associated pneumonia.

Abstract

BACKGROUND: Development of practice-based learning (PBL) is one of the core competencies required for resident education by the Accreditation Council for Graduate Medical Education, and specialty organizations including the American College of Surgeons have formed task forces to understand and disseminate information on this important concept. However, translating this concept into daily practice may be difficult. Our goal was to describe the successful application of PBL to patient care improvement with development of an algorithm for the empiric therapy of ventilator-associated pneumonia (VAP). METHODS: The algorithm development occurred in two phases. In phase 1, the microbiology and timing of VAP as diagnosed by bronchoalveolar lavage was reviewed over a 2-year period to allow for recognition of patterns of infection. In phase 2, based on these data, an algorithm for empiric antibiotic coverage that would ensure that the large majority of patients with VAP received adequate initial empiric therapy was developed and put into practice. The period of algorithm use was then examined to determine rate of adequate coverage and outcome. RESULTS: : In Phase 1, from January 1, 2000 to December 31 2001, 110 patients were diagnosed with VAP. Analysis of microbiology revealed a sharp increase in the recovery of nosocomial pathogens on postinjury day 7 (19% < day 7 versus 47% > or = day 7, p = 0.003). Adequate initial antibiotic coverage was seen in 74%. In Phase 2, an algorithm employing ampicillin- sulbactam for coverage of community- acquired pathogens before day 7 and cefipime for nosocomial coverage > or =day 7 was then employed from January 1, 2002 to December 31, 2003. Evaluation of 186 VAP cases during this interval revealed a similar distribution of nosocomial cases (13% < day 7 versus 64% > or = day 7, p < 0.0001). Empiric antibiotic therapy was adequate in 82% of cases or =day 7: overall accuracy improved to 83% (p = 0.05). Mortality from phase 1 to phase 2 trended toward a decrease (21% versus 13%, p = 0.1). CONCLUSIONS: Application of the concept of PBL allowed for identification of local patterns of infection and development of an institution specific treatment algorithm that resulted in >80% adequate initial empiric coverage for VAP with a trend toward decreased mortality. PBL allows for alteration in practice based on local patterns and outcomes and has the potential to improve patient care.