Decision Analysis of Treatment Strategies for Ventilator-Associated Pneumonia

Abstract & Commentary

Synopsis: In late-onset VAP, survival improved and costs decreased using initial coverage with 3 antibiotics. Mini-BAL did not improve survival but decreased costs and antibiotic usage.

Source: Ost DE, et al. Am J Respir Crit Care Med. 2003;168:1060-1067.

The optimal strategy for managing ventilator-associated pneumonia (VAP) remains controversial. To clarify the tradeoffs, Ost and colleagues developed a decision-analysis model to simultaneously examine outcomes of 16 diagnostic and treatment strategies. The "subjects" were a cohort of immunocompetent critically ill patients with the following characteristics: 1) intubated for 7 days; 2) evidence of late-onset VAP based on Centers for Disease Control and Prevention (CDC) criteria of fever, purulent secretions, leukocytosis, and radiographic infiltrates; and 3) estimated mortality of 20%. Five assumptions were built into the model. First, antibiotics would be chosen based on American Thoracic Society guidelines adapted to local formularies and ICU pathways. Second, antibiotics would be given immediately after the patient met CDC criteria for VAP and continued until diagnostic test results returned. Third, antibiotics would be adjusted to cover any identified pathogens and unnecessary antibiotics discontinued. Fourth, if all cultures were negative and the patient had ongoing severe sepsis or was unstable, antibiotics would be continued. Fifth, if the patient was stable, antibiotics would be discontinued.

There were 4 treatment strategies (zero, 1, 2, or 3 antibiotics) and 4 diagnostic strategies: 1) no diagnostic testing; 2) endotracheal tube aspirate quantitative cultures; 3) bronchoscopic cultures; or 4) nonbronchoscopic mini-BAL quantitative cultures. Initial coverage with 3 antibiotics was better than expectant management (zero antibiotics) or 1 or 2 antibiotics, leading to both improved survival (54% vs 66%) and decreased costs ($55,447 vs $41,483 per survivor). Testing with mini-BAL did not improve survival but did decrease costs ($41,483 vs $39,967) and antibiotic use (63 vs 39 antibiotic days per survivor). From the perspective of minimizing cost, minimizing antibiotic use, and maximizing survival, the best strategy was 3 antibiotics with mini-BAL.

Comment by Leslie A. Hoffman, PhD, RN

VAP frequently complicates the course of critically ill patients on mechanical ventilation and is associated with a high mortality. Survival is highly dependent on selecting the appropriate initial antibiotic. Although diagnostic tests often lead to a change in therapy, it requires time to obtain the results, and when they become available it may be too late to alter survival. Therefore, many advocate that high-risk populations be initially treated with broad-spectrum antibiotic therapy. The decision model that Ost et al used was developed from a literature search that returned 555 citations, later reduced to 111 articles based on search criteria. The final model analyzed multiple outcomes including survival, cost, cost per survivor, antibiotic use, antibiotic use per survivor, and the combined perspective of financial and antibiotic cost per survivor. From a combined perspective, a 3 antibiotic plus mini-BAL strategy was superior to all 1 and 2 antibiotic strategies, irrespective of diagnostic technique, in terms of minimizing antibiotic use and financial cost per additional survivor. Diagnostic testing alone had little effect on survival but was cost-effective because it decreased unnecessary antibiotic usage.

Often overlooked as a technique, decision analysis offers a useful perspective on complex management challenges, such as those presented by VAP. The technique allows simultaneous testing of more options than are possible in a clinical trial. It is ideally suited to situations with multiple clinical options, a large literature base, and continuing uncertainty about the best approach. Decision analysis is subject to accuracy of the basic assumptions used to build the analytic model and unable to project the consequences of future events, such as the emergence of antibiotic resistance and the consequence of this outcome. Given these limitations, it provides an interesting approach to the analysis of a very complex problem.

Leslie A. Hoffman, PhD, RN, Department of Acute/Tertiary Care, School of Nursing, University of Pittsburgh, is Associate Editor of Critical Care Alert.