Evaluation of an Algorithm to Reduce Antibiotic Prescribing for Acute Bronchitis

The rapid rise in antibiotic resistance among common bacteria are adversely affecting the clinical course and health care costs of community-acquired infections. Because antibiotic resistance patterns are strongly correlated with antibiotic use patterns, multiple organizations have declared reductions in unnecessary antibiotic use to be critical components of efforts to combat antibiotic resistance. Among humans, the vast majority of unnecessary antibiotic prescriptions are used to treat acute respiratory tract infections (ARIs) that have a viral etiology. In particular, despite the fact that numerous controlled trials have demonstrated no benefit of antibiotic therapy for patients with acute bronchitis, the majority of patients diagnosed with acute bronchitis continue to receive antibiotic therapy across diverse treatment settings. Recently, the National Committee on Quality Assurance adopted the proportion of adult visits diagnosed as acute bronchitis when an antibiotic was NOT prescribed as a quality measure within the HEDIS data set. Recent results from the HEDIS dataset emphasize the continued high rates of antibiotic prescribing for patients with acute bronchitis. One key factor driving overuse of antibiotics in the management of patients with lower respiratory tract infections-such as acute bronchitis-is diagnostic uncertainty and inaccurate risk estimation of underlying pneumonia in such patients. Recently, our study team has observed substantial reductions in antibiotic prescribing following the incorporation of a diagnostic and treatment algorithm into an acute care setting. This acute cough management algorithm incorporates data on vital signs and symptoms distinguishing patients with community-acquired pneumonia from other patients with acute cough illness, specifically those with acute bronchitis. The acute cough management algorithm has become even more valuable in recent years due to the introduction of quality measures that emphasize the timely administration of antibiotics for patients with community-acquired pneumonia. Thus, strong empirical evidence of the effectiveness of such an algorithm could lead to wide adoption of the algorithm and substantial improvements in antibiotic prescribing. The investigative team is proposing a unique partnership with Geisinger Health System, a large integrated health network, to implement and evaluate the algorithm. Utilizing a cluster-randomized trial design across 33 practice sites, we will address the following aims: 1) To measure the reduction in antibiotic prescribing resulting from incorporation of the algorithm compared to usual care sites utilizing two different implementation strategies, one poster-based and one electronic health record-based, 2) To measure revisits, delayed hospitalizations and net economic costs associated with algorithm implementation, and 3) To evaluate local practice characteristics influencing the level of implementation and ultimate performance success at intervention sites. In a final component of the study, the investigators will partner with NCQA to disseminate study results through the national network of participating plans and stimulate wide spread adoption of the algorithm and quality improvement methods.