Data Driven Stewardship: Using Databases to Target Inappropriate Antimicrobial Use for Viral Respiratory Infections

Purpose: Antimicrobial stewardship programs (ASP) are effective at reducing inappropriate antimicrobial usage that contributes to antimicrobial resistance and adverse medical outcomes. Antibiotic prescribing for outpatient viral acute respiratory tract infections (ARI) has been implicated in contributing to inappropriate antibiotic usage. We sought to develop database-driven algorithms to determine if this was an area of inappropriate antibiotic prescribing for our institution, characterize the scope of the problem, and determine how this could be targeted through antimicrobial stewardship. Methods: Antibiotic prescribing for viral ARI was examined in outpatient primary care clinics and the emergency department (ED) of a large military medical center using the military health system management and analysis reporting tool. We developed an algorithm integrating inclusionary and exclusionary diagnostic codes with pharmacy data to robustly measure inappropriate antibiotic use associated with viral ARI. Adult and pediatric patient encounters from 2013-2017 were evaluated using relevant viral ARI ICD-9 and ICD-10 diagnostic codes; encounters with concurrent diagnoses of asthma/COPD, bacterial pneumonia, bacterial pharyngitis or otitis media were excluded. Frequencies of ARI diagnoses and antibiotic dispersal were calculated. A subgroup analysis was performed looking specifically at viral pharyngitis diagnoses among pediatric patients by incorporating laboratory data to determine if bacterial infections were identified but not coded as such. Results: Our analysis identified 17,986 cases of viral ARI, for which 3,889 (21.6%) were prescribed an unnecessary antibiotic. Pediatric patients had the lowest rate of unnecessary antibiotics prescriptions at 8% of cases. Diagnoses most associated with unnecessary antibiotic prescribing were acute rhinosinusitis (74%) and pharyngitis (21%) for pediatric medicine; acute rhinosinusitis (66.1%), uncomplicated bronchitis (59%), and pharyngitis (26%) for adult medicine; and acute rhinosinusitis (67%), viral pneumonia (70%) and uncomplicated bronchitis/bronchiolitis (57%) for the ED. For the subgroup analysis of pediatric viral pharyngitis encounters receiving antibiotics, we determined that 19% of patients had laboratory data supportive of a diagnosis of streptococcal pharyngitis. However, the vast majority (81%) of pediatric encounters coded with viral pharyngitis were determined to have no alternate diagnosis or laboratory data supportive of a bacterial infection requiring antibiotics, and therefore were appropriately identified as receiving unnecessary antibiotic prescriptions. Conclusion: Medical databases are a tool for ASP, however robust algorithm development and testing is necessary to appropriately measure inappropriate antibiotic usage. By developing an algorithm for evaluation of antibiotic prescribing associated with viral ARI, we found that unnecessary antibiotic prescriptions for viral ARIs were common at our facility, occurring in approximately 1 in 5 ARI encounters facility-wide, and in 1 in 12 pediatric primary care ARI encounters. Even after subgroup analysis matching laboratory data to coding and pharmacy data, the frequency of encounters in which unnecessary antibiotics were prescribed remained high. These findings identified clinic-specific targets for stewardship and benchmarks for measuring improvement.