Discussion

Our study adds to the published literature on the feasibility of collaboration between multiple health systems to create regional antibiograms. Previous published studies by Guarascio, Butler, and Var included a larger number of institutions in their studies including large academic medical centers. ^{Reference Var, Hadi and Khardori9–Reference Butler, Biagi and Gupta11} The AntibiogramDSM study included data from four separate health systems (representing 8 hospitals) in the Des Moines metro. One main goal of this project was to increase awareness locally related to bacterial resistance. Among those providers that were surveyed in the educational session, 82.2% did not have access to an antibiogram. Very few of the surveyed providers could appropriately identify local resistance patterns for common bacteria such as E. coli and S. pneumoniae. In fact, no individual correctly identified the resistance range of all of the antibiotics presented in the survey. Prior to education, there was a tendency to overestimate resistance of S. pneumoniae to many antibiotics (except macrolides which was underestimated). This may possibly result in unnecessarily broad-spectrum therapies for respiratory tract infections on an outpatient basis. However, E. coli resistance to guideline-recommended agents used for UTIs tended to be underestimated by those surveyed, which could lead to treatment failures and repeat exposures to antibiotics. In fact, a recent cohort study showed nonsusceptible, uncomplicated urinary tract infection isolates resulted in 1.4 more prescriptions per patient, higher probability of subsequent healthcare encounters, and $426 higher all-cause costs. ^{Reference Shafrin, Marijam and Joshi15} Furthermore, in a survey, 75% of ambulatory care physicians felt that poor access to antibiograms could possibly impact prescribing and be an important cause of antimicrobial resistance. ^{Reference Harris, Chandramohan, Awali, Grewal, Tillotson and Chopra16}

Our study was multilayered and included the creation of the antibiogram, education, and public availability of the antibiogram. Previous studies by Guarascio and Butler discuss the creation of the regional antibiograms and the compiled susceptibility results. ^{Reference Guarascio, Brickett, Porter, Lee, Gorse and Covvey10,Reference Butler, Biagi and Gupta11} However, these studies did not elaborate on the implementation or utilization of the regional antibiogram. Prior surveys show that ambulatory care physicians recognize the need for additional education related to their prescribing practices, particularly in the area of antibiotic resistance. ^{Reference Harris, Chandramohan, Awali, Grewal, Tillotson and Chopra16} This study utilized a local antibiogram as a tool to support education and allow providers to select agents informed by both national guidelines and local resistance patterns. Participants were able to select more appropriate pharmacotherapy in all four of the case-based questions in our survey following education and viewing the combined antibiogram. This achieved statistical significance in all but the ABS case, which already had high performance in the pre-education survey.

One major concern is the concept of using a mix of cultures from patients from the community presenting to a local hospital and those obtained from within the hospital. While all the health systems in this study built their antibiograms this way, this method may result in inaccurate assumptions about community resistance rates due to inclusion of nosocomial pathogens in the sample, particularly among gram-negative pathogens. ^{Reference McGregor, Bearden and Townes17} A prior comparison of an outpatient antibiogram and inpatient antibiogram found some statistically significant differences in gram-negative resistance to some antimicrobials versus E. coli, Klebsiella, and Pseudomonas. Most variance was within 10%. However, using data sets that include inpatient cultures may be the only method to have adequate isolate counts for outpatient providers to gain access to an antibiogram in some settings. Creation of an outpatient only combined antibiogram would require all health systems to define cultures as inpatient or outpatient consistently, including those with a community-acquired infection, which was not cultured until the patient was admitted to a hospital. It may also require integration of cultures from clinics in the community, which may be operating independently of local health systems. While a combined antibiogram may theoretically present higher resistance rates than the community, a cursory comparison of the 2019 AntibiogramDSM (educational session edition) to SENTRY Public the same year shows lower MRSA rates (37% vs 41.6%) and lower E. coli resistance with ciprofloxacin (25% vs 27.1%) or trimethoprim-sulfamethoxazole (22% vs 32.5%). ¹⁸ Furthermore, the participants in the educational session tended to assume greater resistance rates with common pathogens.

Our study had several other limitations. The methodology for each antibiogram year changed slightly. The initial pilot antibiogram included the largest antibiogram representing 18 mo, then the same institution having some cephalosporins censored for enterobacterales, and another organization did not participate due to COVID. This resulted in significant decreases in total isolates included in the antibiogram per year. Also all health systems used different microbiology testing platforms that were updated during various antibiogram years. This might have included different timing to accept updated breakpoint guidance. In this study, we chose to accept how each institution evaluated the susceptibility breakpoint and applied them to their own antibiograms rather than rely on MIC reporting. Such a method would allow greater uniformity in interpretation, but would be laborious. The health systems also created their individual antibiograms within different time frames varying from every year to every 18 mo. This required coordination with a microbiology lab at one institution and may have resulted in more organisms or antibiotics being included in the data set than if the institution’s official antibiogram was provided.

The live CME was presented in January 2021, and online on-demand education was available during the following year. The timing of the educational opportunities occurred during the COVID-19 pandemic. This resulted in a lower than anticipated participation rate in the study. Several of the education session participants were trainees, which may have impacted preeducation responses. Furthermore, the education assessment relied on immediate post-education assessment as a surrogate for knowledge gained and possible changes in prescribing patterns. Some case-based assessments did not require direct knowledge of local susceptibilities. It was felt that assessment of other learning objectives of the CME program was important to address as well. Regardless, in pre-education, it appeared that there was a knowledge gap in the estimation of local resistance patterns. Availability of local resistance data is crucial to informing the development of infectious disease CME. Future studies should assess if the presence of a combined local antibiogram impacts actual prescribing quality.

Following the completion of this study, CLSI released updated guidance related to the creation of antibiograms. ¹⁹ Included in this document are recommendations for multifacility antibiograms and considerations for collection of data to be included in such antibiograms. Data could be centrally collected by a single laboratory or with isolate-level data to produce an antibiogram similar to methods of a single hospital, but this may be an unrealistic option for competing health systems. Although combining individual antibiograms may lead to some inaccuracies due to variability in development methods, lab abilities, and antibiotics presented, it is a more practical way for separate institutions to merge antibiograms. It requires that each stakeholder communicates changes in their processes, particularly with recent changes in CLSI breakpoints. ²⁰

Health systems should consider collaborating and creating combined antibiograms to be shared with the community. Such tools can become important pieces of public health infrastructure. Combined local antibiograms expand access to information about local resistance patterns and when paired with education can improve antibiotic selection.