Quality Indicators to Measure Appropriate Antibiotic Use: Some Thoughts on the Black Box

Dear Editor, The recent publication by Spivak et al. highlights the heterogeneity in current definitions of appropriate antimicrobial use and gives a comprehensive overview of the challenges in defining and measuring appropriateness [1]. We agree with the authors that ideally, to objectively and reproducibly measure appropriateness, measurement tools are used that facilitate large-scale data collection without the need for expert review. In their review the authors suggest to use the Centers for Disease Control (CDC) audit tools to assess appropriate antimicrobial use. These tools describe—at patient level—patient information (e.g., presence of allergy) and processes of care information (e.g., were initial antibiotics consistent with institutional/national guidelines?). The authors also propagate tools that permit comparative analysis between hospital types. To compare hospitals, hospital wards, or even professionals, it is necessary to aggregate patient level data to group level data. Unfortunately, the authors do not describe how to use the proposed CDC tools to compare entities. To do so, for the chosen entity, a synthesis of patient population information and processes of care information has to be performed, which can be a tremendous challenge. To prevent post hoc analyses on an aggregated level, we suggest to use quality indicators that assess group level appropriateness while using—objective—patient level data. These quality indicators are operationalized with extensive algorithms in which objective data directly originating from the patient record are inserted and are compared to recommended care as described in (hospital) guidelines. This assessment uses objective data from the patient record on processes of care—like antibiotics prescribed, including route and duration, cultures performed, and so forth—and takes known and recommended patient characteristics into account where relevant—like sex, allergy, co-morbidities, severity, previous antimicrobial use, and so forth—to compute an appropriateness score per patient. This score can be aggregated to a professional, ward, or hospital level by dividing the number of patients receiving hospital recommended care (numerator) by the total number of patients to whom the recommended care would apply (denominator). We believe this to be a very objective and reliable method to assess appropriateness. Albeit stringent in its interpretation, at least there will be no discussion about interrater discrepancy. Of course, there will always be a reason for nonadherence to a recommendation that is not documented in the guideline and—thus—not taken up in the algorithm. Therefore, 100% adherence can never be a goal; 80% would be great already. Some variation in adherence to quality indicators is normal and even desirable. Previously, various of such quality indicators were developed. Moreover, their psychometric properties were tested in various hospitals; most of them were applicable in daily practice. Several of these sets were published in this journal [2–4]. We suggest these quality indicators are used next to the suggested audit tools, with the ultimate aim of uniform measurement of appropriate use of antibiotics. Acknowledgments. We publish this correspondence on behalf of our co-researchers (see references): Jan M. Prins and Suzanne E. Geerlings. Potential conflicts of interest. Authors certify no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.