Identification of invasive aspergillosis in electronic health records
Emily Rayens, Jessica Skela, Bradley K Ackerson, Magdalena E Pomichowski, Lance B Price, Sara Y TartofAbstract
Background
Consensus definitions for clinical identification of invasive aspergillosis (IA) are lacking, likely owed to difficulties in diagnosis. While the European Organization for Research and Treatment of Cancer-Mycoses Study Group (EORTC/MSG) has provided benchmark guidelines to define IA, the criteria are highly restrictive. In this study, we constructed alternative definitions of IA using structured data from electronic health records (EHR).
Methods
From 01 October 2015 to 31 March 2022, we identified patients with IA from EHR at Kaiser Permanente Southern California using definitions comprised of combinations of diagnosis codes, antifungal prescriptions, and/or positive mycological findings. Here, we report patient characteristics, including demographic information, underlying comorbidities and immunosuppression, diagnostic measures, as well as morbidity and mortality after diagnosis for the population comprising each IA definition.
Results
Between 86 and 4580 individuals were identified as possible IA cases across seven definitions. All IA definitions that except for that only required positive mycology had similar patient characteristics and outcomes to those meeting 2008 EORTC/MSG criteria. The largest of these cohorts was comprised of 1067 patients compared to the 86 that met EORTC/MSG criteria. Across IA definitions, all-cause mortality ranged from 11.7-33.1% within 6 months following diagnosis. The number of IA cases began trending upwards between 2019 and 2020 and continued steadily through the COVID-19 pandemic.
Conclusions
IA definitions based on structured data vastly increase efficiency and identify clinically significant IA cases that might otherwise go unrepresented. Variability in the number of patients identified between definitions also highlights critical deficits in current IA diagnostics.