Search for a command to run...
Introduction: To maximize the benefits of clinical research, the National Academy of Medicine (NAM) outlined a set of criteria to prioritize diagnoses for study that included prevalence, mortality, and resource use. Prioritization exists for general hospital pediatric care, but not for Pediatric ICU (PICU) care. We aim to use data from the Pediatric Health Information System (PHIS) and Virtual Pediatric System (VPS) databases to prioritize PICU diagnoses based on NAM criteria. Methods: We performed a database study of children admitted to a PICU from 1/1/2017-12/31/2022 at 31 children’s hospitals participating in PHIS and VPS. PHIS contains administrative data including demographics, comorbidities, childhood opportunity index (COI), and standardized costs (“costs”) that are a surrogate for resource use. VPS contains PICU clinical data including primary PICU diagnosis, mortality, and Pediatric Index of Mortality III (PRISM3) scores. We ranked PICU diagnoses based on prevalence, mortality, and total costs. We also ranked inter-hospital cost variability by calculating Intraclass Correlation Coefficients (ICC) for each diagnosis from mixed effects models of cost per case with a random intercept for each hospital and demographics, comorbidities, COI, and PRISM3 as fixed effects. Results: We analyzed 244,544 records with 5,121 deaths. The top 3 ranked PICU diagnoses in each criteria were: Prevalence (n, % of cases): bronchiolitis (n = 30,749, 13%), asthma (n = 17,537, 7%), and diabetic ketoacidosis (n = 13,825, 6%); Deaths (n, percent of deaths): 1) anoxic brain Injury (626 deaths, 12%), traumatic brain injury (TBI, 528 deaths, 10%), and cardiac arrest (464 deaths, 9%); Total costs: acute respiratory failure (ARF, $1,029,932,078), bronchiolitis ($915,140,772), and TBI ($639,342,573); Inter-hospital cost variation: Chiari malformations (ICC 0.72), traumatic intrabdominal organ injury (ICC 0.69), and croup (ICC 0.49). Bronchiolitis and TBI were the only diagnoses ranked in the top 3 in multiple criteria. Conclusions: We utilized a novel combined PHIS-VPS dataset to prioritize PICU diagnoses for research based on prevalence, mortality, and resource use. Based on these criteria, bronchiolitis and TBI are important PICU diagnoses and continued clinical research in these diagnoses is likely to be high impact.