P4A223: Validation of Pediatric Patients Diagnosed with In-hospital Stroke

Background: Although the incidence of pediatric stroke is low, it is associated with significant morbidity and mortality. Little is known about the characteristics and outcomes of pediatric patients diagnosed with in-hospital stroke, and there is no established methodology to identify in-hospital stroke in the International Pediatric Stroke Study (IPSS), an international multisite database. The objective of this study was to validate a methodology to identify patients diagnosed with in-hospital arterial ischemic stroke (AIS) in the IPSS database.

Methods: Utilizing the IPSS database, non-neonatal patients (greater than 29 days old to less than 19 years old) diagnosed with AIS from January 2010 to December 2019 were included. An expert panel was convened to develop eligibility criteria that utilized IPSS variables and/or combinations of variables to identify patients diagnosed with in-hospital AIS. Seven eligibility criteria were identified, and 2 of the criteria were further broken down into 5 subsets based on timing of events linked to chosen variables, leading to a total of 15 eligibility criteria (table 1). To verify the ability of these eligibility criteria to identify in-hospital stroke, a convenience sample from this population was obtained based on center and investigator availability. Eleven volunteer institutions performed a chart review to confirm if identified patients were diagnosed with in-hospital versus out-of-hospital stroke. Descriptive statistics were utilized, and positive predictive values (PPVs) of in-hospital stroke were calculated.

Results: 542 patients diagnosed with potential in-hospital stroke were identified by meeting at least one of the eligibility criteria. 13 patients were initially excluded due to data entry errors. The number of patients identified by each eligibility criteria varied from 4 to 183, with four criteria each identifying 23-33% of the patients. For the convenience sample, 239 patients had chart reviews performed and an additional 17 patients were excluded due to data entry error or lack of medical record access. For each eligibility criteria, PPVs ranged from 31.9-100% (table 2). Utilizing a cutoff PPV of 75% or greater, 11 eligibility criteria were included. This cutoff identified 156 patients of the convenience sample (70.3%) and the combined PPV of in-hospital stroke was 77.6%.

Conclusion: This methodology identifies a large cohort of patients in the IPSS database with in-hospital stroke. A chart review was performed on a sample of ~45% of the identified patients and found a range of PPVs from 31.9-100% for each eligibility criteria. The next steps include utilizing the combination of eligibility criteria based on the cutoff PPV of 75% to identify and characterize the pediatric patients diagnosed with in-hospital stroke within the IPSS database.