(1334) Evaluating Disparities of a New Diagnostic Stewardship Tool for Pediatric Critical Care

Introduction: Quality improvement efforts may impact existing health disparities. We successfully reduced avoidable bacterial cultures with a new clinical decision support (CDS) tool for pediatric intensive care unit (PICU) and pediatric cardiac intensive care unit (PCICU) patients with new onset fever or instability in July 2020. Here, we evaluated for testing practice disparities across demographic groups related to this intervention.

Methods: We examined concordance of blood, urine, and respiratory cultures 12 months pre/post-CDS introduction, considering race, ethnicity, primary language, insurance type, age groups and sex . We analyzed CDS-concordance with a multivariable mixed effect logistic regression model with correlation within a patient using robust variance estimates and assessed for effect modification of the intervention.

Results: Of 2838 cultures, 920 (61%) of the pre-intervention and 957 (72%) of the post-intervention cultures were CDS-concordant. There were no differences in CDS-concordance for race, language, sex, or insurance. Patients' ethnicity labeled “declined to answer” (n=31,1.1%) had increased odds of CDS-concordant cultures compared to non-Hispanic/non-Latino patients (OR 31.2, 95%CI 2.7–360, p< 0.01). Patients >5 years to 15 years were less likely to have CDS-concordant cultures as compared to patients
There was no effect modification with the introduction of the CDS tool for race, language, ethnicity, sex or age and pre-existing differences of ethnicity and age were consistent. While there was not a significant difference in CDS-concordance in the pre-intervention (OR 1.7, 95%CI 0.36-8.32, p=0.49) and the post-intervention periods (OR 0.5, 95%CI 0.13-1.87, p=0.3), international self-pay had a lower odds of CDS-concordance than patients with private insurance in the post-intervention period compared to the pre-intervention period (OR 0.2, 95%CI 0.03–0.82, p=0.03).

Conclusions: We identified differences in CDS-concordance in school-age and non-identified ethnicity groups. Introduction of the CDS tool did not widen or improve these differences and may have reduced minor differences in CDS-concordance for international self-pay patients. Next, we will explore contributing factors for variation in CDS-concordant testing.