(282) Linguistic Isolation Correlates With Length of Stay for Pediatric Oncology Patients in California

Introduction: This is a focused investigation on the state of California, aiming to correlate social drivers of health indicators from patient ZIP codes with medical outcomes of oncology patients in pediatric intensive care units (PICUs).

Methods: We queried the Virtual Pediatric Systems (VPS) database for demographic and admission data for oncology patients in California PICUs (2009-2021). We extracted demographic, clinical, and outcome variables (excluding post-operative patients) using three-digit ZIP codes as a proxy for geographical location. Social drivers of health variables such as linguistic isolation (defined as a household in which no one age >14 speaks only English, or speaks a non-English language and speaks English “very well”), poverty, race, and educational attainment were abstracted from American Community Survey (ACS) data for each 3-digit ZIP code following the methods of Environmental Protection Agency’s (EPA's) EJScreen tool. "Other" racial group included Native Americans, Alaskan Natives, Hawaiian islanders, Pacific Islanders, and two or more races. The study compared social variables with outcomes such as length of stay (LOS), mortality, acuity scores, and type of insurance, using Pearson correlation, and mixed effects generalized regression models.

Results: There was a positive correlation between mortality and minority racial groups (all non-white individuals, specifically Hispanic/Latino) across ZIP codes (2017: p=0.0207, 2020: p=0.0425, 2021: p=0.013), while the "other" racial group showed an inverse correlation (2017: p=0.0025, 2020: p=0.0034, 2021: p=0.02958). Median PICU LOS was positively correlated with linguistic isolation (2108: p=0.05; 2019: p< 0.001; 2020: p< 0.01; 2021: p< 0.001), while controlling for PRISIM III scores (n=7,417). Regression model showed a longer medical PICU LOS for Black or African American patients with a 15.8% increase and Hispanic or Latino patients with a 12.4% increase compared to "white" patients when controlling for gender, percentage linguistic isolation, and year. Additionally, there was a 0.784% increase in medical PICU LOS for each 1-unit increase in percentage linguistic isolation.

Conclusions: Our results demonstrate that minority groups are a heterogenous population and clinical outcomes may vary by racial group.