P2.D067: Comparing different measures of community resources for children with out-of-hospital emergencies

Background: Community measures of social determinants of health correlate with multiple pediatric health outcomes. Three such measures have gained traction in healthcare research: the Child Opportunity Index (COI), Area Deprivation Index (ADI), and Social Vulnerability Index (SVI), yet research comparing performance of these measures is limited. We sought to correlate the COI, ADI and SVI and compare them as predictors of overall and condition-specific emergency medical services (EMS) outcomes.

Methods: We performed a retrospective study of EMS encounters using patient care records from the nationally-representative 2020-2021 ESO Data Collaborative. We included scene EMS encounters for children ( < 18 years old). We used COI, ADI, and SVI at the Census tract level. Scales were internally normalized to the dataset and aligned for uniform directionality. We calculated the agreement between the three measures using a fixed effects intraclass correlation coefficient. We evaluated the association of each index (as a splined predictor) with transport status, identification of cardiac arrest, corticosteroid and bronchodilator use in asthma, naloxone use in overdose, physical restraints or sedation in behavioral health encounters, analgesia, pain score and mental status assessment for injured children using logistic regression. We identified best fitting models with the Aikaike Information Criterion (AIC).

Results: We included 738,892 encounters (median age 10.5 years; IQR 2.9-15.5 years; 50.9% male.) The intraclass correlation between the models indicated good agreement (0.74). When visualizing splined models, there was substantial overlap in relationships between each index and the study outcomes (Figure). Models with COI had better fit for outcomes of non-transport, bronchodilators for asthma, naloxone for ingestions, analgesia for trauma, and analgesia for injured children with elevated pain scores. Models with ADI had better fit for outcomes of cardiac arrest, benzodiazepines for seizure, corticosteroids for asthma, physical restraints and sedation for behavioral health, and for pain and mental status assessments for trauma (Table). There was overlap between the odds of having each outcome of interest between models using the three measures of disparity.

Conclusion: The COI, ADI, and SVI had good correlation and demonstrated similar effect size estimates for a variety of EMS care delivery outcomes. Collectively, these findings support use of these indices for prehospital pediatric research. While the COI includes a wider range of variables specific to children, the SVI and ADI are also appropriate for pediatric research. Investigators should take consider the characteristics of the population of interest and outcomes when choosing an index.