326 - Race, ethnicity, and language data collection practices in children's hospitals

Poster Number: 326
Publication Number: 326.1749

Background: Hospital databases such as the Pediatric Health Information System (PHIS) provide opportunities to design and evaluate large-scale multi-site interventions that address healthcare inequities. Fulfilling this potential depends on the standardization, accuracy, and completeness of demographic data, such as race, ethnicity, and language preference (REaL). However, little is known about variation in collection and reporting of REaL data across hospitals that submit data to PHIS.

Objective: To describe REaL data collection and reporting practices among PHIS hospitals

Design/Methods: We surveyed leaders at 48 PHIS hospitals in summer 2023 about processes, challenges, and opportunities related to 1) REaL data collection and verification in inpatient and emergency department settings, and 2) data transformation for PHIS reporting. The 44-item electronic survey was informed and piloted by focus groups of experts in quality improvement; informatics; and diversity, equity, and inclusion with input from health services researchers and a family representative. Survey recipients were encouraged to engage experts in their hospitals to ensure accurate and complete responses. We analyzed data using descriptive statistics and thematic review of narrative responses.

Results: Forty-one of 48 hospitals (85%) completed the survey. While some practices were consistent across most hospitals (e.g., directly asking parent/caregiver for patient race and ethnicity [93%]), many varied widely for race and ethnicity (Table 1) and language preference (Table 2). Treatment of data when a patient had multiple races and/or ethnicities was especially variable, ranging from the ability to record each identity to only "multiracial"/"multiethnic" to no multiracial/ethnic option at all. This affected how accurately data could be transformed and reported to PHIS. Language preferences also were collected inconsistently, including whether data represented the patient’s (71%) or caregiver’s (63%) preferred language, and whether data reflected spoken (76%), written (49%), or unspecified (24%) language. A small number of hospitals asked teens < 18 to self-report race and ethnicity (20%) or language preference (41%). Less than a third of hospitals regularly asked families to verify the accuracy of REaL data at each encounter.

Conclusion(s): Wide variation in the collection of REaL data among children's hospitals creates challenges for the use of multi-site databases to address healthcare inequities. Our findings highlight the urgent need to develop and implement standards and best-practices for REaL data collection, quality management, and reporting.