Racial disparities in levels of care in an emergency department

Background: The Institute of Medicine (2009) suggests disparate care is more likely to occur in patient-provider interactions where “high time pressure, cognitive complexity, and pressures for cost containment” are present. Each of these conditions is met in emergency medicine, suggesting that emergency department (ED) patients who are socially marginalized racially and/or ethnically may receive levels of care inconsistent with clinical presentation. Objective: To examine whether levels of care at an ED differ by patient race/ethnicity, controlling for diagnosis, insurance status and primary language. Methods: A secondary data analysis of ED visits (n = 101,240) was conducted using medical billing records from March 2009 through September 2010. Data were included in the central analyses if ED registration listed race-data as African-American/black (n = 15,498), Caucasian/white (n = 41,895), Hispanic of color (n = 21,890), or Hispanic white (n = 19,806). Insurance was aggregated as Medicaid, Medicare, private, or self-pay. Primary languages included were English and Spanish. 2009 and 2010 ICD-9 codes and Current Procedural Terminology (CPT) levels 99281-99291 were used to classify levels of care as “adequate” or “inadequate” per diagnosis. A level of care ≥ 5 was considered to be “adequate” for high risk diagnoses (abdominal pain, acute asthma exacerbation, chest pain, myocardial infarction and acute coronary syndrome). A level ≥ 4 was used for analysis of back pain diagnoses. Chi-square analyses examined the relationship between race and adequacy of care stratified by diagnosis, insurance and primary language. Results: Overall, white patients were significantly more likely to receive an adequate level of care than patients listed as another race/ethnicity (p < .001). This was observed in the subset of visits for abdominal pain (p < .001), persisted when controlling for primary language (p < .001), and when restricted to Medicaid (p =.025) and privately-insured (p = .003) patients. When controlling for language and insurance, white privately-insured, English-speaking patients were significantly more likely to receive an adequate level than any other group. Among visits for chest pain, Hispanic-white patients were significantly less likely to receive an adequate level of care than other groups; this effect was mitigated by controlling for primary language and insurance. Conclusion: After controlling for patient diagnosis, insurance and primary language in a diverse ED population, patients of color may receive lower levels of care than white patients for similar medical conditions.