Background: Patients with limited English proficiency (LEP) experience inferior health outcomes. The end-of-life period represents a particularly vulnerable time for LEP patients who may not receive optimal symptom management due to language barriers and other factors. There is limited data on disparities in provider practices around opioid administration based on patient LEP status, particularly at end-of-life and among inpatients on a general medicine service. This study seeks to identify differences in quantities of opioids given to LEP patients as compared to those with English proficiency (EP) while receiving end-of-life care on a hospital medicine service.

Methods: This is an EHR-based retrospective cohort study identifying all adult patients discharged from the hospital medicine service at a single academic teaching hospital from January 2013 to September 2021 (n = 61,836). Patients who had an active “Comfort Measures” order set on discharge were considered to be receiving end-of-life care and were included in our study. Data was collected on patient demographics, diagnosis codes, and medications. LEP status was defined as having a designated primary language other than English and requiring an interpreter during admission. For each patient, all opioid medications received during admission were converted to oral morphine equivalents (OME) and averaged over the entire length of stay to create a daily average OME (mg/day). A multivariable linear regression was used to examine differences in average daily OME for LEP vs EP patients. The model was adjusted for patient demographics (age, gender, race/ethnicity), hospitalization factors (intensive care unit stay, palliative care consult, teaching vs direct care hospitalist service, year of admission), medical diagnoses (cancer pain diagnosis, adjusted Elixhauser comorbidity index, history of substance use disorder), whether patients were prescribed opioids prior to admission, and the interaction between LEP status and race/ethnicity. Year of admission was included to control for temporal trends in prescribing.

Results: We identified 2652 patients who received end-of-life care. Of these patients, 701 had limited English proficiency and 1951 were considered English proficient (Table 1). In unadjusted analysis, LEP patients received 176.8 mg fewer OME per day (p= 0.03). Adjusting for covariates, LEP status was no longer significant; instead, the difference in daily OME was driven by age and cancer-related pain (Table 2). For every increased year in age, patients received, on average, 12.7 mg fewer OME per day (CI -17.8 to -7.5, p< 0.001). Patients with a cancer pain diagnosis, on average, received 298.8 mg more OME per day (CI 70.6 to 526.9, p=0.01). Additionally, patients identifying as Native Hawaiian or Other Pacific Islander received 798.6 mg more OME per day (CI 99.0 to 1498.3, p=0.03, n=54). The LEP subgroup was significantly older and with a smaller proportion of cancer pain diagnoses compared to the EP subgroup, likely accounting for the differences seen in the unadjusted analysis.

Conclusions: Our study found no significant difference in average daily OME administered to LEP patients receiving end-of-life care on a hospital medicine service as compared to EP patients in adjusted analysis. Older patients received significantly fewer opioids while patients with cancer-related pain received significantly more.

IMAGE 1: Table 1. Baseline Patient Characteristics

IMAGE 2: Table 2. Unadjusted and Multivariable Linear Regression Results