Racial and Ethnic Disparities in Long-Acting Injectable Antipsychotic Use in a National Sample of Medicare Beneficiaries With Schizophrenia

Abstract

This cross-sectional study assesses the racial and ethnic disparities in long-acting injectable antipsychotic use in a national sample of Medicare beneficiaries with schizophrenia.

Introduction

Despite its importance in preventing relapse and hospitalization, adherence to antipsychotic medications is a challenge for many individuals with schizophrenia.¹ Long-acting injectable antipsychotics (LAIs) offer a major advantage over traditional, once-daily oral medications given their less-frequent dosing intervals.² While first-generation LAIs are typically administered monthly, second-generation LAIs offer even longer dosing intervals. The American Psychiatric Association’s clinical guidelines have also highlighted numerous other advantages of LAIs over oral antipsychotics.² Previous studies have also shown LAI use, especially second-generation LAI use, to be associated with better outcomes relative to oral antipsychotics.^2,3,4 Despite renewed interest in addressing health inequity in the US, limited research exists on racial disparities in LAI use and none on Medicare, which insures half of US patients with schizophrenia.⁵

Methods

In this cross-sectional study, we used national 2018-2019 Medicare claims to identify all continuously enrolled fee-for-service Medicare Part D beneficiaries with schizophrenia who filled 1 or more prescriptions for antipsychotic medication (eMethods in Supplement 1). The outcomes were (1) any LAI use among all antipsychotic users and (2) any second-generation LAI use among all LAI users. Logistic regressions controlling for sociodemographic, clinical, and Part D plan policy factors were used to estimate adjusted differences in outcomes by the race and ethnicity designations available in Medicare data (eMethods in Supplement 1). Sensitivity analysis was conducted using the Research Triangle Institute race and ethnicity code variable. Analyses were repeated among 2 subgroups: (1) those with a greater need for LAIs (ie, poor antipsychotic adherence, schizophrenia-related hospitalization, and/or schizophrenia-related emergency department visit in 2018) and (2) those with minimal or no out-of-pocket costs for LAIs (ie, eligible for full Part D low-income subsidy [LIS]). This study was deemed exempt from review and informed consent by the University of Pennsylvania Institutional Review Board because it constituted secondary data research and posed minimal risk to individuals included in the data set. We followed the STROBE reporting guideline. Data were analyzed using SAS Enterprise, version 7.15 HF9 (SAS Institute Inc).

Results

The final sample included 210 686 beneficiaries (mean [SD] age, 56.1 [14.1] years; 28.7% were ≥65 years; 60.6% were male; 66.4% were White; 90.9% were eligible for full LIS). Overall, 22.6% beneficiaries used an LAI; two-thirds (66.4%) of the LAI users received a second-generation LAI. Prior to adjustment, large racial and ethnic differences were observed in the rate of LAI use (Table 1) and second-generation LAI use (Table 2). After adjustment, the differences narrowed substantially in any LAI use (23.5% Black, 23.0% Hispanic, 22.9% other, and 22.2% White). However, among LAI users, Black patients (60.1%) continued to have a substantially lower rate of second-generation LAI use than White patients (69.9%) (adjusted odds ratio [AOR], 0.63; 95% CI, 0.60-0.66). Hispanic patients (64.9%) also showed a lower rate of second-generation LAI use relative to White patients (69.9%) (AOR, 0.79; 95% CI, 0.71-0.87). Similar findings were observed in the subgroup and sensitivity analyses.

Table 1. Rates of Any LAI Use Among All Fee-for-Service Medicare Beneficiaries With Schizophrenia Using Antipsychotics in 2019.

Race and ethnicity category	No. of antipsychotic usersa	Rate of any LAI use		AOR (95% CI)c
		Unadjusted, No. (%)b	Adjusted, %c
All antipsychotic users (main sample, n = 210 686)
Black	48 965	15 021 (30.7)	23.5	1.38 (1.31-1.46)
Hispanic	9340	2424 (26.0)	23.0	1.22 (1.10-1.35)
White	139 872	26 990 (19.3)	22.2	1.00 [Reference]
Otherd	12 509	3232 (25.8)	22.9	1.19 (1.09-1.30)
Subgroup of antipsychotic users with full Part D LIS status in 2019 (n = 191 409)
Black	46 245	14 462 (31.3)	24.6	1.38 (1.31-1.46)
Hispanic	9019	2379 (26.4)	24.1	1.23 (1.10-1.36)
White	124 678	25 379 (20.4)	23.2	1.00 [Reference]
Otherd	11 467	3092 (27.0)	24.0	1.20 (1.09-1.32)
Subgroup of antipsychotic users with history of poor antipsychotic adherence, schizophrenia-related hospitalization, and/or schizophrenia-related ED visits in 2018 (n = 77 665)
Black	21 216	7325 (34.5)	29.0	1.31 (1.23-1.40)
Hispanic	3737	1131 (30.3)	28.1	1.16 (1.02-1.31)
White	48 274	11 662 (24.2)	27.1	1.00 [Reference]
Otherd	4438	1450 (32.7)	28.3	1.2 (1.07-1.34)
Sensitivity analysis using RTI race and ethnicity variable in main sample (n = 210 686) e
Black	48 439	14 901 (30.8)	23.5	1.39 (1.32-1.46)
Hispanic	17 085	3949 (23.1)	22.7	1.13 (1.05-1.23)
White	133 410	25 810 (19.3)	22.2	1.00 [Reference]
Otherd	11 752	3007 (25.6)	22.8	1.17 (1.06-1.28)

^a The number of antipsychotic users serves as the denominator for the calculation of the rates of any long-acting injectable antipsychotic (LAI) use for each race and ethnicity category.

^b Unadjusted rates were observed rates of the number of users of any LAI divided by the number of antipsychotic users in each race and ethnicity category.

^c Adjusted rates and adjusted odds ratios (AORs) estimated from logistic regressions controlling for sociodemographic factors (age, sex, original reason for Medicare eligibility, census region, metropolitan status of residence, and low-income subsidy [LIS] status), clinical factors in previous year (antipsychotic use and adherence, number of Elixhauser comorbidities, any schizophrenia-related hospitalization, and any schizophrenia-related emergency department [ED] visit if no evidence of schizophrenia-related hospitalization), and Part D plan policy factors (mean percentage of all available LAIs on formulary and mean percentage of LAIs on formulary requiring prior authorization).

^d Included Asian, North American Native, and Other unspecified races.

^e Medicare’s original data on race and ethnicity comes from Social Security Administration records. The Research Triangle Institute (RTI) race and ethnicity code variable enhances this original race and ethnicity designation in the Medicare data using first and last name algorithms to identify more beneficiaries as Hispanic or Asian.

Table 2. Rates of Any Second-Generation LAI Use Among All Fee-for-Service Medicare Beneficiaries With Schizophrenia Using LAIs in 2019.

Race and ethnicity category	No. of LAI usersa	Rate of any second-generation LAI use		AOR (95% CI)c
		Unadjusted, No. (%)b	Adjusted, %c
All LAI users (n = 47 667)
Black	15 021	8906 (59.3)	60.1	0.63 (0.60-0.66)
Hispanic	2424	1720 (71.0)	64.9	0.79 (0.71-0.87)
White	26 990	18 664 (69.2)	69.9	1.00 [Reference]
Otherd	3232	2368 (73.3)	68.2	0.92 (0.85-1.01)
Subgroup of LAI users with full Part D LIS status in 2019 (n = 45 312)
Black	14 462	8706 (60.2)	61.3	0.63 (0.61-0.66)
Hispanic	2379	1699 (71.4)	66.2	0.79 (0.72-0.88)
White	25 379	17 874 (70.4)	70.9	1.00 [Reference]
Otherd	3092	2285 (73.9)	69.0	0.91 (0.83-1.00)
Subgroup of LAI users with history of poor antipsychotic adherence, schizophrenia-related hospitalization, and/or schizophrenia-related ED visits in 2018 (n = 21 568)
Black	7325	4230 (57.7)	58.7	0.68 (0.63-0.72)
Hispanic	1131	765 (67.6)	61.0	0.75 (0.65-0.86)
White	11 662	7739 (66.4)	67.1	1.00 [Reference]
Otherd	1450	1018 (70.2)	64.8	0.9 (0.79-1.02)
Sensitivity analysis using RTI race and ethnicity variable in all LAI users (n = 47 667) e
Black	14 901	8824 (59.2)	60.0	0.63 (0.60-0.66)
Hispanic	3949	2779 (70.4)	66.3	0.84 (0.78-0.91)
White	25 810	17 837 (69.1)	69.9	1.00 [Reference]
Otherd	3007	2218 (73.8)	68.8	0.95 (0.87-1.04)

^a The number of long-acting injectable antipsychotic (LAI) users serves as the denominator for the calculation of the rates of any second-generation LAI use for each race and ethnicity category.

^b Unadjusted rates are observed rates of the number of users of any second-generation LAI divided by the number of any LAI users in each race and ethnicity category.

^c Adjusted rates and adjusted odds ratios (AORs) estimated from logistic regressions controlling for sociodemographic factors (age, sex, original reason for Medicare eligibility, census region, metropolitan status of residence, and low-income subsidy [LIS] status), clinical factors in previous year (antipsychotic use and adherence, number of Elixhauser comorbidities, any schizophrenia-related hospitalization, and any schizophrenia-related emergency department [ED] visit if no evidence of schizophrenia-related hospitalization), and Part D plan policy factors (mean percentage of all available LAIs on formulary and mean percentage of LAIs on formulary requiring prior authorization).

^d Included Asian, North American Native, and other unspecified races.

^e Medicare’s original data on race and ethnicity comes from Social Security Administration records. The Research Triangle Institute (RTI) race and ethnicity code variable enhances this original race and ethnicity designation in the Medicare data using first and last name algorithms to identify more beneficiaries as Hispanic or Asian.

Discussion

In this national study of Medicare beneficiaries with schizophrenia, statistically significant albeit small racial and ethnic differences were observed in any LAI use after controlling for sociodemographic, clinical, and plan factors. However, among patients using an LAI, Black and Hispanic patients were substantially less likely to use second-generation LAIs (and more likely to use first-generation LAIs) relative to White patients. Similar disparities in second-generation LAI use existed even among subgroups wherein patients had a potential need for LAIs or faced lower out-of-pocket cost barriers for second-generation LAIs. Future research should examine how these disparities in use translate into clinical and economic outcomes. Our study was limited to fee-for-service Medicare beneficiaries, and results may not generalize to Medicare Advantage beneficiaries. Given that several second-generation LAIs are now available with longer dosing intervals (every 2, 3, or 6 months) that may facilitate improved antipsychotic adherence, our study highlights the need for further research to examine reasons for these disparities and develop appropriate interventions.

Supplement 1.

eMethods.

Supplement 2.

Data Sharing Statement