Specialty drug use for autoimmune conditions varies by race and wage among employees with employer-sponsored health insurance

Abstract

BACKGROUND:

The relationship between race and ethnicity, wage status, and specialty medication (SpRx) use among employees with autoimmune conditions (AICs) is poorly understood. Insight into sociodemographic variations in use of these medications can inform health equity improvement efforts.

OBJECTIVE:

To assess the association of race and ethnicity and wage status on SpRx use and adherence patterns among employees with AICs enrolled in employer-sponsored health insurance.

METHODS:

In this observational, retrospective cohort analysis, data were obtained from the IBM Watson MarketScan database for calendar year 2018. Employees were separated into race and ethnicity subgroups based on employer-provided data. Midyear employee wage data were used to allocate employees into the following annual income quartiles: $47,000 or less, $47,001-$71,000, $71,001-$106,000, and $106,001 or more. The lowest quartile was further divided into 2 groups ($35,000 or less and $35,001-$47,000) to better evaluate subgroup differences. Outcomes included monthly days SpRx-AIC supply, proportion of days covered (PDC), and medication discontinuation rates. Generalized linear regressions were used to assess differences while adjusting for patient and other characteristics.

RESULTS:

From a sample of more than 2,000,000 enrollees, race and ethnicity data were available for 617,117 (29.8%). Of those, 47,839 (7.8%) were identified as having an AIC of interest, with prevalence rates of AICs differing by race within wage categories. Among those with AICs, 5,358 (11.2%) had filled at least 1 SpRx-AIC prescription. Following adjustment, except for the highest wage category, prevalence of SpRx-AIC use was significantly less among Black and Hispanic subpopulations. Black patients had significantly lower SpRx-AIC use rates than White patients (≤$35,000: 4.9 vs 9.4%, >$35,000-$47,000: 5.5 vs 10.6%, >$47,000-$71,000: 8.5 vs 11.1%, and >$71,000-$106,000: 9.1 vs 12.7%; P <0.001 for all). For Hispanic patients, prevalence rates were significantly lower than White patients in 3 different wage categories (≤$35,000: 4.5 vs 9.4%, >$35,000-$47,000: 6.1 vs 10.6%, and >$71,000-$106,000: 8.6 vs 12.7%; P < 0.001). PDC and 90-day discontinuation rates did not differ among race and ethnicity groups within the respective wage bands.

CONCLUSIONS:

Race and ethnicity and wage-related disparities exist in SpRx use, but not PDC or discontinuation rates for treatment of AICs among non-White and low-income populations with employer-sponsored insurance, and may adversely impact clinical outcomes.

Plain language summary

Biologic drugs are used to treat diseases like rheumatoid arthritis and psoriasis. Non-White and/or low-wage workers with these diseases may be less likely to take these medications. As a result, these individuals may have more symptoms and their diseases may get worse. More needs to be done to make sure that all individuals have access to these medications.

Implications for managed care pharmacy

Managed care pharmacy practitioners are uniquely positioned to identify and take steps to mitigate observed disparities in specialty biologic medication use for treatment of autoimmune conditions among non-White and low-wage workers. Doing so has the potential to yield more equitable health outcomes in these subpopulations.

Researchers, policymakers, health care providers, and employers have been concerned about existing inequities in health care for several decades.¹ The recent COVID-19 pandemic has exacerbated that concern, impacting disadvantaged communities and populations to a greater extent.² Among adults with employer-sponsored health insurance, analyses have further quantified disparities in disease burden and management effectiveness on the basis of race and ethnicity.³ Others have evaluated the relationship between income and health care utilization, in which low-wage workers exhibit significantly greater use of the emergency department and avoidable hospitalizations, as well as reduced compliance with preventive care services.⁴ These insights have fueled efforts to identify and correct health care disparities that result from systemic inequities.⁵

More recent analyses have examined the intersectionality between race and ethnicity and sociodemographic factors.^6,7 These studies illustrate how race and ethnicity can moderate the effect of sociodemographic variables on health outcomes and have helped to provide greater clarity in focusing opportunities to mitigate existing health inequities. Said differently, because individuals with multiple contributors to health inequities often experience worse outcomes than those with a single contributor,⁶ therefore, a deeper understanding of intersectionality can help to target subsequent interventions toward those with the greatest opportunity for potential benefit.

Relative to other health care service types, specialty pharmaceuticals (SpRx) have been responsible for substantial disproportionate annual cost growth for employers during the past 12 years.⁸ Employer efforts to manage SpRx costs by implementing increased cost-sharing from enrollees have contributed to disparities in use of these medications, with a lower prevalence of SpRx use among low-wage earners with autoimmune conditions (AICs).⁹ However, race and ethnicity were not included in this analysis, limiting their value in understanding the intersectionality between wage and race and ethnicity in the setting of SpRx use.

With growing employer interest in health equity, a deeper understanding of the association of both wage and race and ethnicity with SpRx can inform targeted efforts to achieve more equitable condition management outcomes. Accordingly, the goal of the current study was to evaluate patterns of SpRx use by race and ethnicity and income for individuals with autoimmune disorders and enrolled in employer-sponsored health insurance.

Methods

This study examined the specialty drug use patterns among 2,071,980 active employees who were enrolled in selfinsured employer-sponsored health insurance coverage for the entire 2018 calendar year, and for whom employer-reported wage data, as well as race and ethnicity data, were available. The research dataset was derived from the IBM Watson Health MarketScan Database, which consists of deidentified outpatient, inpatient, and pharmaceutical claims of approximately 40-50 million privately insured patients each year. These claims originate from more than 150 large employers offering self-insured health insurance coverage for eligible enrollees residing in all 50 states. The database includes patient demographics, benefits eligibility, health benefits design information, and medical and pharmacy claims data. Claims data include actual (reimbursed) payment amounts and health plan and employee spending. The analysis met the required provisions of the Health Insurance and Portability and Accountability Act of 1996 that all data specific to individual patients be nonidentifiable to protect health information. Institutional review board review was not sought.

The study population (Supplementary Table 1 ^{(372KB, pdf)} , available in online article) included individuals with 1 of the following conditions often treated with a specialty biologic medication: rheumatoid arthritis, atopic dermatitis, multiple sclerosis, psoriasis (plaque psoriasis and psoriatic arthritis), Crohn’s disease, and asthma. International Classification of Diseases, Tenth Revision principal diagnosis codes (Supplementary Table 2 ^{(372KB, pdf)} ) from the medical claims data were used to identify these subpopulations, which were then aggregated into a single population defined as having at least 1 of these conditions. The identified conditions were selected for inclusion based on the fact that SpRx medications are often used in treatment, and if prescribed, are used chronically, and not typically in an intermittent or limited manner, such as for treatment of cancers or hepatitis C, respectively.

Biologic specialty medications used to treat any of the identified AICs were identified using the US Food and Drug Administration Purple Book.¹⁰ The full list of SpRx-AIC included in the analysis is shown in Supplementary Table 3 ^{(372KB, pdf)} . More than 90% of SpRx-AICs were available through the pharmacy plan relative to the medical plan; therefore, we excluded SpRx-AIC billed through the medical plan from our analysis.

RACE AND ETHNICITY COHORT, WAGE COHORT, AND CLINICAL DATA DEFINITIONS

Race and ethnicity subgroups were based on employer-provided data and included enrollee categorization in a single data field as White, Black, Hispanic, American Indian/Alaskan Native, Asian-Pacific Islander, Native Hawaiian-Other Pacific, or 2 or more races. Because of small numbers in the latter 4 categories, these were combined in the analysis into a single (Other) subgroup.

Midyear employee wage data included in MarketScan were used to allocate employees into annual income quartiles: $47,000 or less, $47,001-$71,000, $71,001-$106,000, and $106,001 or more, as previously described.⁹ The lowest quartile was further divided into 2 groups ($35,000 or less and $35,001-$47,000) to enable better understanding of differences in health care use at lower-wage levels. Our prior research has demonstrated the use of this approach in characterizing significant differences in employee behaviors without the need for more granular wage categories.⁴ Our intent in doing so was to characterize population subgroups on the basis of proportions of workers rather than using more traditional wage bands, which effectively consolidate low-wage workers into larger categories, while separating higher-earning workers into a series of smaller subpopulations. We examined the association of wage category and race and ethnicity with use of SpRx-AIC for treatment in aggregate for all AICs.

USE AND OUTCOMES MEASURES

Use measures included adherence to specialty medications (using a proportion of days covered [PDC] metric) and specialty medication discontinuation rate (defined as a period of 90 or more days without specialty medication). Outcomes included the likelihood of treatment with a specialty medication and number of days supplied per patient. Full definitions are included in Supplementary Table 4B ^{(372KB, pdf)} . Employee filling of a SpRx-AIC prescription was a primary outcome measure and was based on the presence of any pharmacy claims for condition-specific medications during the study period.

STATISTICAL ANALYSIS

Regression analysis was performed for SpRx-AIC adherence and discontinuation by wage group and race and ethnicity indicators, and covariates: age and sex categories, ZIP code–based median household income,¹¹ geographic census region, health plan contract type (individual, individual plus spouse, individual plus child, individual plus children, or family), net deductible as a percentage of annual wages, comorbid condition prevalence (Charlson Comorbidity Index¹² and Psychiatric Diagnostic Groupings¹³), an indicator for being a salaried employee, an indicator for being part of a union, and an indicator for living in a rural area (full definitions of explanatory variables included in Supplementary Table 4A ^{(372KB, pdf)} ).

Continuous outcomes (specialty medication prevalence of use, adherence, and discontinuation) were modeled using generalized linear models with a Gaussian family, and binary outcomes were modeled using generalized linear models with a logit link. In the descriptive and multivariate analyses, we compare the race and ethnicity groups within each of the respective wage bands, with values for the White subgroup used as the comparator, because of it being the largest subpopulation. All comparisons were based on α = 0.0125 (the Bonferroni correction for multiple comparisons, a conservative approach). This methodology was adapted from a prior similar claims data analysis.⁹

Results

EMPLOYEE CHARACTERISTICS

Of the initial dataset containing more than 2 million employees, race and ethnicity data were available for 617,117 (29.8%). Of those, 47,839 (7.8%) were identified as having an autoimmune disorder of interest. Table 1 shows the demographic distribution of these individuals, including race, ethnicity, and wage. A more detailed distribution of the demographic variables incorporated into the analysis is provided in Supplementary Table 5 ^{(372KB, pdf)} . Employees in the lower-wage categories were more likely to be non-White, younger, female, and hourly workers in comparison with employees in higher-wage categories. Employees in the lowest wage category were more than twice as likely to be enrolled in an employee-only benefit plan compared with employees in higher-wage groups and were more than 5 times more likely to pay deductibles as a greater proportion of wages (1.6% of wages or greater) than employees in the highest wage group. In addition, employees in all lower-wage categories had significantly higher scores on both the Charlson Comorbidity Index and Psychiatric Diagnostic Groups than employees in the highest wage group, indicating a higher disease burden of both chronic physical and psychiatric illnesses.

TABLE 1.

Proportion of Employees With Autoimmune Diagnosis Using Specialty Medication by Annualized Wage Category (Adjusted and Unadjusted)

Annualized wage band	Black	Hispanic	All othera	White	Total
Employees, N	556	351	345	4,106	5,358
Wage category ≤$35,000
N	67	56	32	301	456
Unadjusted mean	5.34	4.99	8.84	10.56	8.86
Predicted mean (95% CI)	4.9b (3.6-6.7)	4.5b (3.2-6.4)	7.8 (5.0-12.0)	9.4 (8.1-11.0)
Wage category $35,001-$47,000
N	80	47	31	419	577
Unadjusted mean	5.97	6.69	7.91	11.32	10.07
Predicted mean (95% CI)	5.5b (4.1-7.3)	6.1b (4.2-8.7)	7.4 (4.7-11.4)	10.6 (9.3-12.0)
Wage category $47,001-$71,000
N	179	109	70	1,108	1,466
Unadjusted mean	8.86	8.70	7.75	12.14	11.18
Predicted mean (95% CI)	8.5b (7.0-10.2)	8.9 (7.0-11.2)	7.6b (5.7-10.2)	11.1 (10.2-12.0)	—
Wage category $71,001-$106,000
N	149	87	89	1,251	1,576
Unadjusted mean	9.21	8.55	8.86	13.37	12.47
Predicted mean (95% CI)	9.1b (7.5-11.1)	8.6b (6.6-11.1)	8.9b (6.8-11.4)	12.7 (11.8-13.6)
Wage category ≥$106,001
N	81	52	123	1,027	1,283
Unadjusted mean	12.54	11.40	12.07	13.41	11.49
Predicted mean (95% CI)	11.9 (9.0-15.5)	11.0 (7.8-15.2)	11.5 (9.2-14.3)	12.7 (11.7-13.7)

^aIncludes American Indian/Alaskan Native, Asian-Pacific Islander, Native Hawaiian-Other Pacific, or individuals reporting 2 or more races.

^bP value less than 0.001, relative to the White population in the corresponding wage category, adjusted for demographic and health variables using generalized linear models.

Of this population subset with an autoimmune disorder, 5,358 (11.2%) had at least 1 claim for a SpRx-AIC. Table 1 shows the proportion of individuals by race and ethnicity with an AIC and use of SpRx-AIC by wage category following adjustment for demographic and other variables. Notably, the prevalence rates of the combined AICs differ by race and ethnicity within wage categories, with Black patients having significantly lower rates compared with White patients in all but the top wage category (≤$35,000: 4.9 vs 9.4%, >$35,000-$47,000: 5.5 vs 10.6%, >$47,000-$71,000: 8.5 vs 11.1%, and >$71,000-$106,000: 9.1 vs 12.7%; P < 0.001 for all). For Hispanic patients, prevalence rates were significantly lower than White patients in 3 different wage categories (≤$35,000: 4.5 vs 9.4%, >$35,000-$47,000: 6.1 vs 10.6%, and >$71,000-$106,000: 8.6 vs 12.7%; P < 0.001). For individuals in the Other category, prevalence rates of SpRx-AIC use were significantly lower in the 2 middle wage quartiles (>$47,000-$71,000: 7.6 vs 11.1%, P = 0.002; and >$71,000-$106,000: 7.6 vs 12.7%; P = 0.001).

Study results examining the use of SpRx-AIC indicate that following adjustment, except for the highest wage category, medication use was significantly less among Black and Hispanic patients (Table 1). Black patients had significantly lower SpRx-AIC use rates than White patients (≤$35,000: 4.9 vs 9.4%, >$35,000-$47,000: 5.5 vs 10.6%, >$47,000-$71,000: 8.5 vs 11.1%, and >$71,000-$106,000: 9.1 vs 12.7%; P < 0.001 for all). For Hispanic patients, prevalence rates were significantly lower than White patients in 3 different wage categories (≤$35,000: 4.5 vs 9.4%, >$35,000-$47,000: 6.1 vs 10.6%, and >$71,000-$106,000: 8.6 vs 12.7%; P < 0.001).

PATTERNS OF MEDICATION USE

Table 2 details the SpRx-AIC PDC by race and ethnicity and income following adjustment for demographic and health variables. Although PDC was generally lower among non-White subpopulations in relation to White populations, no significant differences were observed within any of the different wage categories.

TABLE 2.

Specialty Medication Percentage of Days Covered by Race and Ethnicity and Annualized Wage Category (Adjusted and Unadjusted)

	Black	Hispanic	All othera	White	Total
Employees, N	556	351	345	4,106	5,358
Wage category ≤$35,000
N	67	56	32	301	456
Unadjusted mean	0.73	0.72	0.78	0.75	0.75
Predicted mean (95% CI)	0.74 (0.67-0.81)	0.74 (0.66-0.82)	0.78 (0.67-0.88)	0.75 (0.75-0.81)
Wage category $35,001-$47,000
N	80	47	31	419	577
Unadjusted mean	0.70	0.74	0.77	0.76	0.75
Predicted mean (95% CI)	0.70 (0.63-0.77)	0.76 (0.68-0.85)	0.79 (0.68-0.89)	0.76 (0.73-0.79)
Wage category $47,001-$71,000
N	179	109	70	1,108	1,466
Unadjusted mean	0.71	0.67	0.72	0.76	0.75
Predicted mean (95% CI)	0.71 (0.67-0.76)	0.67 (0.61-0.72)	0.72 (0.65-0.79)	0.76 (0.74-0.78)
Wage category $71,001-$106,000
N	149	87	89	1,251	1,576
Unadjusted mean	0.69	0.72	0.70	0.77	0.76
Predicted mean (95% CI)	0.70 (0.65-0.75)	0.73 (0.67-0.79)	0.70 (0.64-0.77)	0.77 (0.75-0.78)
Wage category ≥$106,001
N	81	52	123	1,027	1,283
Unadjusted mean	0.74	0.74	0.75	0.77	0.76
Predicted mean (95% CI)	0.74 (0.68-0.81)	0.75 (0.67-0.83)	.75 (0.70-0.80)	0.77 (0.75-0.79)

^aAmerican Indian/Alaskan Native, Asian-Pacific Islander, Native Hawaiian-Other Pacific, or 2 or more races.

Table 3 provides a similar analysis of a 90-day discontinuation rate following adjustment for demographic and health variables. Like the PDC results, non-White subpopulations tended to have greater discontinuation rates, but the adjusted differences were not significant within any of the different wage categories.

TABLE 3.

90-Day or Greater Discontinuation of Specialty Medications by Race and Ethnicity and Wage Category (Adjusted and Unadjusted)

	Black	Hispanic	All othera	White	Total
Employees, N	556	351	345	4,106	5,358
Wage category ≤$35,000
N	67	56	32	301	456
Unadjusted mean	0.30	0.36	0.32	0.32	0.32
Predicted mean (95% CI)	0.26 (0.67-0.81)	0.27 (0.66-0.82)	0.26 (0.67-0.88)	0.28 (0.71-0.78)
Wage category $35,001-$47,000
N	80	47	31	419	577
Unadjusted mean	0.35	0.34	0.26	0.29	0.30
Predicted mean (95% CI)	0.29 (0.00-1.00)	0.27 (0.00-1.00)	0.18 (0.00-1.00)	0.26 (0.00-1.00)
Wage category $47,001-$71,000
N	179	109	70	1,108	1,466
Unadjusted mean	0.34	0.42	0.34	0.30	0.32
Predicted mean (95% CI)	0.34 (0.25-0.43)	0.42 (0.31-0.55)	0.34 (0.21-0.50)	0.30 (0.26-0.34)
Wage category $71,001-$106,000
N	149	87	89	1,251	1,576
Unadjusted mean	0.40	0.33	0.31	0.28	0.30
Predicted mean (95% CI)	0.37 (0.27-0.48)	0.31 (0.20-0.45)	0.34 (0.22-0.48)	0.28 (0.25-0.32)
Wage category ≥$106,001
N	81	52	123	1,027	1,283
Unadjusted mean	0.35	0.29	0.32	0.27	0.28
Predicted mean (95% CI)	0.34 (0.21-0.49)	0.27 (0.14-0.44)	0.30 (0.21-0.42)	0.27 (0.24-0.31)

^aAmerican Indian/Alaskan Native, Asian-Pacific Islander, Native Hawaiian-Other Pacific, or 2 or more races.

Discussion

For workers with employer-sponsored health insurance, health care affordability concerns have continued to grow, particularly for low-wage earners.¹⁴ The preponderance of employer-sponsored health insurance incorporates a “one size fits all” approach to health plan premiums, irrespective of employee income.¹⁵ As a result, affordability concerns for low-wage workers have contributed to this subpopulation being more likely to forgo or delay appropriate use of health care services.⁴

An understanding of race, ethnicity, and income data provides additional insight into health care affordability concerns. According to the US Census Bureau, African American and Hispanic individuals have lower median household income than their White counterparts, ($48,297 and $57,981 vs $77,999, respectively).¹⁶ Not surprisingly, as subpopulations, Black and Hispanic men and women are disproportionately overrepresented among lower-wage workers.¹⁷ As a result, these racial and ethnic subpopulations are likely subject to greater affordability pressures in relation to health care expenses than White individuals.

Our findings highlight the association of minority race and ethnicity and lower income with reduced patient use of SpRx-AIC. In the context of this analysis, employee annual income and race and ethnicity represent 2 distinct, yet interrelated variables associated with differences in SpRx-AIC use. With respect to income alone, this analysis of a subset from our earlier study confirms the previous finding that a smaller proportion of low-wage workers with AICs use SpRx-AIC medications relative to higher-wage earners. When race and ethnicity is included as an intersectional variable, we show that Black and Hispanic individuals are disproportionately negatively affected compared with White individuals, with a lower proportion of these racial and ethnic minorities using SpRx-AIC. To our knowledge, this is the first study of SpRx-AIC utilization using a national dataset that incorporates an intersectional analysis of both race and income as independent variables.

Of note, in this study, no consistent patterns in medication discontinuation or PDC by race and ethnicity or wage category were observed. Low-wage and racial and ethnic minority employees exhibited similar behaviors in relation to their higher-earning or White counterparts. This finding contrasts with other studies in which low-wage employees exhibit reduced chronic medication adherence relative to their higher-earning workers.⁴ One reason may be that the SpRx-AIC medications directly facilitate symptom relief and resolution, whereas medications for hyperlipidemia or hypertension, for example, are largely preventive. One explanation could be that SpRx-AIC use by individuals in racial or ethnic minorities or low-wage categories may be limited to those with more clinically significant AIC symptoms, whereas medication adherence is more likely because of greater perceived therapeutic benefit. Additional research is necessary to assess the strength of this hypothesis.

Other research has identified similar associations between sociodemographic attributes and SpRx-AIC medication use. However, those analyses have largely comprised single-center studies and focused principally on either race or income. In 2 different health center studies, Black patients with rheumatoid arthritis were more likely to receive steroids and conventional disease-modifying agents, and less likely to receive SpRx-AICs, in comparison with non-Black individuals though socioeconomic status (SES) was not included in these analyses.^18,19 Black individuals and other minority groups were found to be less likely to use biologic medications for psoriasis than White patients.²⁰ Multiple-year delays in SpRx-AIC prescribing were identified for Black and Hispanic patients and those receiving public insurance treated in a county hospital clinic relative to a nearby private clinic where predominantly White patients were seen.²¹

The implications of these findings are significant. A national study of patient-functional status in rheumatoid arthritis revealed worsening functional status in association with decreasing SES, as well as more rapid rates of functional status decline among low SES patients.²² In combination with our results, these findings suggest that greater use of SpRx-AIC could improve functional status as well as mitigate the rate of functional status decline in this subpopulation. Additionally, recent research has highlighted race- and ethnicity-related differences in response to treatment with specific SpRx-AICs,²³ suggesting that the selection of specific agents on the basis of race may yield enhanced benefit. Further research into understanding the role of specific health system and patient-related factors that contribute to reduced SpRx-AIC use in these subpopulations, along with a better understanding of the impact on clinical outcomes can help to focus efforts to mitigate the observed disparities.

LIMITATIONS

This study has a number of limitations. First, although the study size was large, it comprised data from self-insured employers that contracted with IBM Watson for claims analysis and may not be generalizable to all employers or noncommercially insured patients. A notable strength of the study was a national sampling of patients, reducing the likelihood that the observed findings reflected those of a single health care entity.

Further, as noted, the MarketScan dataset included race and ethnicity detail for about 30% of individuals because these data were voluntarily submitted by employers and not broadly available for the time period of the study. Unfortunately, many employers are still excluding race and ethnicity data from their medical claims databases, so our findings are a pragmatic reflection of employer practices. That noted, we did perform sensitivity analyses to determine the adequacy of the race and ethnicity data, which revealed consistent allocation of individuals by race and ethnicity in the different wage categories. Accordingly, for the analysis, we assumed that the partial data were representative of the broader MarketScan population.

Also, our claims-based analysis did not allow us to control for differences in severity between the different race and ethnicity or wage groups. Further, our use of aggregate data for all AICs may have skewed the results as a function of race and ethnicity-based differences in condition-specific prevalence rates. We acknowledge that associations exist between AIC prevalence and race and ethnicity for psoriasis,²⁴ asthma,²⁵ atopic dermatitis,²⁶ multiple sclerosis,²⁷ and ulcerative colitis,²⁸ but not apparently so for rheumatoid arthritis.²⁹ Similarly, income-based differences in prevalence rates of AICs may also exist.³⁰

As previously discussed, our use of claims data as the basis for this analysis prevents us from differentiating among the potential contributors to the observed low SpRx-AIC use rates. Additional data, including clinician prescribing patterns, clinician-patient discussions about treatment options, and initial prescription abandonment rates, will help to identify the relative importance of these causative factors. Use of claims data for this analysis limited our ability to determine the specific factors contributing to reduced prevalence of SpRx-AIC use observed in the study. Considerations include, and are not limited to, patient-related factors such as abandonment of initial prescriptions because of side effects, efficacy, cost or other concerns,^31,32 health literacy limitations,³³ lack of trust in the health care system,³⁴ unmet social needs priorities,³⁵ lack of access to patient support programs,³² and/or access to authorized or specialist prescribers of SpRx-AIC.³⁶ Clinician-related factors may include race, ethnicity or socioeconomic-related implicit bias,³⁷ or failure to discuss cost/affordability concerns.³⁸ Employment-related factors include inequitable benefit design³⁹ in addition to use of copay accumulator adjustment programs.⁴⁰ Our findings could also reflect the result of shared decision-making between patients and clinicians regarding risk-benefit considerations regarding SpRx use.⁴¹ Insight into the relative significance of these contributors will necessitate use of other data sources, which were not available for this analysis. Additional research is needed to understand and target the drivers underlying the findings in this study.

Additionally, we were unable to ascertain as to the prevalence of use of copayment accumulator assistance programs (CAAP) among individuals in each of the race and ethnicity and income subgroups. It is possible that a skewed distribution of CAAP as a component of pharmacy benefits within the different subgroups may contribute to the observed outcomes. During 2018, about one-third of employers were planning on incorporating these programs as a cost-containment measure.⁴² Given the number of employers and the size of the overall study population, we believe that a significant impact of CAAP implementation on the results our analysis is unlikely.

Finally, our analysis incorporated individual wages. Income from other household members may have confounded the observed results. To control for this, and as in prior studies,^4,9 we adjusted for zip codelevel median household income. To further account for differences in individual vs household income, we also adjusted for health plan enrollment category (single, employee plus spouse, employee plus family).

Conclusions

In conclusion, in a national sample of individuals with employer-sponsored health insurance, our study provides compelling evidence for both race and income-related variation in SpRx-AIC use. In contrast, no association between race, ethnicity, or income was observed in relation to PDC or SpRx-AIC discontinuation rates. Future research should evaluate extent to which the observed disparities in care are associated with worse clinical outcomes. Efforts to improve health equity should include attention to minimize these observed disparities in SpRx-AIC use.

Funding Statement

This study was funded by the National Pharmaceutical Council.