Copay assistance use and prescription abandonment across race, ethnicity, or household income levels for select rheumatoid arthritis and oral oncolytic medicines

Abstract

BACKGROUND:

Disparities in prescription abandonment may exacerbate health inequities. Whether copay assistance is associated with changes in prescription abandonment across different patient groups is unknown.

OBJECTIVE:

To assess disparities in copay assistance use; prescription abandonment across race, ethnicity, or income; and association of copay use with prescription abandonment and whether it differs across race, ethnicity, or household income.

METHODS:

This pooled, cross-sectional study assessed claims-level prescription data linked to a consumer database containing information on race, ethnicity, and household income for commercially insured patients. The first prescription for rheumatoid arthritis (RA) or oral oncolytic medicines from 2016 to 2020 was included. Logistic regression models measured odds of copay assistance use (copay/discount cards or free-trial voucher) and prescription abandonment (prescription not filled within 30 days of health plan approval). Interaction terms for copay assistance use by race, ethnicity, and income were tested.

RESULTS:

The sample included 67,674 patients prescribed RA medications and 9,560 prescribed oral oncolytic medications. Copay assistance use across race, ethnicity, and income ranged from 28.2% to 31.1% (RA medicines) and 27.2% to 36.7% (oral oncolytic medicines). Among those prescribed RA medicines and not using copay assistance, Black/African American, Hispanic patients, and those with household incomes less than $50,000 were more likely to abandon prescriptions than White patients and patients with household incomes more than $200,000 (odds ratio [OR] [95% CI], P value: Black/African American: 1.17 [1.06-1.29], P < 0.01; Hispanic: 1.11 [1.01-1.22], P = 0.03; income <$50,000: 1.24 [1.11-1.37], P < 0.01). Among patients using oral oncolytic medicines and not using copay assistance, there was no racial or ethnic difference in prescription abandonment. Patients using oral oncolytics with household incomes less than $50,000 were more likely to use copay assistance (1.34 [1.12-1.61], P < 0.01), but also more likely to abandon their prescriptions if not using copay assistance (1.44 [1.12-1.85], P < 0.01). Copay assistance was associated with a 79% (RA) and 71% (oral oncolytics) lower odds of prescription abandonment (0.21 [0.19-0.24], P < 0.01; 0.29 [0.24-0.36], P < 0.01), which did not differ across race, ethnicity, or income levels (P > 0.05).

CONCLUSIONS:

Copay assistance has potential to narrow disparities in prescription abandonment for commercially insured Black/African American or Hispanic patients taking RA medicines and patients with household incomes less than $50,000; however, efforts to improve access to copay assistance are needed. Copay assistance, as a factor facilitating equal access to medicines, is an important consideration when evaluating policies that impact access to copay assistance programs.

Plain language summary

Is copay assistance related to the likelihood of patients of different race, ethnicity, or income picking up rheumatoid arthritis (RA) or oncology prescriptions? Black/African American and Hispanic patients with RA medicine and patients with incomes less than $50,000 were more likely to not pick up prescriptions. Copay assistance reduced the likelihood of patients not picking up their prescription similarly across race, ethnicity, and income levels, and thus has the potential to narrow the identified prescription use differences.

Implications for managed care pharmacy

This study found significant racial, ethnic, and income disparities in both the use of copay assistance and prescription abandonment for select RA and oral oncolytic medications. Among commercially insured patients, the association of copay assistance with prescription abandonment did not vary across race, ethnicity, or income levels. Improving access to copay assistance may narrow health inequities and disparities of care. These findings should be considered when evaluating policies, which impact access to copay assistance programs.

Among adult patients in the United States prescribed medications, approximately 25% have challenges with out-of-pocket (OOP) costs.¹ Patients with financial barriers may choose to not fill their prescriptions, also known as prescription abandonment.² In 2019, approximately 9% of all prescriptions were abandoned at retail pharmacies, although this rate increased to approximately 60% when cost sharing was more than $500 per fill.² The impact of patients’ nonadherence to their medicines has been estimated to result in as much as $100-$300 billion in avoidable health care costs each year and thus has been a focus for policymakers and health care providers to identify solutions that may reduce patient OOP costs.

Although many studies have focused on OOP costs and benefit designs^3,4 as a factor driving prescription abandonment, less is known on whether these factors may impact prescription abandonment equally across different races, ethnicities, or incomes. Previous research has found associations between race, ethnicity, or income with levels and types of insurance coverage, with patients from racial and ethnic minority groups being more likely to lack insurance coverage and less likely to be privately insured.^5,6 Additionally, a 2021 survey found that a greater proportion of Black, Hispanic, and lower income patients reported difficulties in affording their medicines.¹ Although limited, data suggest there may be disparities in race, ethnicity, or income in prescription abandonment,^2,7,8 as well as for related outcomes including medication adherence⁹ or prescription use.¹⁰ Given the implications of prescription abandonment on poorer clinical outcomes,¹¹ disparities in abandonment may also lead to disparities in health outcomes. Furthermore, previous research has shown prescription abandonment for specialty drugs, such as oral oncolytics and anti-inflammatory drugs to be as high as 67%³ and 52.3%,¹² respectively. Thus, the disparities in prescription abandonment in these therapeutic areas could further widen existing differences in health outcomes. Thus far, there have been few solutions evaluated that address disparities in prescription abandonment.

One potential mechanism to alleviate financial barriers to prescription medication use is copay assistance programs. Copay assistance programs may take different forms, such as manufacturer copay cards and patient assistance foundations, and work by reducing or eliminating insurance cost-sharing requirements for patients, thereby enabling patients to access their prescriptions with lower or no OOP cost. The use of copay assistance is particularly prevalent in specialty drug areas, in which copay assistance use has been estimated to be 20% in oncology and up to 65% among patients using specialty drugs.^13,14 In addition to reducing OOP costs, research has also suggested that copay assistance programs may improve medication adherence^12,15 and other clinical outcomes.¹⁶ Given that copay assistance programs reduce the OOP burden for patients, in theory these programs should reduce the disparities in access to medicines; however, there is a lack of data thus far on the association between copay assistance and achieving equal health outcomes for all.

Furthermore, there are limited data on whether copay assistance programs reach the populations most in need. Awareness,¹⁷ eligibility criteria, and application processes vary by copay assistance program and may be complex for some patients,^18-20 creating additional barriers to access. Additionally, policies, such as the anti-kickback law in Section 1128B(b) of the Social Security Act, prohibit the use of manufacturer copay assistance for patients enrolled in federal programs such as Medicare.²¹ Although some health care delivery systems have robust processes in place to help patients in need access copay assistance, their availability is not uniform.

Given the potential of copay assistance to reduce OOP costs and disparities in prescription abandonment, particularly in specialty drug areas in which the risk of prescription abandonment and copay assistance use are high, the objectives of this analysis are as follows: To (1) describe whether there are disparities in use of copay assistance by race, ethnicity, or income; (2) understand whether there is a disparity in prescription abandonment by race, ethnicity, or income for which copay assistance may potentially help address; and (3) understand the association between use of copay assistance and prescription abandonment and whether it may differ by race, ethnicity, or income.

Methods

DATA SOURCES

This was a pooled, cross-sectional study (Figure 1) using the real-world IQVIA Longitudinal Access and Adjudicated Dataset (LAAD) pharmacy claims data linked to Experian consumer data. The LAAD pharmacy data are a deidentified claims-level dataset sourced from a number of pharmacy points of sale serving approximately 300 million patients and is estimated to represent approximately 92% of prescriptions dispensed at retail pharmacies and approximately 70% of prescriptions through mail order pharmacies, depending on the specialty. The LAAD data capture transactions that occur during the adjudication process, including when the health plan approves/rejects the claim, as well as whether the prescription is paid for and picked up from the pharmacy. A claim in LAAD also has information on whether any copay assistance (manufacturer’s copay card, other discount card, or free-trial voucher) for the transaction is associated with the payment; however, the use of e-coupon vouchers and denial conversion programs are not captured in the dataset. This database has been used in prior research to study prescription use.^22,23

FIGURE 1.

Study Design

VARIABLE DEFINITIONS

Race, ethnicity, and household income were obtained from the Experian Consumer Demographics Data, ConsumerView, which contains deidentified consumer information for more than 311 million consumers and 126 million households, collected from a variety of public and proprietary sources, including self-reported information, public records, and purchase transaction information. Linkage of the datasets was done at the patient level via an anonymized cryptographic token process, which has been validated with proprietary methods, allowing the Experian data to be appended to the LAAD dataset as nonidentified consumer information and providing insight into patient demographics. Income is based on Experian’s multiple proprietary statistical methodologies to predict household income in the most recent year captured and was grouped into 5 categories (<$50,000, $50,000-$74,999, $75,000-$99,999, $100,000-$199,999, and >$200,000). Race and ethnicity classification was based on Experian’s proprietary name and geography technology and summarized into 6 categories (White, Black/African American, Hispanic, Asian, Other Race [ie, Native American, Polynesian, etc], and unknown). Although separate social constructs, race and ethnicity are reported together in the data. Race, ethnicity, and income variables reflect the most current data captured that were available in the Experian database at the time of analysis. The match rate between the 2 databases for this analysis was approximately 49%.

Variables included in the LAAD dataset include age (categorized into age groups 18-34, 35-49, and 50-64 years), sex, zip code, state, prescribing provider type (ie, rheumatologist, oncologist, primary care physician, etc), and OOP cost (categorized into groups $0-$9, $10-$39, $40-$124, $125-$499, and $500+). State was further used to describe geographic region based on the 9 US Census Bureau divisions.²⁴ Zip code was used to determine rural or urban status.²⁵

PATIENT SELECTION

Select branded drugs from 2 therapeutic areas were examined in this analysis: rheumatoid arthritis (RA) medicines (adalimumab, etanercept, tofacitinib, 12 palbociclib, upadacitinib) and oral oncolytics (12 palbociclib, alectinib, ibrutinib, 12 palbociclib, nilotinib, lenalidomide, abiraterone, sorafenib, dasatinib, enzalutamide) (Supplementary Table 1 ^{(315.9KB, pdf)} , available in online article). The prescription for that branded drug was included in the analysis if (1) the claim was the patient’s first for that brand (as determined by looking back until January 1, 2016), (2) the claim was approved by a commercial payer (ie, Medicare, Medicaid and other insurance types excluded), and (3) the patient had commercial insurance during the study period of January 1, 2016, through December 31, 2020. The decision to include 2020 data in the analysis, despite the COVID-19 pandemic, was based on examining yearly abandonment rates in which we did not observe notable departures from abandonment trends overall and by therapeutic area, race, ethnicity, or income categories. Lastly, we removed patients who were not flagged as receiving copay assistance, yet had differences in their initial and final OOP costs (9,764 patients [11.2%] removed), to minimize potential confounding due to other types of assistance or plan programs not detected or specified in the dataset (ie, e-coupons or copay maximizer programs) (Supplementary Table 2 ^{(315.9KB, pdf)} ).

OUTCOMES

Two outcomes were assessed in this analysis, the receipt of copay assistance and prescription abandonment (Supplementary Figure 1 ^{(315.9KB, pdf)} ). The receipt of copay assistance was defined as the presence of manufacturer copay card, other discount card, or free-trial voucher on the patient’s first claim during the study period for a selected branded product. Prescription abandonment was assessed by examining whether the patient’s first attempt to fill a prescription was filled within 30 days from the day of approval by the health plan. The analysis was restricted to the patient’s first fill attempt for one of the branded products listed above, given this is most likely when patients would initiate the use of copay assistance.

STATISTICAL METHODS

Patient characteristics were reported as descriptive statistics using frequency and percentage distributions. Statistical differences, defined as P < 0.05, between copay card assistance users and nonusers were assessed using Pearson’s chi-square test.

To examine potential disparities in the use of copay assistance, a logistic regression was used to measure the association between race, ethnicity, and household income with copay assistance use while adjusting for potential confounders. Secondly, to assess which patient groups may be at a higher likelihood of abandoning prescriptions and thus potentially benefit from copay assistance, we used a multivariable logistic regression to estimate the association between race, ethnicity, and household income with prescription abandonment among patients who did not receive copay assistance. Lastly, the association between copay assistance and prescription abandonment and whether it varied by race, ethnicity, or income was assessed using 2 adjusted multivariable logistic regression models. The first model assessed the association of the use of copay assistance with prescription abandonment. In the second model, to determine whether any association of copay assistance and prescription abandonment varied by race, ethnicity, or income, 2 interaction terms were added and tested using a chi-square test. Model covariates included sex, age, race, ethnicity, household income, region, prescribing provider type, drug, year, initial OOP cost, and rural/urban. Each association was assessed separately for patients prescribed RA medicines and patients prescribed oral oncolytics, with results reported as odds ratios (ORs) with 95% CIs. Sensitivity analyses excluding prescriptions in 2020 were conducted to assess the robustness of results when excluding prescriptions during the COVID-19 pandemic. All analyses were conducted using STATA version 14.2. To guide reporting of results, the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) study checklist²⁶ was used. This study only used Health Insurance Portability and Accountability Act-compliant deidentified patient data; thus, institutional review board approval was not required for this study.

Results

PATIENT CHARACTERISTICS

The sample included 67,674 patients prescribed advanced RA medicines and 9,560 patients who were prescribed an oral oncolytic during the study period. The characteristics of patients by receipt of copay assistance were similarly distributed in terms of age, with most aged 50-64 years (RA medicines: 52.7% [no copay assistance] and 51.8% [copay assistance]; oral oncolytics: 82.0% [no copay assistance] and 79% [copay assistance]). Race, ethnicity, and region were also similarly distributed in both therapeutic area cohorts (Table 1). A greater proportion of the Black/African American and Hispanic patient populations had household incomes less than $50,000 compared with other income categories, whereas the Asian patient population had a greater proportion of household incomes in the $100,000 to $199,000 and more than $200,000 categories compared with other income categories (Supplementary Figure 2 ^{(315.9KB, pdf)} ), with the percentages across racial or ethnicity ranging from 28.6% to 31.2% for RA medicines and from 27.2% to 36.7% for oral oncolytics. Approximately one-third of patients (RA medicines 28.2%, oral oncolytics 35.1%) in the lowest income category used copay assistance.

TABLE 1.

Patient Characteristics by Copay Assistance Use and Therapeutic Area

	Rheumatoid arthritis medicines			Oral oncolytic medicines
	No copay assistance n = 47,228	Copay assistance n = 20,446	P value	No copay assistance n = 6,410	Copay assistance n = 3,150	P value
Age (in years), n (%)
18-34	7,120 (15.1)	3,131 (15.3)	0.08	183 (2.9)	133 (4.2)	<0.01
35-49	15,221 (32.2)	6,733 (32.9)		968 (15.1)	529 (16.8)
50-64	24,887 (52.7)	10,582 (51.8)		5,259 (82.0)	2,488 (79.0)
Sex, n (%)
Female	31,013 (65.7)	13,188 (64.5)	<0.01	3,159 (49.3)	1,860 (59.0)	<0.01
Male	16,200 (34.3)	7,227 (35.3)		3,250 (50.7)	1,288 (40.9)
Unknown	15 (0.0)	31 (0.2)		1 (0.0)	2 (0.1)
Race and ethnicity, n (%)
White	35,915 (76.0)	15,679 (76.7)	0.09	4,702 (73.4)	2,320 (73.7)	0.02
Black/African American	4,108 (8.7)	1,750 (8.6)		658 (10.3)	381 (12.1)
Asian	940 (2.0)	426 (2.1)		179 (2.8)	72 (2.3)
Hispanic	4,252 (9.0)	1,702 (8.3)		532 (8.3)	238 (7.6)
Other	875 (1.9)	379 (1.9)		166 (2.6)	62 (2.0)
Unknown	1,138 (2.4)	510 (2.5)		173 (2.7)	77 (2.4)
Household income, n (%)
$0-$49,999	10,916 (23.1)	4,287 (21.0)	<0.01	1,345 (21.0)	729 (23.1)	<0.01
$50,000-$74,999	9,831 (20.8)	4,336 (21.2)		1,348 (21.0)	674 (21.4)
$75,000-$99,999	8,538 (18.1)	3,862 (18.9)		1,125 (17.6)	578 (18.3)
$100,000-$199,999	13,241 (28.0)	5,852 (28.6)		1,775 (27.7)	843 (26.8)
$200,000+	4,702 (10.0)	2,109 (10.3)		817 (12.7)	326 (10.3)
Region, n (%)
New England	1,811 (3.8)	543 (2.7)	<0.01	202 (3.2)	54 (1.7)	<0.01
Mid-Atlantic	5,429 (11.5)	2,178 (10.7)		762 (11.9)	244 (7.7)
East North Central	8,865 (18.8)	3,557 (17.4)		1,057 (16.5)	559 (17.7)
West North Central	4,431 (9.4)	2,066 (10.1)		614 (9.6)	328 (10.4)
South Atlantic	8,580 (18.2)	3,879 (19.0)		1,132 (17.7)	628 (19.9)
East South Central	3,468 (7.3)	1,814 (8.9)		432 (6.7)	376 (11.9)
West South Central	6,839 (14.5)	3,700 (18.1)		1,110 (17.3)	445 (14.1)
Mountain	2,911 (6.2)	1,094 (5.4)		432 (6.7)	211 (6.7)
West	4,573 (9.7)	1,461 (7.1)		650 (10.1)	301 (9.6)
Other/unknown	321 (0.7)	154 (0.8)		19 (0.3)	4 (0.1)
Provider type, n (%)
Other	20,827 (44.1)	9,247 (45.2)	0.01	1,208 (18.8)	582 (18.5)	0.66
Rheumatologist/oncologist	26,401 (55.9)	11,199 (54.8)		5,202 (81.2)	2,568 (81.5)
Rural/urban, n (%)
Urban	43,680 (92.5)	19,016 (93.0)	0.02	6,055 (94.5)	2,966 (94.2)	0.55
Rural	3,548 (7.5)	1,430 (7.0)		355 (5.5)	184 (5.8)
Year, n (%)
2016	10,966 (23.2)	7,229 (35.4)	<0.01	1,688 (26.3)	802 (25.5)	0.02
2017	8,781 (18.6)	4,863 (23.8)		1,311 (20.5)	622 (19.7)
2018	9,311 (19.7)	3,201 (15.7)		1,271 (19.8)	602 (19.1)
2019	10,107 (21.4)	2,739 (13.4)		1,141 (17.8)	652 (20.7)
2020	8,063 (17.1)	2,414 (11.8)		999 (15.6)	472 (15.0)

RACIAL AND ETHNIC GROUP AND INCOME DISPARITIES WITH USE OF COPAY ASSISTANCE

RA Medicines. Overall, 30.2% of prescribed RA medicines used copay assistance. Among patients receiving RA medicines, the use of copay assistance was approximately 30% across all races and ethnicities (White 30.4%, Black 29.9%, Hispanic 28.6%, Asian 31.2%). After adjusting for baseline characteristics, Black/African American and Hispanic patients had a 7% (adjusted OR [95% CI]: 0.93 [0.88-1.00], P = 0.04) and 10% (0.90 [0.85-0.97], P < 0.01) lower odds of using copay assistance compared with White patients (Figure 2). The use of copay assistance across income categories ranged from 28% to 31%, with those in the lower household income groups having an 8% and 21% lower odds of using copay assistance compared with those with household incomes more than $200,000 (adjusted OR [95% CI]: $50,000-$74,000: 0.92 [0.86-0.98], P = 0.01; $0-$49,000: 0.79 [0.74-0.84], P < 0.01).

FIGURE 2.

Association of Race, Ethnicity, and Income With Use of Copay Assistance

Oral Oncolytics. In total, 32.9% of prescribed oral oncolytic medicines used copay assistance. In contrast, for patients receiving oral oncolytics, there was no association between race and ethnicity and receipt of copay assistance (Figure 2). Furthermore, a differing trend from RA medicines was also observed for the association of income with copay assistance use, with those in the lowest income group having 34% greater odds of receiving copay assistance than those with household incomes more than $200,000 (adjusted OR [95% CI]: 1.34 [1.12-1.61], P < 0.01). Similar trends for both RA medicines and oral oncolytics were observed in the sensitivity analyses excluding prescriptions during the COVID-19 pandemic (2020) (Supplementary Table 3 ^{(315.9KB, pdf)} ).

RACIAL AND ETHNIC GROUPS AND INCOME DISPARITIES IN PRESCRIPTION ABANDONMENT

RA Medicines. Among patients not receiving copay assistance, Black/African American patients had a 17% (adjusted OR [95% CI]: 1.17 [1.06-1.29], P < 0.01) greater odds of prescription abandonment and Hispanic patients had an 11% (adjusted OR [95% CI]: 1.11 [1.01-1.22], P = 0.03) greater odds of prescription abandonment relative to White patients (Figure 3). Patients in the lowest income category ($0-$49,000) were more likely to abandon their prescriptions compared with those in the highest income category ($200,000+) (adjusted OR [95% CI]: 1.24 [1.11-1.37], P < 0.01).

FIGURE 3.

Association of Race, Ethnicity, and Income With Prescription Abandonment Among Patients Not Using Copay Assistance

Oral Oncolytics. There was no significant difference in prescription abandonment by race and ethnicity. Patients in the income categories of $0-$49,000, $50,000-$74,000, and $100,000-$199,000 had higher odds of abandonment compared with the highest income category (Figure 3). Similar trends for both RA medicines and oral oncolytics were observed in the sensitivity analyses excluding prescriptions during the COVID-19 pandemic (2020) (Supplementary Table 3 ^{(315.9KB, pdf)} ).

ASSOCIATION OF COPAY ASSISTANCE WITH PRESCRIPTION ABANDONMENT ACROSS RACE, ETHNICITY, AND INCOME

RA Medicines. Use of copay assistance was associated with a 79% lower odds of prescription abandonment for patients using RA medicines (adjusted OR [95% CI]: 0.21 [0.19-0.24], P < 0.01) (Table 2). In the model, which included interaction terms to assess whether the association of copay assistance with prescription abandonment depended on race, ethnicity, or income, neither of the interaction terms was statistically significant (copay assistance use × race and ethnicity: P = 0.30, copay assistance use × income: P = 0.13).

TABLE 2.

Association of Copay Assistance Use With Prescription Abandonment

	Odds ratio (95% CI)	P value
Model 1: copay assistance use (ref. no copay assistance use)
Rheumatoid arthritis medicines	0.21 (0.19-0.24)	<0.01
Oral oncolytic medicines	0.29 (0.24-0.36)	<0.01
Model 2: include interaction terms
Rheumatoid arthritis medicines
Copay card use × race and ethnicity	—	0.30
Copay card use × income	—	0.13
Oral oncolytic medicines
Copay card use × race and ethnicity	—	0.32
Copay card use × income	—	0.58

Model 1 was adjusted for sex, age, region, race and ethnicity, income, prescribing specialist, product, year, initial out-of-pocket cost, and rural/urban. Model 2 included the same covariates as Model 1 with the addition of interaction terms for copay card use × race and ethnicity, and for copay card use × income.

ref. = reference.

Oral Oncolytics. Use of copay assistance was associated with a 71% lower odds of prescription abandonment for patients using oral oncolytics (adjusted OR [95% CI]: 0.29 [0.24-0.36], P < 0.01) (Table 2). In the model, which included interaction terms to assess whether the association of copay assistance with prescription abandonment depended on race, ethnicity, or income, neither of the interaction terms was statistically significant (copay assistance use × race and ethnicity: P = 0.32, copay assistance use × income: P = 0.58).

Discussion

In this pooled cross-sectional study using real-world data, we found racial or ethnic and income disparities in both the use of copay assistance and prescription abandonment. To our knowledge, this is the first study to assess the association of copay assistance with disparities in prescription abandonment. In patients receiving RA medicines, Black/African American, Hispanic, and lower income patients were not only more likely to abandon their prescriptions but also less likely to use copay assistance (for their first prescription) compared with White and higher income patients, respectively. For patients receiving oral oncolytics, no racial or ethnic disparities were observed; however, a disparity in prescription abandonment between the highest and lower household income levels was observed. In both cases, we found that the association of copay assistance with a reduction in prescription abandonment did not differ by race, ethnicity, or income, suggesting that copay assistance could have a role in improving equal access to medicines among patient groups in which there is a disparity in prescription abandonment.

Despite the finding that the use of copay assistance may be associated with similar effects on prescription abandonment across racial or ethnic and income groups, the disparities in use of copay assistance present a barrier to enabling patients to realize any potential equity effects associated with copay assistance. The findings of this study differ from a recent study that found non-White patients were more likely to use copay assistance (on any prescription) in immunology.²⁷ This difference may be due to our study focusing only on the use of copay assistance on the first fill attempt for a particular medicine, and it is possible that there may be a disparity in the timing of the receipt of copay assistance (considering subsequent fills) or the disparities observed may not persist when accounting for switching to other medicines. Nonetheless, further efforts may be needed to establish or target medication assistance programs to connect patients more likely to abandon prescriptions with copay assistance programs and help narrow the disparity in use of copay assistance and subsequent drug abandonment. Many oncology clinics have financial navigator programs, which may explain the observation that lower income oral oncolytic patients were more likely to use copay assistance, as well as the lack of difference in use of copay assistance by race and ethnicity, which was found in the RA cohort. That said, only 35% of the lowest income oral oncolytic cohort used copay assistance. Given the potential reduction of prescription abandonment with improved access to copay assistance, the current low levels of copay assistance use represent a potential opportunity to further improve access to prescribed medicines, particularly for therapeutic areas with higher rates of abandonment such as these.^3,12

Our findings of disparities among race and ethnicity (in RA) and income in prescription abandonment are consistent with previous studies, which also found these factors to be associated with specialty drug abandonment. Huang et al⁷ found that Black patients were more likely to abandon their prescribed oral preexposure prophylaxis in preventing HIV compared with White patients. Additionally, Streeter et al⁴ found lower income to be associated with a higher abandonment rate among patients prescribed oral oncolytic therapy. Despite the growing evidence of these disparities, little is known on the reasons behind these observations. Given that studies have previously shown that patient exposure to cost sharing are also associated with prescription abandonment,^3,28 one possible explanation is that the level of insurance benefits may vary among race, ethnicity, or income; however, further research is needed to explore the role of coverage level/benefits as a mediator.

The finding that copay assistance may narrow disparities in prescription abandonment suggests a role of copay assistance beyond simply lowering OOP costs. Given the COVID-19 pandemic has brought greater attention to the health inequity issues persistent in our health care system,²⁹ the equity impacts of copay assistance are an important consideration and may contribute to pharmacoequity, a concept defined as “ensuring all individuals, regardless of race, ethnicity, socioeconomic status or availability of resources, have access to the highest-quality medications required to manage their health needs.”³⁰ Thus, when evaluating policies and programs that may impact access to copay assistance programs, the impact on pharmacoequity should also be considered. For example, when evaluating the impact of copay accumulator programs, which exclude copay assistance amounts from patients’ deductibles and OOP maximums, it may be important not only to assess their effects in terms of clinical and economic outcomes but also to consider their potential impact on access to medicines for underserved populations.

Although our study examined the association of copay assistance with patients’ first prescription abandonment, it is unknown whether copay assistance may have any association with narrowing disparities for outcomes such as adherence or persistence, for which longer term or cumulative cost barriers may play a role. Although the literature assessing the impact of copay assistance is limited, literature reviews have suggested copay assistance to be positively associated with adherence.^15,16 Additionally, a study found that copay assistance was associated with longer treatment persistence with anaplastic lymphoma kinase inhibitors.³¹ Nonetheless, future research is warranted to understand whether the findings here on copay assistance and prescription abandonment extend to adherence and persistence.

LIMITATIONS

There are several limitations in this study to consider. First, this analysis was a pooled cross-sectional analysis of 2 therapeutic areas and, as such, has limitations to the interpretation and generalizability of the results. For example, income may not be captured at the time of the prescription, and it was difficult to determine if patients in the beginning of the study period were truly a new start, both of which may influence the abandonment rate. Additionally, the 2 therapeutic areas may not be generalizable to all specialty drugs or therapeutic areas. Although disparities were observed in both RA medicines and oral oncolytics, the differences in disparities between them highlight factors that may drive disparities specific to each therapeutic area. Second, the match rate between the databases was approximately 50%, thus only representing a subset of patients in the database. It is unknown how the unmatched patients may have impacted the observed disparities, but this is nonetheless a limitation of the available data sources to examine the research question. Additionally, the variables matched on have limitations to note. Race and ethnicity were estimated rather than self-reported, and income may not reflect the affordability of a medication to a patient given household size is not accounted for. Furthermore, there may be other unmeasured confounders related to socioeconomic status that may impact prescription abandonment, such as transportation availability or food insecurity. Lastly, we observed that fewer patients had a prescription in 2020, likely due to the COVID-19 pandemic. It is possible that many patients did not seek initial treatment during this period, although it is unknown how this might influence the disparities in abandonment observed in this study. Nonetheless, abandonment trends were similar when excluding this time period in sensitivity analyses, thus this limitation may not have a significant impact on our findings.

Conclusions

Among commercially insured patients prescribed RA medicines or oral oncolytics, copay assistance is associated with lower prescription abandonment, which did not differ by race, ethnicity, or income levels. Thus, although use of copay assistance has the potential to narrow disparities in prescription abandonment for Black/African American or Hispanic patients taking RA medicines and patients with household incomes less than $50,000 (RA medicines and oral oncolytic medicines), efforts to improve and ensure equal access to copay assistance is needed to realize any such potential outcome. These findings shed light on the role of copay assistance as a factor facilitating equal access to medicines, an important consideration when evaluating policies that impact access to copay assistance programs.