Non–cost-related sources of medication nonadherence in the Medicare population

Abstract

Significant research and attention to date have focused on cost-related medication nonadherence as rising prescription drug prices worsen affordability and access for many Americans. This study investigated self-reported sources of medication nonadherence, measuring both cost- and non–cost-related medication nonadherence among community-dwelling Medicare Part D beneficiaries in 2022. A total of 13.7% of beneficiaries (4 589 843) reported some type of medication nonadherence; 7.5% reported medication nonadherence related to cost and 6.2% reported for non-cost reasons. Beneficiaries reporting food insecurity, poor functional status, and lack of understanding of the Part D benefit were more likely to report both types of medication nonadherence after adjustment for sociodemographic factors. Beneficiaries receiving the Low-Income Subsidy had lower odds of reporting cost-related but greater odds of reporting non–cost-related medication nonadherence. These findings suggest that non–cost-related sources of medication nonadherence, such as beneficiary preferences or beliefs, understanding of their health situation or insurance coverage, and ability to fill a prescription, are significant contributors to overall nonadherence. Non–cost-related medication nonadherence should be considered alongside recent changes to the Part D benefit and in future Part D Centers for Medicare and Medicaid Services (CMS) Innovation Center models, such as the proposed Medicare $2 Drug List Model, in order to maximize the impact of these initiatives.

Introduction

Over 50 million Medicare beneficiaries have prescription drug coverage under Medicare Part D,¹ and previous studies have shown that 10%–25% of this population fail to fill or take their prescriptions as prescribed.^2,3 Medication nonadherence (MNA) can lead to worse health outcomes and increased or unnecessary spending.^2-5 Understanding sources of MNA will help interested parties measure the impact of planned changes to the Medicare Part D program, including the Inflation Reduction Act of 2022 (IRA), which will include several Part D benefit changes aimed at improving prescription drug access, affordability, and overall Medicare beneficiary health. Additionally, understanding the sources of MNA will help inform future Part D Centers for Medicare and Medicaid Services (CMS) Innovation Center models, such as the proposed Medicare $2 Drug List (M2DL) Model⁶ that is under development and seeks to address cost- and non–cost-related barriers that patients face in accessing and filling their prescriptions.

Previous studies of MNA among Medicare beneficiaries have focused on cost-related MNA and identified higher rates of nonadherence among beneficiaries who are under 65 years of age, those who have low incomes, belong to a historically marginalized population, or have worse health.^2-4 For example, a recent report by the Office of the Assistant Secretary for Planning and Evaluation estimated that a total of 3.7 million Medicare beneficiaries did not fill prescriptions due to cost in 2019: 4.4% of beneficiaries aged 65 years or older and 17.8% of beneficiaries under the age of 65 years.³ In another recent study, Dusetzina et al² estimated cost-related MNA for the 65-years-and-older population in 2022 at 20.2%. Similarly, Nekui et al⁵ studied cost-related MNA in Medicare beneficiaries, controlling for Medicare Part D enrollment and type of prescription drug coverage. Nekui et al observed that beneficiaries enrolled in the Part D Low-Income Subsidy (LIS) program, whose premiums and out-of-pocket drug costs are lowered, were less likely to report cost-related MNA compared with near-poor and beneficiaries who did not receive the LIS.

Some literature considers non–cost-related (other) MNA,^7-10 but studies of other MNA among the Medicare population are sparse. Specific categories and nomenclature for other MNA vary, but commonly cited causes include a beneficiary's preferences or beliefs, such as trust in provider or fear of side effects; knowledge and understanding of their health situation or their insurance coverage, including belief in drug effectiveness, self-denial of conditions, or knowledge of what is covered by insurance; and ability to fill a prescription, such as poor cognitive function or limited access to a pharmacy.

The current study expands on previous Medicare cost-related MNA research by characterizing both cost-related MNA and other MNA, seeking to quantify the prevalence of other MNA and understand the characteristics of beneficiaries that report MNA. This study also adds to the current literature by introducing new beneficiary characteristics associated with MNA, such as difficulty with transportation, knowledge of Medicare drug benefits, and Part D plan benefit type. Finally, this study examines the association of LIS enrollment with cost-related and other MNA among Part D beneficiaries.

Data and methods

Data collection

This study utilized a core set of self-reported measures in the RXMED and ACCSSMED segments of the 2022 Medicare Current Beneficiary Survey (MCBS) Survey File Limited Data Set (LDS)¹¹ as outcome measures. The MCBS is a longitudinal survey of Medicare beneficiaries administered by the CMS and is used to produce nationally representative estimates of Medicare beneficiaries. Most interviews in the current analysis were conducted by telephone due to the COVID-19 pandemic. The MCBS LDS also includes CMS administrative data, such as enrollment, prescription drug claims, and plan characteristics. In this study, the authors matched additional insurance information to MCBS data to enhance categorization of prescription drug plan benefits.

The 2022 MCBS collected information during the summer of 2023 from Medicare beneficiaries who were enrolled in Medicare in 2022.¹¹ The information included sociodemographic characteristics, health status and functioning, access to care, general health, use of prescription drugs and other forms of health care, and self-reported ability to understand the Medicare prescription drug benefit.

Beneficiaries were defined as food insecure if they reported that they sometimes or often had no money to buy more food when they ran out, could not afford balanced meals, cut the size or frequency of meals due to cost, ate less because they did not have enough money for food, or were hungry because they did not have enough money for food. Health insurance status was coded based on a combination of administrative and self-reported data; enrollment was used to determine a beneficiary's Medicaid, Medicare Advantage, or LIS status. This study also evaluated Part D plan benefit type, comparing basic and enhanced Medicare Advantage prescription drug plans (MAPDs) and standalone prescription drug plans (PDPs).

The analysis for this paper was generated using SAS software, version 9.4, of the SAS System for Windows (2013; SAS Institute, Inc). SAS and all other SAS Institute, Inc, product or service names are registered trademarks or trademarks of SAS Institute, Inc, Cary, NC, USA.

Definitions of nonadherence

Estimates reflect beneficiaries living in the community only and exclude beneficiaries living in facilities, such as long-term care facilities or nursing homes.

This study categorized cost-related and other MNA into mutually exclusive categories through 2 series of questions; one in the MCBS fall questionnaire (Path 1) and another in the MCBS winter questionnaire (Path 2) (see Appendix Figure 1).

In Path 1, a beneficiary experienced cost-related MNA if they indicated during the past year that they had sometimes or often taken smaller doses or skipped doses of a prescription to make it last longer, and in conjunction also reported delayed filling a prescription due to cost or failed to fill a prescription due to cost. Beneficiaries experienced other MNA if they indicated that, during the past year, they had sometimes or often taken smaller doses or skipped doses of a prescription to make it last longer but did not also report cost concerns. Beneficiaries meeting neither of these criteria were categorized as having no MNA.

In Path 2, a beneficiary experienced cost-related MNA if they indicated that, during the past year, there were medicines prescribed for them that they did not get, and in conjunction selected “Thought it would cost too much” as the reason for which they did not receive the medication. Beneficiaries experienced other MNA if they indicated that, during the past year, there were medicines prescribed for them that they did not get but did not select “Thought it would cost too much.” Beneficiaries who did not indicate that they failed to receive medication were categorized as having no MNA.

If beneficiaries experienced MNA and indicated that nonadherence was due to cost, they were placed in the cost-related MNA group. If beneficiaries experienced MNA but did not indicate that nonadherence was due to cost, they were placed in the other MNA group. If they had no MNA in either path, they were placed in neither group. Some beneficiaries classified as “Other MNA” indicated why they were nonadherent, with indicated reasons including trouble obtaining the drug, not liking the reaction to the drug or not believing it would help, substituting other drugs, or taking smaller doses to make the drug last.

Analytic methods

To account for the stratified, unequal-probability, multistage sampling in the MCBS, balanced repeated replication (BRR) method weighting was used to estimate characteristics of the national population of community-dwelling Medicare beneficiaries who were enrolled in a PDP or MAPD for at least 1 month in 2022. Certain Part D plan types that only apply to a certain subset of the Medicare population, such as Employer Group Waiver Plans (EGWPs), Program of All-Inclusive Care for the Elderly (PACE) plans, and institutional type plans, were excluded from this analysis. This BRR method allows the analysis to produce accurate point and standard error estimates that account for intra-cluster correlations.³ The weighted population size of the study sample was approximately 33.3 million beneficiaries.

The authors used Rao-Scott chi-square tests with BRR weighting to examine the association of key predictors with outcomes of cost-related MNA and other MNA. Predictors included sociodemographic characteristics, insurance coverage, health status, and access-to-care measures derived from MCBS data.

Multivariate logistic regression models with BRR weighting were used to produce odds ratios and estimated standard errors that demonstrated strengths of associations between these same variables and outcomes. The models were adjusted for age (<65 years old, ≥65 years), race (non-Hispanic White, non-Hispanic Black, Hispanic, other), gender (male, female), education (high school graduate or less, more than high school), and general health (good to excellent health, poor to fair health). These covariates were chosen to control for overall health status and complexity of medication regimen among beneficiaries, as previous studies have shown that the likelihood of nonadherence grows as patients are prescribed more medications.⁴

Results

Weighted frequencies of the study sample (Table 1) showed that 58.6% of Medicare Part D beneficiaries received their Part D coverage from MAPDs (predominantly plans that offer enhanced benefits), while the other 41.4% received coverage from PDPs (which were more evenly split between basic and enhanced benefit types). Approximately one-quarter of beneficiaries were dually eligible for Medicaid (25.9%) or received the LIS (29.7%) during some point in the year. Almost one-third (30.7%) of Medicare beneficiaries reported that their Part D benefit was difficult to understand. A significant proportion of the population had difficulties with transportation (34.8%), or food insecurity (22.0%).

Table 1.

Summary of population characteristics.

	Unweighted n	Weighted n	Weighted % of sample	SE of %
Total	4579	33 259 735
Age
Under 65 y	928	5 005 473	16.5%	0.7%
65+ y	3079	25 415 567	83.5%	0.7%
Gender
Male	1991	13 986 978	42.1%	0.8%
Female	2588	19 272 757	57.9%	0.8%
Race
Non-Hispanic White	3253	24 221 703	72.8%	1.0%
Non-Hispanic Black	508	3 720 977	11.2%	0.8%
Hispanic	576	3 278 415	9.9%	0.8%
Other/mixed/missing	242	2 038 640	6.1%	0.6%
Income to poverty ratio
≤138%	1617	10 438 294	31.4%	1.1%
>138%	2962	22 821 441	68.6%	1.1%
Food insecurity
Any food insecurity	1036	7 301 983	22.0%	0.8%
No food insecurity	3537	25 921 851	78.0%	0.8%
Educational level
HS graduate or less	1958	13 558 427	40.9%	1.1%
Some college or greater	2600	19 593 282	59.1%	1.1%
Metropolitan status
Metropolitan	3623	27 430 101	82.5%	1.2%
Micropolitan/small town/rural	950	5 800 331	17.5%	1.2%
MAPD/PDP type
MAPD enhanced	2324	16 823 730	50.6%	1.1%
MAPD basic	438	2 654 516	8.0%	0.6%
PDP enhanced	960	7 759 081	23.3%	0.9%
PDP basic	857	6 022 407	18.1%	0.8%
Dual
Any dual eligibility for the year	1354	8 628 325	25.9%	1.0%
Not dually eligible	3225	24 631 410	74.1%	1.0%
LIS
Any for the year	1548	9 875 424	29.7%	1.0%
No LIS	3031	23 384 311	70.3%	1.0%
Thinks Medicare drug benefit is easy to understand
Yes	2891	21 168 422	69.3%	0.9%
No	1277	9 372 393	30.7%	0.9%
General health
Good to excellent	3499	25 596 128	77.6%	0.7%
Poor to fair	1041	7 388 147	22.4%	0.7%
Transportation difficulty due to health
Yes	1679	11 492 853	34.8%	0.8%
No	2875	21 578 966	65.2%	0.8%
Functional limitations
No ADLs/IADLs	2491	19 245 973	57.9%	1.0%
IADLs only	769	4 994 313	15.0%	0.7%
1–2 ADLs	843	5 824 280	17.5%	0.6%
3+ ADLs	476	3 195 168	9.6%	0.6%

Abbreviations: ADL, activity of daily living; HS, high school; IADL, instrumental activity of daily living, LIS, Low-Income Subsidy; MAPD, Medicare Advantage prescription drug plan; PDP, standalone prescription drug plan.

Source: Medicare Current Beneficiary Survey (MCBS) 2022 Survey File (unweighted n = 4579; weighted n = 33.2 million). Results weighted to represent the national population of Medicare beneficiaries enrolled in a Part D plan at any point in 2022.

Table 2 presents the prevalence of other MNA and cost-related MNA by different sociodemographic characteristics of Medicare beneficiaries. Overall, 13.7% of beneficiaries reported MNA; 7.5% reported cost-related MNA and 6.2% reported other MNA. This table also presents logistic regression results for other MNA and cost-related MNA, controlling for age, race, gender, education, income, and general health status. A greater proportion of beneficiaries under 65 years old reported other MNA (10.1%) compared with older beneficiaries. Other MNA was more common among beneficiaries with an income-to-poverty ratio of 138% or less compared with higher-income beneficiaries, but the same was not true for cost-related MNA. These differences in other MNA by age and income remained significant after adjustment for sociodemographic characteristics.

Table 2.

Medication nonadherence (other MNA and cost-related MNA) and adjusted odds ratio estimates by sociodemographic data.

	Other MNA					Cost-related MNA
	%	SE of %	OR	95% CI		%	SE of %	OR	95% CI
				Lower	Upper				Lower	Upper
Total	6.2	0.4				7.5	0.5
Age
Under 65 y	10.1	1.3	1.28	0.88	1.85	11.6	1.4	1.44	0.99	2.09
65+ y	5.7	0.5	Ref			7.1	0.5	Ref
Gender
Male	5.6	0.6	Ref			6.4	0.6	Ref
Female	6.5	0.5	1.13	0.89	1.44	8.2	0.6	1.37	1.02	1.83
Race
Non-Hispanic White	6.2	0.5	Ref			7.1	0.5	Ref
Non-Hispanic Black	6.6	1.1	0.68	0.44	1.06	7.9	1.2	1.00	0.65	1.54
Hispanic	6.7	1.4	0.71	0.41	1.23	6.8	1.4	0.93	0.57	1.52
Other/mixed/missing	3.6	1.3	0.45	0.18	1.11	11.2	2.9	1.61	0.83	3.14
Income to poverty ratio
≤138%	9.5	0.9	2.09	1.45	3.01	7.3	0.8	0.65	0.45	0.94
>138%	4.8	0.4	Ref			7.5	0.6	Ref
Educational level
HS graduate or less	6.3	0.7	0.71	0.49	1.04	7.4	0.7	0.9	0.66	1.23
Some college or greater	6.1	0.6	Ref			7.5	0.5	Ref
Metropolitan residence
Metropolitan	6.4	0.5	Ref			7.4	0.5	Ref
Micropolitan/small town/rural	4.9	1.1	0.6	0.35	1.02	7.6	1.2	1.02	0.69	1.5

Abbreviations: HS, high school; MNA, medication nonadherence; OR, odds ratio; Ref, reference.

Source: Medicare Current Beneficiary Survey (MCBS) 2022 Survey File (unweighted n = 4579; weighted n = 33.2 million). Results weighted to represent the national population of Medicare beneficiaries enrolled in a Part D plan at any point in 2022. Multivariate results adjusted for age, race, gender, education, income, and general health status.

Table 3 displays the associations between other and cost-related MNA among beneficiaries, grouped by insurance characteristics. Differences in other MNA between plan benefit type (ie, enhanced and basic plans) were more pronounced than were seen between PDPs and MAPDs; beneficiaries with MAPD basic plans had the highest proportion of other MNA (10.1%), although they were not significantly higher than other plan benefit types after adjustment for sociodemographic characteristics.

Table 3.

Associations between medication nonadherence and health insurance characteristics among Medicare Part D beneficiaries.

	Other MNA					Cost-related MNA
	%	SE of %	OR	95% CI		%	SE of %	OR	95% CI
				Lower	Upper				Lower	Upper
Total	6.2	0.4				7.5	0.5
MAPD/PDP type
MAPD enhanced	6.1	0.7	Ref			6.7	0.6	Ref
MAPD basic	10.1	1.6	1.14	0.71	1.83	5.3	1.0	0.67	0.40	1.10
PDP enhanced	4.9	0.8	0.85	0.52	1.4	7.8	1.0	1.27	0.86	1.87
PDP basic	6.2	1.0	0.76	0.46	1.27	10.4	1.6	1.47	0.95	2.29
Medicare-Medicaid dually eligible
Any dual eligibility for the year	9.9	1.1	1.61	0.96	2.71	6.5	0.9	0.45	0.27	0.75
Not dually eligible	4.9	0.5	Ref			7.8	0.5	Ref
Low-Income Subsidy
Any for the year	10.4	1.1	2.40	1.53	3.76	6.6	0.9	0.43	0.25	0.77
No LIS	4.5	0.4	Ref			7.8	0.5	Ref
Thinks Medicare drug benefit is easy to understand
Yes	5.3	0.4	Ref			5.0	0.5	Ref
No	9.4	1.1	1.72	1.30	2.28	14.3	1.2	3.07	2.29	4.12

Abbreviations: LIS, Low-Income Subsidy; MAPD, Medicare Advantage prescription drug plan; MNA, medication nonadherence; OR, odds ratio; PDP, standalone prescription drug plan; Ref, reference.

Source: Medicare Current Beneficiary Survey (MCBS) 2022 Survey File (unweighted n = 4579; weighted n = 33.2 million). Results weighted to represent the national population of Medicare beneficiaries enrolled in a Part D plan at any point in 2022. Multivariate results adjusted for age, race, gender, education, income, and general health status.

After adjustment for sociodemographic characteristics, beneficiaries who receive the LIS had significantly greater odds of other MNA than beneficiaries who did not receive LIS (odds ratio [OR]: 2.40; 95% CI: 1.53–3.76) but had significantly lesser odds of cost-related MNA than non-recipients (OR: 0.43; 95% CI: 0.25–0.77). In other words, beneficiaries who received the LIS were less likely to report MNA due to cost compared with beneficiaries who did not receive the LIS, but more likely to report MNA due to other reasons.

Beneficiaries who perceived that the Medicare drug benefit was difficult to understand had higher rates of MNA compared with their peers, both for cost-related MNA and other MNA.

Table 4 shows the associations of MNA with transportation, food insecurity, and general health. Transportation difficulties and food insecurity were associated with significantly higher other and cost-related MNA. Beneficiaries who were in worse general health or reported a greater number of functional limitations also saw higher rates of other and cost-related MNA.

Table 4.

Medication nonadherence and adjusted odds ratio estimates by health status and social drivers of health.

	Other MNA					Cost-related MNA
	%	SE of %	OR	95% CI		%	SE of %	OR	95% CI
				Lower	Upper				Lower	Upper
Total	6.2	0.4				7.5	0.5
Transportation difficulty due to health
Yes	9.1	0.9	1.76	1.24	2.5	11.5	1.0	2.05	1.48	2.82
No	4.7	0.5	Ref			5.2	0.5	Ref
Food insecurity
Any food insecurity	11.8	1.2	2.23	1.49	3.33	17.2	1.4	4.40	3.21	6.03
No food insecurity	4.7	0.5	Ref			4.8	0.4	Ref
General health
Good to excellent	5.4	0.5	Ref			5.9	0.5	Ref
Poor to fair	9.4	1.1	1.52	1.10	2.10	13.0	1.5	2.33	1.60	3.40
Functional limitations
No ADLs/IADLs	3.9	0.4	Ref			5.7	0.6	Ref
IADLs only	8.8	1.8	1.92	1.07	3.47	8.4	1.2	1.30	0.81	2.08
1–2 ADLs	9.2	1.4	2.21	1.29	3.80	9.1	1.1	1.45	0.95	2.23
3+ ADLs	11.2	1.8	2.72	1.63	4.53	14.3	2.2	2.05	1.16	3.64

Abbreviations: ADL, activity of daily living; IADL, instrumental activity of daily living; MNA, medication nonadherent; Ref, reference.

Source: Medicare Current Beneficiary Survey (MCBS) 2022 Survey File (unweighted n = 4579; weighted n = 33.2 million). Results weighted to represent the national population of Medicare beneficiaries enrolled in a Part D plan at any point in 2022. Multivariate results adjusted for age, race, gender, education, income, and general health status.

Discussion

This study expands on previous Medicare cost-related MNA research by characterizing both cost-related and other MNA, and found that, while cost-related issues remain a significant source of MNA (7.5% of the weighted population), other sources of MNA (6.2% of the weighted population) also play a large role. The results from unadjusted chi-square tests and the multivariate models showed similar patterns of MNA among beneficiaries. For example, regression results demonstrated a strong association between other MNA and social drivers of health, poor health, and lack of understanding of the Part D benefit. Food insecurity, low income, LIS, and severe functional limitations increased the likelihood of reporting other MNA by more than 2 times. Similar results were found for cost-related MNA, although the magnitude of association was generally higher for cost-related MNA compared with other MNA. The comparison between Part D plan types (MAPD and PDP, basic and enhanced) was inconclusive in assessing plan benefit designs’ role in other MNA. Overall, the results for other MNA suggest that the same groups of beneficiaries experiencing cost-related MNA are also more likely to experience other MNA.

The LIS findings in this study may have important policy significance. Results for cost-related MNA reported by LIS beneficiaries showed no relationship between LIS enrollment and cost-related MNA. These results align with previous research⁵ and were expected, as LIS beneficiaries’ premiums and out-of-pocket drug costs are lowered as part of their Part D benefit. This subsidy program helps LIS beneficiaries afford their medications and those beneficiaries are therefore less likely to cite cost as a barrier to medication adherence.

Results for the other outcome measure (other MNA) were not anticipated, were in the opposite direction, and showed a strong relationship between LIS enrollment and non-cost reasons for MNA: beneficiaries enrolled in the LIS were almost 2.5 times more likely to report other MNA compared with non-LIS beneficiaries. These findings highlight a less studied aspect of MNA, which is the potential for differential impact of cost vs non-cost reasons for MNA on certain populations, and suggest that, specifically for LIS beneficiaries, more focus and attention must be given to identify opportunities to reduce non-cost reasons for MNA.

Limitations

The study has several limitations. Self-reported nonadherence is subject to a beneficiary's recall ability, which may be less accurate over time and as more drugs are prescribed to a particular beneficiary. When comparing overall nonadherence estimates in this study with past and recent estimates, this study may be underestimating current levels of nonadherence.^2-5 Further, dividing specific MCBS measures into cost and other MNA relies on author judgment. Certain reasons for not obtaining a medication as prescribed, such as “Not covered by insurance” or “Obtained/used samples,” were included in the other MNA categorization because those options did not directly reference cost. In addition, the mutually exclusive definition of MNA assigned beneficiaries to a single category, when, in reality, beneficiaries may be experiencing both types of nonadherence (this is another potential source of underestimation of MNA). Still, these decisions to categorize responses were considered necessary to estimate other MNA, which is the primary focus of this study.

Conclusion

Understanding the factors that lead to MNA in the Medicare population is of critical importance to interested parties who want to maximize the value of the Part D benefit. While significant research and attention have focused on cost, this study emphasizes that non-cost factors need to be considered in order to improve adherence among Part D beneficiaries. Further, beneficiaries who are more likely to face cost-related MNA are also more likely to experience nonadherence for other reasons. Several Part D provisions enacted in the IRA (P.L. 117-169) have the potential to address cost-related MNA, including those that reduce out-of-pocket costs like the annual cap on out-of-pocket drug costs (capped at $2000 in 2025). There are also provisions in the IRA with the potential to address other MNA by simplifying the benefit for the beneficiary, including cost-sharing smoothing under the Medicare Prescription Payment Plan and a standard $35 cap on cost-sharing for a month's supply for each covered insulin product.

The proposed M2DL model could also help test ways to address both types of MNA, while drawing on key findings from this study as the model continues to be developed. The M2DL model proposes testing whether a simplified approach to offering low-cost, clinically important generic drugs can improve medication adherence, lead to better outcomes, and improve beneficiary and prescriber satisfaction with the Part D benefit. This study highlights a strong relationship between a beneficiary's lack of knowledge/understanding of the Part D benefit and increases in both forms of nonadherence. The M2DL model aims to decrease nonadherence by simplifying the Part D benefit for both beneficiaries and prescribers, setting a fixed, low-copay (up to $2 per month supply) across all cost-sharing phases of the Part D drug benefit. This study also highlights an opportunity to improve adherence by focusing on non–cost-related reasons for nonadherence (other MNA), such as beneficiary preferences or beliefs (including trust in provider and effectiveness of the drug), understanding of their health situation or insurance coverage, and ability to fill a prescription—all areas that may be positively impacted by the M2DL model intervention. Finally, lessons learned from this study about the characteristics of beneficiaries more likely to face nonadherence can also inform model design. For example, the M2DL model may consider including LIS beneficiaries as there still appear to be non–cost-related opportunities to improve adherence in this population, as well as vulnerable populations.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

The content of this article was not developed under grants. Data were collected under contract with NORC at University of Chicago (contract: 75FCMC19D0092; task order: 75FCMC21F0001).