Citizenship status and cost‐related nonadherence in the United States, 2017–2021

Abstract

Objective

To assess inequities in prescription medication use and subsequent cost‐related nonadherence (CRN) and cost‐saving strategies by citizenship status in the United States.

Data Sources/Study Setting

National Health Interview Survey (2017–2021).

Study Design

This cross‐sectional study examined noncitizen (n = 8596), naturalized citizen (n = 12,800), and US‐born citizen (n = 120,195) adults. We also examined older adults (≥65 years) separately, including noncitizens without Medicare (a group of importance given their immigration‐related barriers to health care access). Multiple mediation analysis was used to examine differences in CRN and determine whether economic, health care, and immigration factors explain observed inequities.

Principal Findings

Noncitizens (41.9%) were less likely to use prescription medications than naturalized (60.5%) and US‐born citizens (68.2%). Among prescription medication users, noncitizens (13.8%) were more likely to report CRN than naturalized (9.5%) and US‐born citizens (11.0%). CRN differences between noncitizens and naturalized citizens (OR 1.38, 95% CI 1.21–1.44) and between noncitizens and US‐born citizens (OR 1.23, 95% CI 1.07–1.35) were explained by insurance status and food insecurity. Only 4.9% of medication users turned to alternative therapies to lower their drug costs, but there were no substantial differences across citizenship status. More medication users requested lower‐cost prescriptions (19.0%); however, noncitizens were less likely to make these requests. Older noncitizens without Medicare, of whom 23.9% requested lower‐cost drugs, were an exception. Noncitizens (5.8%), particularly older noncitizens without Medicare (21.8%), were more likely to import their drugs than naturalized (3.5%) and US‐born citizens (1.2%).

Conclusions

Noncitizens experience a high burden of cost‐related barriers to prescription medications. Efforts to reduce these inequities should focus on dismantling health care and food access barriers, regardless of citizenship status.

What is known on this topic

• Due to federal and state policies that systematically exclude certain immigrants from health care access, noncitizens disproportionately encounter cost‐related barriers to health care, including prescription medications.

• While noncitizens are less likely to use clinically indicated medications, little is known regarding inequities in cost‐related nonadherence (CRN) and cost‐saving strategies (e.g., drug importation) by citizenship status.

What this study adds

• Noncitizens are less likely to use prescription medications than naturalized and US‐born citizens.

• Noncitizens are also more likely to experience CRN; this inequity was explained by differences in insurance status and food security.

• Noncitizens were more likely to attempt to mitigate their cost barriers by importing potentially unsafe and ineffective drugs from abroad.

1. INTRODUCTION

In 2019, there were approximately 42.4 million immigrants (i.e., foreign‐born) adults living in the United States, including 19.8 million noncitizens. ¹ Due to federal and state policies that systematically exclude certain immigrants from formal labor and safety‐net programs, ² , ³ noncitizens–including documented (lawfully present) and undocumented immigrants–disproportionately encounter cost‐related barriers to health care, ⁴ including prescription medications. ⁵ , ⁶ For example, undocumented immigrants, who account for half of all noncitizens, ⁷ are excluded from many jobs that offer employer‐based coverage and are barred from most publicly funded insurance programs (i.e., Medicare, Medicaid, and Affordable Care Act [ACA] Marketplace coverage) and other safety‐net programs (e.g., Temporary Assistance for Needy Families). ⁴ Many documented immigrants are also ineligible for public insurance because eligibility depends on visa status, duration of residence, and state of residence. ⁴ Compared to US‐born and naturalized citizens (immigrants who have gained US citizenship via naturalization), noncitizens also disproportionately experience adverse social determinants of health, such as poverty and food insecurity, ⁸ , ⁹ , ¹⁰ that may impact their ability to access or afford essential health care. Therefore, citizenship status may contribute to health care barriers, including inequitable access to prescription medications.

Although prescription medications are critical for the prevention and management of chronic conditions, noncitizens are less likely to use clinically indicated drugs. ⁵ , ⁶ Medication nonadherence is a major concern because 30%–50% of Americans prescribed medications do not adhere to their regimens, ¹¹ , ¹² , ¹³ which is associated with adverse health outcomes, including an increased risk of death. ¹⁴ , ¹⁵ While medication nonadherence is complicated and multifaceted, ¹¹ cost‐related nonadherence (CRN) is an important problem. One in 10 prescription medication users report CRN and there are substantial inequities associated with economic and health care barriers. ¹⁶ , ¹⁷ , ¹⁸ However, it is not known whether noncitizens, who are more likely to experience these barriers, also disproportionately report CRN.

Noncitizens may be more likely to turn to cost‐saving strategies because of barriers in the affordability of health care and prescription medications. Cost‐saving strategies may include behaviors such as requesting lower‐cost prescriptions, taking alternative therapies (e.g., dietary supplements), and importing drugs from abroad. However, these strategies may not be feasible nor clinically advisable. For example, lower‐cost prescriptions may not be available or clinically appropriate for specific conditions (e.g., novel cancer regimens). ¹⁹ While many source countries for imported drugs have high regulatory standards (including Canada and Mexico), ²⁰ , ²¹ , ²² individuals may obtain medications from countries without appropriate regulations or unregulated pharmacies. ²³ Furthermore, without clinical oversight of alternative therapies or drugs obtained from abroad, individuals may have worse outcomes due to inappropriate self‐medication and inadequate follow‐up with a health care professional.

While inequities in CRN and the use of cost‐saving strategies may adversely impact the health of immigrants, there is limited information about these medication access barriers by citizenship status. Understanding these inequities is critical considering the number of policies that exclude immigrants, especially noncitizens, from health care and medication access. ²⁴ , ²⁵ This study may also inform new and proposed municipal and state public insurance programs designed to bridge the gap in health care access between noncitizens and citizens in certain areas. ⁴

Using data from the National Health Interview Survey (NHIS), we assessed inequities in prescription medication use and, among those taking medications, CRN and cost‐saving strategies by citizenship status. We also evaluated which factors contribute to CRN inequities across citizenship status using mediation analysis. We hypothesized that CRN is more common among noncitizens due to worse economic status and health care access.

2. METHODS

2.1. Study sample

NHIS is a nationally representative cross‐sectional interview survey that uses a multistage probability survey design to select samples of dwelling units. ²⁶ We used data from the 2017 to 2021 adult sample (≥18 years) (obtained from IPUMS). ²⁷ Although NHIS is generally conducted face‐to‐face, in March 2020 the survey shifted to telephone interviews due to the Coronavirus disease 2019 (COVID‐19) pandemic. ²⁸ Surveys were conducted in English or Spanish; interpreters were used when necessary.

We excluded participants who did not report their place of birth or citizenship status (n = 2733) or had missing indicators for prescription medication use (n = 616) or CRN (n = 29) (Figure S1). Our analytic sample had 141,969 participants (noncitizens: 8596; naturalized citizens: 12,800; US‐born citizens: 120,195), including 98,929 prescription medication users (noncitizens: 3751; naturalized citizens: 8142; US‐born citizens: 87,028).

2.2. Measures

We examined several outcomes to characterize medication access, including prescription medication use, CRN, and the use of cost‐saving strategies (Table S1 summarizes the questionnaire items used). First, we described self‐reported prescription medication use in the last year. Then, we examined CRN – operationalized as an affirmative response to items asking whether participants needed but could not afford medication in the previous year or, if to save money, they had skipped medication doses, taken less medicine than prescribed, or delayed filling their prescriptions – among those who take prescription medications. Finally, we examined cost‐saving strategies among prescription medication users based on whether participants asked their clinicians for lower‐cost medications, used alternative therapies, or bought prescription drugs from another country to save money in the last year (collected during 2017–2018).

Citizenship status, the exposure, was determined by the questions: (1) Were you born in the United States or a US territory? and (2) Are you a citizen of the United States? ²⁹ We categorized citizenship status as US‐born citizen (born in the US or to US‐born parents), naturalized citizen (foreign‐born and reports US citizenship), and noncitizen (foreign‐born and does not report US citizenship). Fewer than 1% of adults sampled refused to report this information.

We examined inequities among Latinxs separately because they constitute the plurality of naturalized citizens and noncitizens in the US. ¹ We also examined older adults (≥65 years) separately; focusing on noncitizens without Medicare. Older noncitizens may be excluded from premium‐free Medicare part A because they (1) lack documentation to reside in the US (undocumented immigrants are prohibited from accessing federally‐funded insurance), (2) hold a non‐eligible documentation status (i.e., are not lawful permanent residents or “green card” holders), (3) have not held a permanent residency permit for at least 5 years, or (4) have not paid Medicare taxes long enough to qualify for the program (usually at least 10 years) due to their current or past documentation status or their duration of residence in the US. ⁴ Nearly all older citizens report Medicare access (97%), while only 81% of noncitizens report access. ¹

To determine if the citizenship status‐related inequities observed during the 2017–2021 study period also persisted during the COVID‐19 pandemic, we reported the prevalence of prescription medication use and CRN from 2020 to 2021 as a sensitivity analysis.

Important clinical factors included age, sex, and number of self‐reported chronic diseases (asthma/ chronic obstructive pulmonary diseases, arthritis, high cholesterol, hypertension, diabetes, kidney disease, liver disease, cardiovascular disease [coronary heart disease, myocardial infarction, or stroke], and cancer). Economic factors included education (< or ≥ high school education), poverty (household income below poverty line), and low food security (30‐day, 10‐item food security score ≥5). ³⁰ , ³¹ Health care access was based on reporting a usual health care provider and insurance coverage. Immigration factors included foreign‐language preference (responded to interview in Spanish, collected during 2017/2018), and years in US. We also reported race/ethnicity across citizenship status.

2.3. Statistical analysis

Descriptive statistics were used to estimate the prevalence of prescription medication use and, among prescription medication users, the prevalence of CRN and use of cost‐saving strategies. Differences were determined using Pearson χ² tests. Descriptive statistics and statistical tests were weighted using Taylor linearization methods to incorporate sample weights that adjust for differential probability of selection and nonresponse.

Logistic regressions were used to evaluate the associations between citizenship status and CRN to enable mediation analyses which are described below. We chose to forgo survey weights in examining the associations because our mediation analysis method cannot incorporate weighting. However, the magnitude and directionality of the univariate association between citizenship status and CRN were very similar regardless of modeling choice or survey weights (Table S2), suggesting these choices may lead to minimal differences in measures of association in this study.

We conducted mediation analyses to examine which factors explain the association between citizenship status and CRN. Our conceptual framework considered economic, health care, and immigration‐related factors to be on the causal pathway between citizenship status and CRN (Figure 1). ³² We, therefore, modeled these variables as potential mediators. Informed by the Institute of Medicine's definition of health care disparities, or differences not justified by clinical characteristics or patient preferences, ³² , ³³ , ³⁴ clinical factors were considered potential confounders. Race/ethnicity, a fundamental cause of health inequities, ² was not considered a confounder or mediator of the association between citizenship status and CRN. Because citizenship status varies across racial/ethnic groups (primarily due to structural racism, where the immigration system discriminates against certain groups more than others), ² , ³ citizenship status would be considered a mediator of the known association between race/ethnicity and CRN. ¹⁶ , ¹⁷

FIGURE 1.

Path diagram demonstrating proposed total effect of citizenship status on cost‐related nonadherence (CRN).

Performing formal mediation analysis allows the identification of factors that may explain the relationship between citizenship status and CRN. In addition to conceptual considerations, factors may mediate this relationship if the following conditions are met: (1) potential mediators must be associated with the outcome (CRN) after adjusting for the exposure (citizenship status), and (2) potential mediators must be associated with the exposure (citizenship status). ³⁵ , ³⁶ If neither condition was met, it was not considered a mediator nor included in the final model. For more information, see Table S3, which presents which health care, economic, and, where appropriate, immigration factors (i.e., comparisons between naturalized citizens and noncitizens) were considered mediators based on the conditions above.

We used the multiple mediation analysis method based on the counterfactual framework proposed by Yu and Li. ³⁵ , ³⁶ This method was implemented using the R mma package. ³⁷ We used a generalized method with multiple additive regression trees. ³⁷ This approach was selected because it can measure mediating effects for multiple potential mediators; is flexible in terms of the types of outcome, exposure, mediator, and confounding variables that can be included in the model; and can better account for complex exposure‐mediator effects and multicollinearity than linear methods. ³⁵ , ³⁶ , ³⁷ We reported the total, direct, and indirect effects of citizenship status on CRN through odds ratios (OR) and relative effect sizes (RE percentages) using 100 bootstrap iterations to accommodate uncertainty. ³⁷ , ³⁸

Previous studies have demonstrated that compared to US‐born citizens, noncitizens—but not naturalized citizens—have worse access to prescription medications. ⁵ , ⁶ Therefore, in our mediation analysis, we compared noncitizens to naturalized and US‐born citizens (i.e., both citizen groups serving as reference groups). Our mediation analysis focused on all racial/ethnic groups in aggregate; however, given known inequities in medication use associated with citizenship status among Latinxs, ⁵ , ⁶ we replicated our mediation analysis among Latinx adults.

p values were 2‐sided. Analyses were completed using R version 4.2.2. As this study relied on de‐identified public data, Institutional Review Board approval was not required (45 CFR ×46.102[f]).

3. RESULTS

In this national sample, 8.0% were noncitizens, 10.1% were naturalized citizens, and 81.9% were US‐born citizens (representing 19.4, 24.6, and 199.7 million adults, respectively) (Table 1). Compared to naturalized and US‐born citizens, noncitizens were younger and reported fewer chronic conditions; were more likely to lack a high school education, live in poverty and have low food security; and had less access to a usual source of care and health insurance. Most immigrants were people of color; Latinxs alone constituted 60.2% of noncitizens and 34.8% of naturalized citizens. Sixty‐three percent of noncitizens have lived in the US for at least 10 years compared to 95.2% of naturalized citizens. Noncitizens were also more likely to have limited English proficiency than naturalized citizens. We observed similar patterns among older adults (Table S4) and prescription medication users (Table S5).

TABLE 1.

Characteristics of adults in the United States, by citizenship status.

	Characteristics a % (CI)
	Total	Citizenship status b
		US‐born citizen	Naturalized citizen	Noncitizen
US population, no.	243,746,425	199,678,148	24,629,214	19,439,063
Sample, no.	141,591	120,195	12,800	8596
Overall	100.0	81.9 (81.7–82.2)	10.1 (9.9–10.3)	8.0 (7.8–8.2)
Clinical factors
Age‐y
Median (IQR)	47.0 (32.0–62.0)	47.0 (31.0–63.0)	52.0 (40.0–64.0)	40.0 (31.0–51.0)
<44	45.9 (45.6–46.2)	45.7 (45.4–46.1)	34.4 (33.4–35.5)	62.3 (61.0–63.5)
45–64	33.0 (32.7–33.3)	32.1 (31.8–32.4)	42.1 (41.0–43.2)	30.4 (29.3–31.6)
≥65	21.1 (20.9–21.4)	22.2 (21.9–22.5)	23.5 (22.6–24.3)	7.3 (6.7–7.9)
Men	48.3 (48.0–48.7)	48.4 (48.1–48.8)	46.5 (45.4–47.6)	50.0 (48.7–51.3)
Chronic conditions c
0	43.9 (43.6–44.2)	42.0 (41.6–42.4)	44.2 (43.2–45.3)	63.1 (61.9–64.4)
1	23.0 (22.7–23.2)	23.2 (22.9–23.5)	23.8 (22.9–24.7)	19.6 (18.6–20.7)
≥2	33.1 (32.8–33.4)	34.8 (34.5–35.2)	32.0 (31.0–33.0)	17.3 (16.3–18.3)
Economic factors
<High school	10.3 (10.1–10.6)	7.8 (7.6–8.0)	13.9 (13.1–14.7)	31.7 (30.4–32.9)
<1.0 poverty‐to‐income ratio	10.5 (10.3–10.7)	9.4 (9.2–9.6)	9.8 (9.2–10.4)	22.4 (21.3–23.6)
Low food security	7.9 (7.7–8.1)	7.6 (7.4–7.8)	6.9 (6.3–7.4)	12.1 (11.2–13.0)
Health care factors
No usual source of care	12.8 (12.5–13.0)	11.9 (11.6–12.1)	9.8 (9.2–10.5)	26.0 (24.9–27.2)
Health insurance
Private	69.5 (69.2–69.8)	71.9 (71.6–72.2)	69.2 (68.2–70.2)	45.2 (43.9–46.4)
Public alone	20.8 (20.5–21.1)	20.5 (20.2–20.8)	23.3 (22.4–24.2)	20.4 (19.3–21.5)
Uninsured	9.7 (9.5–9.9)	7.6 (7.4–7.8)	7.5 (6.9–8.1)	34.5 (33.2–35.8)
Demographic & immigration factors
Race/ethnicity
White	63.6 (63.2–63.9)	73.6 (73.2–73.9)	23.4 (22.6–24.3)	11.4 (10.6–12.1)
Black	11.6 (11.4–11.8)	12.2 (11.9–12.4)	10.8 (10.1–11.5)	6.7 (6.0–7.5)
Latino	16.3 (16.0–16.6)	9.8 (9.5–10.0)	34.8 (33.8–35.8)	60.2 (58.9–61.4)
Asian	5.9 (5.7–6.0)	1.5 (1.4–1.6)	29.3 (28.3–30.3)	20.8 (19.8–21.8)
≥10 y in United States d	–	–	95.2 (94.7–95.7)	62.9 (61.6–64.2)
Limited english proficiency e	6.0 (5.7–6.3)	1.2 (1.1–1.4)	18.9 (17.4–20.5)	39.3 (37.2–41.5)

Abbreviations: CI, confidence interval; IQR, interquartile range; no., number; US, United States; y, years.

Note: Data from the National Health Interview Survey (2017–2021).

^a Estimates were weighted.

^b All differences across citizenship status were statistically significant (p < 0.01).

^c Self‐reported history of asthma/ chronic obstructive pulmonary disease, arthritis, high cholesterol, hypertension, diabetes, kidney disease, liver disease, cardiovascular disease, or cancer.

^d Among foreign‐born adults (naturalized citizens or noncitizens).

^e Collected from 2017 to 2018.

Noncitizens were less likely to use prescription medications than naturalized and US‐born citizens (41.9%, 60.5%, and 68.2%, respectively) (Figure 2). Similar inequities across citizenship status were observed among Latinxs and across age groups. While most older adults used prescription medications, older noncitizens without Medicare reported substantially lower use.

FIGURE 2.

Prescription medication use among US adults, by citizenship status. [Color figure can be viewed at wileyonlinelibrary.com]

Among prescription medication users, noncitizens were more likely to experience CRN than naturalized and US‐born citizens (13.8%, 9.5%, and 11.0%, respectively) (Table 2). These inequities were evident in the Latinx population and across age groups. However, the magnitude of this inequity was largest among older adults, where noncitizens, especially those without Medicare, were more likely to report CRN than naturalized and US‐born citizens. In other words, a disproportionate number of older noncitizens without Medicare did not get their medications (18.4%), skipped doses (8.9%), took fewer medications (10.2%), and delayed filling their prescriptions (9.0%) due to costs (Table S6). CRN inequities by citizenship status were also evident among patients with at least one chronic condition (Table S7), where noncitizens were more likely to report CRN than naturalized and US‐born citizens (15.4%, 9.9%, and 12.0%, respectively).

TABLE 2.

Cost‐related nonadherence (CRN) and cost‐saving strategies among adult prescription medication users, by citizenship status.

Race/ethnicity	Age group‐y	CRN or cost‐saving strategy‐% (CI) a					p
		Overall	Citizenship status
			US‐born citizen	Naturalized citizen	Noncitizen	Noncitizens ≥65 y without medicare b
CRN c
All races/ethnicities	All ages	11.0 (10.8–11.3)	11.0 (10.8–11.3)	9.5 (8.6–10.3)	13.8 (12.4–15.2)	–	<0.01
	18–64	12.8 (12.5–13.1)	12.9 (12.5–13.2)	11.1 (10.0–12.2)	14.6 (13.0–16.2)	–	<0.01
	≥65	6.6 (6.3–6.9)	6.6 (6.2–6.9)	6.1 (5.1–7.1)	8.9 (6.3–11.6)	22.8 (13.0–32.6)	0.01
Latinx	All ages	14.2 (13.3–15.1)	14.0 (12.8–15.2)	11.6 (10.0–13.3)	17.4 (15.3–19.4)	–	<0.01
	18–64	15.5 (14.4–16.5)	14.9 (13.5–16.3)	13.7 (11.6–15.9)	18.2 (15.9–20.5)	–	0.02
	≥65	8.7 (7.3–10.1)	9.1 (7.1–11.1)	6.9 (4.9–8.8)	12.0 (7.7–16.4)	26.0 (12.9–39.1)	0.03
Asked clinicians for lower costs medications d
All races/ethnicities	All ages	19.0 (18.5–19.5)	19.8 (19.2–20.3)	14.5 (12.8–16.2)	14.0 (11.8–16.1)	–	<0.01
	18–64	19.4 (18.8–20.1)	20.3 (19.6–20.9)	15.5 (13.4–17.6)	13.6 (11.3–16.0)	–	<0.01
	≥65	17.8 (17.0–18.5)	18.5 (17.6–19.3)	12.2 (9.8–14.6)	16.2 (10.6–21.8)	23.9 (3.0–44.8)	<0.01
Latinx	All ages	17.0 (15.4–18.6)	17.5 (15.4–19.6)	15.5 (12.2–18.7)	17.5 (14.2–20.8)	–	0.56
	18–64	17.5 (15.7–19.3)	17.7 (15.3–20.1)	16.3 (12.2–20.4)	18.1 (14.5–21.8)	–	0.79
	≥65	14.7 (11.8–17.7)	16.4 (12.3–20.4)	13.3 (8.3–18.3)	13.5 (5.7–21.2)	35.7 (6.4–65.0)	0.18
Alternative medicine use d
All races/ethnicities	All ages	4.9 (4.6–5.2)	4.9 (4.6–5.2)	4.8 (3.8–5.8)	4.8 (3.5–6.2)	–	0.98
	18–64	5.9 (5.5–6.3)	6.0 (5.6–6.3)	5.7 (4.4–7.0)	5.2 (3.6–6.7)	–	0.61
	≥65	2.2 (1.9–2.5)	2.1 (1.8–2.5)	2.7 (1.6–3.8)	2.7 (0.5–4.8)	3.5 (0.0–9.5)	0.75
Latinx	All ages	5.3 (4.4–6.3)	5.2 (4.0–6.5)	5.0 (3.1–6.9)	6.0 (3.8–8.1)	–	0.79
	18–64	6.0 (4.8–7.1)	5.9 (4.5–7.3)	5.7 (3.2–8.1)	6.4 (3.9–8.8)	–	0.92
	≥65	2.5 (1.3–3.7)	1.5 (0.4–2.7)	3.3 (0.9–5.7)	3.4 (0.3–6.4)	5.2 (0.0–14.5)	0.44
Drug importation d
All races/ethnicities	All ages	1.7 (1.5–1.8)	1.2 (1.1–1.4)	3.5 (2.5–4.4)	5.8 (4.4–7.3)	–	<0.01
	18–64	1.6 (1.3–1.8)	1.1 (0.9–1.2)	3.7 (2.4–5.0)	5.6 (4.0–7.1)	–	<0.01
	≥ 65	1.9 (1.6–2.2)	1.6 (1.3–1.9)	2.9 (1.6–4.2)	7.4 (3.7–11.0)	21.8 (2.8–40.7)	<0.01
Latinx	All ages	4.0 (3.1–5.0)	1.8 (0.8–2.8)	5.9 (3.6–8.3)	7.7 (5.4–10.0)	–	<0.01
	18–64	4.1 (3.0–5.1)	1.9 (0.8–3.1)	6.4 (3.4–9.4)	7.2 (4.8–9.7)	–	<0.01
	≥65	3.9 (2.3–5.5)	1.1 (0.2–2.0)	4.7 (1.7–7.7)	10.4 (4.0–16.8)	28.8 (1.8–55.8)	<0.01

Abbreviations: CI, confidence interval; CRN, cost‐related nonadherence; US, United States; y, years.

Note: Data from the National Health Interview Survey (2017–2021).

^a Estimates were weighted. Differences were tested using χ² tests.

^b Older adults without Medicare are a group of special interest because they experience several immigration‐related exclusions to Medicare, a near‐universal program.

^c Collected from 2017 to 2021.

^d Collected from 2017 to 2018.

In unadjusted regressions, compared to US‐born citizens, naturalized citizens (OR 0.82, 95% CI 0.75–0.88) were less likely to report CRN while noncitizens were more likely to do so (OR 1.23, 95% CI 1.07–1.35) (Table S8). Other characteristics that were significantly associated with a higher risk of CRN in unadjusted regressions included younger age, male gender, reporting multiple chronic conditions, lacking a high school education, living in poverty, having low food security, lacking a usual source of care, having no health insurance, being Black or Latinx, and having limited English proficiency.

In our mediation analysis (Table 3), the 23% higher odds of CRN among noncitizens compared to US‐born citizens was explained by economic and health care factors, including health insurance and food security. If these factors were equal across groups (direct effect), CRN would be slightly lower among noncitizens compared to US‐born citizens (OR 0.89, 95% CI 0.78–0.98). The 38% higher odds of CRN among noncitizens compared to naturalized citizens was also explained by economic and health care factors, including health insurance, food security, usual source of care, and poverty. If these factors were equal across groups, CRN among noncitizens and naturalized citizens would be equivalent (OR 1.00, 95% CI 0.93–1.05). Duration of residence in the US did not significantly explain observed CRN differences between noncitizens and naturalized citizens. These findings were similar among Latinxs (Table S9).

TABLE 3.

Association between citizenship status and cost‐related nonadherence among prescription medication users: estimates from multiple mediation models.

Mediator a	OR (95% CI) b	RE, % (95% CI) c
Noncitizen versus US‐born citizen (ref.)
Total effects	1.23 (1.07–1.35)	–
Direct effects	0.89 (0.78–0.98)	−55.9 (−289.6 to 88.5)
Indirect effects	1.37 (1.31–1.44)	155.9 (11.5 to 389.6)
Low food security	1.08 (1.06–1.12)	37.4 (2.2 to 102.9)
No usual source of care	1.02 (1.02–1.04)	11.0 (−0.7 to 33.3)
Uninsured	1.23 (1.18–1.27)	102.5 (1.8 to 257.9)
Noncitizen versus naturalized citizen (ref.)
Total effects	1.38 (1.21–1.44)	–
Direct effects	1.00 (0.93–1.05)	0.1 (−31.3 to 19.8)
Indirect effects	1.38 (1.26–1.42)	99.9 (80.2 to 131.3)
<1.0 poverty‐to‐income ratio	1.02 (1.00–1.03)	6.2 (0.1 to 11.1)
Low food security	1.07 (1.04–1.10)	22.2 (13.6 to 32.7)
No usual source of care	1.02 (1.01–1.04)	6.5 (0.9 to 16.3)
Uninsured	1.23 (1.17–1.27)	65.8 (50.5 to 95.7)
≥10 y in United States	1.00 (0.96–1.01)	−0.6 (−14.5 to 4.9)

Abbreviations: CI, confidence interval; OR, odds ratio; Ref., reference; RE, relative effect; US, United States; y, year.

Note: Data from the National Health Interview Survey (2017–2021).

^a Potential mediators included economic, health care, and immigration factors, as relevant. Only the indirect effects of potential mediators that fulfilled each of the following conditions are included: (1) variable is associated with the outcome (CRN) after adjusting for the exposure (citizenship status); (2) variable is associated with the exposure (citizenships status) (see Table S3).

^b Models were adjusted for clinical factors (age, sex, and number of chronic conditions), which were considered confounders, based on the Institute of Health definition of “health care disparities”.

^c Can be interpreted as proportion of the total effect explained by mediators. The sum of relative effects of the individual mediators may not equal the total indirect effect because of correlation and overlapping mediation effects among mediators that is reflected in the total indirect effect but not the individual mediators.

Five percent of prescription medication users turned to alternative therapies and there were no significant differences across citizenship status (Table 2). Approximately 20% of prescription medication users requested lower‐cost prescriptions but noncitizens were less likely to make these requests. Older noncitizens without Medicare, of whom 23.9% requested lower‐cost drugs, were an exception, noncitizens were more likely to import their medications than naturalized and US‐born citizens (5.8%, 3.5%, and 1.2%, respectively). These inequities were most pronounced among older adults, where older noncitizens, especially without Medicare, were much more likely to import their drugs than their naturalized and US‐born citizen counterparts.

In our sensitivity analysis (Table S10), we found that noncitizens were also less likely to use prescription medications than naturalized and US‐born citizens during the COVID‐19 pandemic (41.3%, 61.3%, and 68.1%, respectively). Among prescription medication users, reports of CRN were more common among noncitizens than among naturalized and US‐born citizens during the pandemic (13.9%, 8.3%, and 9.5%, respectively). Similar inequities were found among Latinxs and across age groups.

4. DISCUSSION

Despite the critical public health importance of medication access, we found that noncitizens were less likely to use prescription drugs and more likely to report CRN than naturalized and US‐born citizens. These inequities were most pronounced among older adults and persisted in the COVID‐19 era. In a mediation analysis designed to elucidate which factors explain CRN inequities across citizenship status, we found that inequities between noncitizens and citizens are explained by differences in insurance status, followed by food insecurity, and (to a lesser extent) access to a usual source of care. This study demonstrates that structural barriers resulting in inequitable health care and food access lead to cost‐related underuse of medications that disproportionately affect noncitizens.

Even among those who take prescription medications, 14% of noncitizens report CRN in the past year versus 9.5% of naturalized and 11% of US‐born citizens. Inadequate medication access–including higher rates of CRN–is associated with adverse health outcomes, ranging from suboptimal control of cardiovascular risk factors to increased risk of death, ¹⁴ , ¹⁵ , ³⁹ and may drive health inequities by citizenship status. For example, a recent study found that young Latinx noncitizens are at 40% higher risk of death than their US‐born counterparts. ⁴⁰ Given the magnitude of CRN inequities associated with citizenship status among nonelderly Latinxs, CRN may contribute to existing mortality inequities.

Because approximately 90% of older adults use prescription medications, CRN is a major concern in this population. Medicare, a near‐universal federal insurance program for older adults, notably excludes many noncitizens, including all undocumented immigrants and many documented immigrants. ⁴ Our sample exemplifies this exclusion, where more than one in 10 older noncitizens were uninsured, including half of noncitizens without Medicare, versus less than 1% of citizens. Consequently, the magnitude of CRN inequities was larger among older adults, where 9% of noncitizens, including nearly a quarter of those without Medicare, report CRN versus 6.5% of citizens. It is crucial to address these inequities because most noncitizens reach late adulthood in the US, ⁴¹ when their risk of chronic conditions requiring pharmacologic treatment increases substantially.

One in five prescription medication users turned to cost‐saving strategies to manage the high cost of medications. With little more than anecdotal evidence, immigrants–especially Latinxs–are thought to disproportionately use alternative therapies due to cultural differences and health care barriers. ⁴² , ⁴³ However, we found that alternative therapies are rarely used as a cost‐saving strategy (<5%) and that there were no differences by citizenship status.

Nearly 20% of prescription medication users ask their clinicians for lower‐cost drugs. However, noncitizens were less likely to ask than US‐born citizens. Without a usual source of care, it may be more challenging to request medication changes, even among those with health insurance. Therefore, this inequity may stem from health care inequities, where a quarter of noncitizens lack a usual source of care compared to approximately 10% of citizens. Nonetheless, lower‐cost medications are not always feasible, especially to treat cancer or chronic conditions where titration with expensive therapies is common (e.g., diabetes). Except for therapeutic substitution (e.g., changing the prescription to a generic version or another drug in the same class), it is unjust to relegate those without health care access, including noncitizens, to medications that may be less clinically appropriate.

Drug importation has been hailed as a solution–or at least a stopgap–for the problems of increasing drug prices ⁴⁴ and increasingly inaccessible medications ⁴⁵ in the US. Yet less than 2% of Americans, including 6% of noncitizens, import medications to reduce drug costs. Drug importation also exists in a legal gray zone, where only certain medications from Canada can be legally imported by US residents. ²² More concerningly, there are safety issues with drug importation. Due to lackluster enforcement of good manufacturing practices and bioequivalence standards, substandard and falsified medications constitute more than 10% of the global drug supply and upward of 70% of drugs in certain countries in Africa and Asia. ²⁰ Even in countries with robust enforcement, substandard medications often enter the supply chain in retail settings, including via unregulated online pharmacies. ⁴⁶ The safest way to obtain medications from abroad may be to travel across the Northern or Southern US borders and visit a licensed retail pharmacy abroad. Ironically, noncitizens, especially those without documentation, may be less able to physically import medications in this manner due to their tenuous immigration status. Instead, noncitizens may disproportionately rely on online pharmacies and local resellers to obtain imported drugs, potentially increasing their risk of obtaining substandard and falsified drugs through poorly regulated channels. ⁴⁶ Finally, without clinical oversight and follow‐up, individuals who import drugs may self‐prescribe inappropriate medications, fail to manage their health conditions, and increase their risk of drug–drug or drug‐supplement interactions.

Relying on cost‐saving strategies to manage cost‐related medication barriers will continue to entrench existing inequities because there is little evidence that using these individual‐level strategies reduces nonadherence. Instead, this study suggests that CRN inequities by citizenship status were primarily attributable to differences in health insurance and food security, and that addressing these barriers may reduce cost‐related barriers to medications.

Food insecurity is associated with CRN because individuals often forgo medications to cover other necessities, including food. ⁴⁷ , ⁴⁸ , ⁴⁹ Noncitizens may be especially vulnerable to food insecurity, and subsequent CRN because they are more likely to experience poverty and are excluded from safety‐net programs designed to address food access. ⁸ , ⁹ , ¹⁰ For example, the Supplemental Nutrition Assistance Program (SNAP), a federal program that provides a monthly allowance to low‐income households to purchase food with an electronic card, is known to alleviate hunger and reduces CRN by lowering the out‐of‐pocket food cost (therefore, allowing individuals to afford their medications). ⁴⁷ , ⁴⁸ However, income‐eligible households with noncitizen members are less likely to participate in SNAP because only citizens and some documented noncitizens are eligible (notably, all undocumented immigrants and most lawful permanent residents who have held their permit for <5 years are ineligible). ⁸ , ⁵⁰ , ⁵¹ Even when mixed‐status households (members with differing citizenship and documentation statuses) participate in SNAP, ⁵¹ they may receive fewer benefits because ineligible members do not count towards the monthly food allowance.

In recent years, federal policies have continued to exclude certain noncitizens from access to programs designed to alleviate barriers to health care and food access. For example, the ACA explicitly excludes undocumented immigrants from enrollment in Marketplace plans and blocks states from using federal funds to cover undocumented immigrants in their Medicaid programs. ⁴ As a result, inequities in insurance coverage by citizenship status have persisted in the last decade, ⁵² aggravating inequities in prescription medication access.

The former Trump administration expanded the Public Charge Rule, which allows the government to refuse lawful permanent resident permits or visa applications for documented immigrants who are determined to be dependent on public assistance. ⁵³ Until 2019, the government considered cash‐based public assistance to determine if an applicant was a “public charge”, but the expansion allowed the government to consider most federal, state, and local assistance programs, including Medicaid, subsidized health insurance purchased through the ACA Marketplace, and SNAP. ⁵³ While the Biden administration rescinded this expansion in 2022, ⁵³ it is thought that 1000s of immigrants still forgo safety‐net health care and food assistance programs due to lingering fears. ⁸ , ⁵⁰ , ⁵⁴ Therefore, policies that explicitly or implicitly exclude or discourage noncitizens from publicly funded safety‐net programs, even temporarily, may worsen existing inequities in prescription medication access.

States and localities have a critically important role in promoting equitable health care access and improving access to prescription medications. For example, states should consider expanding Medicaid to income‐eligible adults (regardless of citizenship status) using their funds. Recently, California, Illinois, and Oregon expanded their Medicaid rolls to include undocumented older adults. ⁴ In 2022, California announced plans to expand its Medicaid program to all low‐income residents, including undocumented immigrants, by 2024. ⁵⁵ States can also use their funds to improve food access among noncitizens, especially those excluded from federally funded programs; doing so may reduce hunger and improve access to medications. ⁴⁷ , ⁴⁸ In 2023, California became the first state to expand its equivalent of SNAP (CalFresh) to undocumented immigrants older than 55. ⁵⁶ Cities and counties can also enact health programs that provide comprehensive care (including prescription drug coverage) to all income‐eligible noncitizens, as the governments of Washington DC and New York City have done. ⁵⁷ , ⁵⁸

5. STRENGTHS AND LIMITATIONS

To our knowledge, this is the first study to examine citizenship status and its potential impact on inequities in CRN and cost‐saving strategies. We used a nationally representative survey with a sufficiently large sample to examine differences across citizenship status. We also developed a mediation model to evaluate factors underlying observed inequities in CRN across citizenship status.

Despite these strengths, our study has limitations. First, immigrants may overreport citizenship or decline to participate in NHIS due to stigma or fear associated with their status. However, nearly all adults surveyed responded to questions about their place of birth and citizenship, and the percentage of immigrants in NHIS is similar to national estimates. ¹ Second, NHIS does not distinguish between noncitizens with or without documentation; the latter may have worse access to prescription medications. ⁵ Third, we could not examine primary nonadherence or the proportion of adults who received a prescription but did not fill it. However, we observed similar inequities in CRN by citizenship status among those who did not use prescription medications (Table S11), suggesting that primary nonadherence is also a concern among noncitizens. Nonetheless, these limitations may attenuate the observed inequities in prescription medication access by citizenship status.

For our mediation analysis that evaluated citizenship status and its association with CRN, we were unable to examine several important explanatory factors (e.g., the number and costs of medications or medication coverage among those who are insured) and neighborhood social determinants of health (e.g., residence in marginalized areas or areas with substantial Immigration and Customs Enforcement presence). We were also unable to account for immigration factors that may serve as proxies for acculturation, such as language preference, in our primary analysis. However, among NHIS participants sampled from 2017 to 2018 (when this variable was available), language preference did not meet the conditions necessary to be considered a mediator (Table S3). Furthermore, duration of residence in the US, another proxy for acculturation, did not substantially mediate any of the associations between citizenship status and CRN. Finally, our study suggests that if mediating factors were equal across groups, CRN might be lower or equivalent for noncitizens compared to US‐born citizens; this finding warrants additional research.

6. CONCLUSION

Noncitizens experience a high burden of cost‐related barriers to prescription medications, including higher rates of CRN. Because differences in insurance status and food insecurity explained CRN inequities between noncitizens and citizens, efforts to reduce these inequities should focus on dismantling barriers to health care and food access, regardless of citizenship status. Future work should examine the impact of inequitable access to prescription medications on immigrant health.

AUTHOR CONTRIBUTIONS

JS Guadamuz originated the idea and design of this article, conducted the analyses, wrote the first draft of the article, and contributed critical revisions for important intellectual content. All authors reviewed, revised, and approved the final version of the article.

FUNDING INFORMATION

No funding was received to conduct this study.

CONFLICT OF INTEREST STATEMENT

JS Guadamuz reported current employment with Flatiron Health Inc, an independent subsidiary of the Roche group, and ownership of Roche stock. DM Qato does not report potential conflicts of interest.

Supporting information

Data S1. Supporting Information

Figure S1. Sample selection.

Table S1. Questionnaire items to define prescription medication use‐cost‐related nonadherence and the use of cost‐saving strategies.

Table S2. Univariate association between citizenship and with cost‐related nonadherence, measured using multiple modeling approaches.

Table S3. Selection process for variables that meet the conditions to be considered mediators of the association between citizenship status and cost‐related nonadherence (CRN) among prescription medication users.

Table S4. Characteristics of older adults in the united states, by citizenship status.

Table S5. Characteristics of adult prescription medication users in the united states, by citizenship status.

Table S6. Types of cost‐related nonadherence reported among prescription medication users, by citizenship status.

Table S7. Cost‐related nonadherence among prescription medication users with chronic diseases, by citizenship status.

Table S8. Factors associated with cost‐related nonadherence.

Table S9. Association between citizenship status and cost‐related nonadherence among latinx prescription medication users: estimates from multiple mediation models.

Table S10. Prescription medication use and cost‐related nonadherence (CRN) among adults in the US during the COVID‐19 pandemic, by citizenship status.

Table S11. Cost‐related nonadherence (CRN) among adult who do not use prescription medications, by citizenship status.

Guadamuz JS, Qato DM. Citizenship status and cost‐related nonadherence in the United States, 2017–2021. Health Serv Res. 2023;58(Suppl. 2):175‐185. doi: 10.1111/1475-6773.14185