Association Between Opioid Prescriptions and Non–US-Born Status in the US

This cross-sectional study examines the patterns of opioid prescribing among non–US-born individuals in the US.


Introduction
The opioid epidemic continues to be a major public health threat that has been associated with high rates of overdose and heroin addiction across the US. Deaths from overdose that involve opioid use number nearly 50 000 every year. 1 A recent study 2 by the Society of Actuaries estimates that the opioid epidemic cost the US economy more than $630 billion from 2015 to 2018. Adverse consequences from opioid use are particularly pronounced among low-income and publicly insured individuals. 3 The association of opioid use and overdose with race and socioeconomic status is well established. For example, 2 previous studies 3,4 using national data on inpatient admissions and Medicare Part D data reported that patients taking opioids were more likely to reside in low-income areas. White patients are more likely than black and Hispanic patients to be hospitalized and die of prescription opioid use, 3,4 but there is evidence that prescription opioid fatalities may be increasing more rapidly among patients in minority groups than among white patients. 5 A study 6 on hospital discharges associated with opioid overdose found significantly higher opioid overdose rates in low-income zip codes compared with high-income zip codes. However, little is known about the association of opioids with non-US-born status in the US. Non-US-born individuals experience high rates of poverty and often work in occupations with high risk of musculoskeletal injury, such as construction and farm industries, which are associated with high rates of opioid prescriptions. 7 Furthermore, noncitizen populations have reduced access to care, which might be associated with opioid prescriptions. 8 To address the knowledge gap about the prevalence of prescription opioid use in non-US-born communities, we used a large-scale, nationally representative data set, the Medical Expenditure Panel Survey (MEPS), to examine the association between opioid prescriptions and non-US-born status, particularly for those clinically diagnosed with pain.

Methods
This cross-sectional study used the 2016-2017 MEPS, a nationally representative, in-person survey database compiled by the Agency for Healthcare Research   in the Supplement). 10 On the basis of ICD-10-CM codes, the CCI adjusted for 17 comorbidities that may increase participants' risk of mortality. The CCI was categorized as 0, 1, and 2 or higher. Length of US residency was asked of non-US-born individuals. This variable was categorized as less than 5 years of residency vs 5 or more years of residency in the US.

Statistical Analysis
Univariate analyses included descriptive statistics stratified by non-US-born status and a comparison of receipt of any opioid prescription and number of days prescribed between US-born and non-USborn respondents. Multivariable logistic regression estimated receipt of an opioid prescription, and multivariable negative binomial regression modeled the number of days prescribed for respondents receiving any opioid prescription. These regression models were also stratified between patients diagnosed with chronic pain, acute pain, and neither chronic nor acute pain. To examine whether length of residency, defined as less than 5 years and 5 or more years, for non-US-born individuals is associated with opioid prescriptions, we used multivariate logistic regression modeling of any opioid prescription and days' supply if prescribed for only the non-US-born population. Sensitivity and stepwise regression analyses were performed. All analyses were weighted and adjusted for complex survey design. Statistical significance was considered to be P < .05 using Stata software, version 16 (StataCorp).

Results
The sample size was 48 729. After listwise deletion, the analytical sample size was 48 162, with a missing rate of 1.2%. Because of the low missing rate, imputation methods were not performed. The Among non-US-born individuals, 6.7% had less than 5 years of US residency. Nearly half (47.1%) of non-US-born individuals were Hispanic compared with 9.5% of US-born individuals. US-born individuals were more likely than non-US-born individuals to be 65 years or older (21.3% vs 16.3%) and less likely to be married (50.0% vs 61.8%). One in 4 non-US-born individuals had less than a high school education (24.7% vs 11.3% for US-born individuals). Non-US-born individuals also had greater

JAMA Network Open | Public Health
Association Between Opioid Prescriptions and Foreign-Born Status in the US likelihood of poverty (14.1% vs 10.1%) and lack of insurance (18.0% vs 6.1%) than US-born individuals.
However, non-US-born individuals tended to be healthier than US-born individuals. For example, 82.0% of non-US-born individuals did not have comorbidities vs 74.7% of US-born individuals, and a lower percentage of non-US-born individuals had received a diagnosis of chronic (29.3% vs 41.0%) or acute pain (14.2% vs 23.5%) than US-born individuals.
Non-US-born individuals were significantly less likely to receive any opioid prescription within a 12-month period than US-born individuals (Figure 1 Among respondents receiving an opioid prescription, multivariable negative binomial regression modeled the incidence rate ratio (IRR) of number of days' supply of opioid prescriptions.

Non-US born
Error bars indicate 95% CIs. Additional multivariable regressions examined the association of length of US residency in years among non-US-born individuals with receiving an opioid prescription ( Table 2). Non-US-born individuals with less than 5 years of residency in the US were significantly less likely to receive a prescription for opioids than were those with longer residency after adjustment for type of pain and other confounding factors (AOR, 0.51; 95% CI, 0.30-0.88). Results for length of residency were not statistically significant for days' supply.

JAMA Network Open | Public Health
We also stratified the analyses by health care setting and examined differences in opioid prescriptions by non-US-born status (eTable 5 in the Supplement). Health care settings included office-based physician, outpatient, emergency department, inpatient, and dental offices.
Approximately half of opioid prescriptions were from office-based physicians, but there were no statistically significant differences in practitioner setting by non-US-born status. As additional sensitivity analysis, we excluded opioid prescriptions associated with dental services, but results were not substantively different for non-US-born vs US-born status in the receipt of any opioid prescription (AOR, 0.64; 95% CI, 0.56-0.74) or for number of days' supply (IRR, 0.74; 95% CI, 0.61-0.88).
We compared unadjusted differences in the receipt of opioid prescriptions by non-US-born status with differences adjusted for socioeconomic status, access to care, and comorbidities (eTable 6 in the Supplement). Although differences between US-and non-US-born individuals decreased with these additional factors, there were no statistically significant differences across AORs.
We examined interactions in the probability of receiving opioid prescriptions and non-US-born status with chronic pain, acute pain, health insurance, poverty status, and age (eTable 7 in the Supplement). In every case, non-US-born individuals had lower likelihood of receiving an opioid prescription compared with US-born individuals, although the difference was not statistically

Non-US born
Error bars indicate 95% CIs.

Discussion
The opioid epidemic shows few signs of abating, with overdose deaths associated with opioid use increasing every year since 2012. 1 Despite the important implications of the epidemic for public health, little is known about its population health consequences for non-US-born communities. We used nationally representative data on sociodemographic factors, medical conditions, and prescriptions to examine the association of opioid use with non-US-born status in the US. Non-USborn individuals were significantly less likely to receive a prescription for opioid medications than US-born individuals regardless of whether they were clinically diagnosed with pain. For example, although 1 in 4 US-born individuals diagnosed with chronic pain were prescribed opioids, only 15% of non-US-born individuals received an opioid prescription. Furthermore, among patients receiving prescriptions for opioids, non-US-born patients were generally treated for a substantially shorter period than US-born patients. US-born individuals with diagnosed chronic pain received more than 70 days' supply of opioids in contrast to the 44-day supply for non-US-born individuals with chronic pain. However, we did not find statistically significant differences in prescriptions for opioids among patients diagnosed with only acute pain. In addition, our data did not show statistically significant differences in the source of opioid prescriptions across practitioner settings (office-based physician, outpatient, emergency department, inpatient, and dental) between US and non-US-born respondents. Interacted regression analysis showed that white non-Hispanic, non-US-born individuals had a significantly higher likelihood of receiving opioid prescriptions than did Hispanic or Asian non-US-born individuals. Differences in opioid prescriptions were not significantly different across racial/ethnic groups for US-born individuals.
Reasons for differences in prescribing of opioids between non-US-born and US-born individuals are unclear and warrant further research. One interpretation of our findings is that non-US-born individuals may be less likely to report experiencing pain than US-born individuals. However, we adjusted for clinical diagnosis of chronic and acute pain using ICD-10-CM codes, which mitigates this possibility. Another possible explanation for the differences is the specialties of practitioners with prescribing authority that treat non-US-born vs US-born individuals. For example, a recent article 11 reported that 22.3% of dental prescriptions were for opioids, such as oxycodone, for patients of US dentists compared with 0.6% of prescriptions written by dentists in England. Prior research 12 found that non-US-born individuals were significantly less likely to use dental services than US-born individuals. However, our sensitivity analysis suggested that differences in opioid prescriptions by non-US-born status remained after opioid prescriptions associated with dental services were excluded. Other research using Medicare Part D prescription claims data indicates that nurse practitioners and physician assistants accounted for three-quarters of prescribers having the highest 5% of opioid prescription proportions. 13 Furthermore, most high rates of opioid prescribing were concentrated in urgent care clinics. 13 In our data, US-born individuals are more likely to report a nurse practitioner or physician assistant as their usual source of care (4.5%) than non-US-born individuals (1.1%). Other research has demonstrated that non-US-born individuals also use urgent care settings, such as emergency departments, at lower rates than US-born individuals. [14][15][16] In addition, there have been long-standing issues with racial/ethnic disparities in quality of care in the treatment of pain, which may help explain the differences observed in our data. 17,18 Prior literature has documented inadequate pain management for racial/ethnic minority groups even after adjustment for demographic factors and pain intensity. 18 However, an important implication of the lower likelihood of opioid prescribing among non-US-born individuals in our study is that it may have helped protect non-US-born communities from the high rates of opioid-related mortality occurring in the US. 19 It is also unclear whether non-US-born individuals are more likely to receive non-pharmaceutical-based therapies for pain if they are less likely to be prescribed opioid medications than US-born individuals. Further research is needed to explore this hypothesis.
Furthermore, our analysis found that non-US-born individuals with less than 5 years of residency in the US were less likely to receive a prescription for opioids than those with longer residency. Our analysis adjusted for access to care measures, such as poverty and uninsured status. However, recently immigrated individuals experience significant legal barriers to qualify for federally funded benefits programs. Under the Personal Responsibility and Work Opportunities Act of 1996, 8 undocumented immigrants and most authorized immigrants are ineligible for Medicaid, Medicare, and other benefit programs unless funding is provided by the state. As a result, compared with longer-residency non-US-born individuals, non-US-born individuals with shorter residency may have different characteristics of insurance coverage, types of practitioners seen, or access to care difficulties associated with the likelihood of receiving a prescription.

Limitations
This study has limitations. Although MEPS attempts to validate medical care use of respondents, respondents may have recall bias on using any medical care in the prior 12-month period. However, it is unclear whether this recall bias systematically differs between US-born and non-US-born individuals. The non-US-born individuals may also be less likely to participate in surveys than US-born respondents, which may affect representativeness of the MEPS. In addition, we do not have data on whether prescribed opioid use resulted in differing rates of opioid misuse or abuse between US-born and non-US-born individuals. Differences in temporal trends in opioid prescribing are unclear and warrant further research. Also, the sample of non-US-born respondents reporting acute pain is limited in size.

Conclusions
Our findings suggest that non-US-born individuals are significantly less likely to receive an opioid prescription than US-born individuals, particularly among those with diagnosed chronic pain.