Impact of perceived access and treatment knowledge on medication preferences for opioid use disorder

Abstract

Background:

Medications for opioid use disorders (MOUD) are effective, but most people with opioid use disorder (OUD) do not receive treatment. Prior research has explored patients’ structural barriers to access and perceptions of MOUD. Little research has considered treatment knowledge and perceptions outside of the patient population. Members of the public without OUD themselves (e.g., family, friends) can significantly influence treatment decisions of persons with OUD. Considering these gaps, we conducted an original survey with a diverse sample of U.S. adults to explore knowledge and preferences toward OUD treatments.

Methods:

We conducted an online survey with 1505 White, Black, and Latino/a Americans including a small percentage (8.5%) with self-reported lifetime OUD. The survey used vignettes to describe hypothetical patients with OUD, provide basic treatment information (i.e., methadone, buprenorphine, naltrexone, non-medication treatment), and then assessed treatment preferences. Using multivariable logistic regression, we examined associations between covariates of interest (e.g., perceived access, knowledge, demographics) and preference for MOUD versus non-medication treatment.

Results:

There were 523 White, 502 Black, and 480 Latino/a respondents. Across racial/ethnic subsamples, respondents had the greatest knowledge of non-medication treatments, with Black (72.7%) and Latino/a (70.2%) respondents having significantly greater knowledge compared to White respondents (61.8%). However, after viewing the vignette, a greater proportion of respondents chose methadone (35.8%) or buprenorphine (34.8%) as their first-choice treatment for hypothetical patients. Multivariable logistic regression suggested that among Black respondents, those with knowledge of non-medication treatment were more likely to choose MOUD than those without (OR=2.41, 95%ci=1.34–4.34). Perceived treatment access did not impact treatment choice.

Conclusions:

Across racial groups, knowledge and perceived access to non-medication treatment was greater than for MOUD, but many still selected MOUD as a first-choice treatment. Significant findings emphasized the importance of treatment knowledge around decision-making, highlighting opportunities for tailored education efforts to improve uptake of evidence-based treatment.

INTRODUCTION

Overdose deaths involving opioids remain a significant cause of death within the United States. In 2022, 81.8% of all fatal drug overdoses in the United States involved opioids.¹ Currently, three evidence-based treatments are available to treat opioid use disorder (OUD); methadone, buprenorphine, and naltrexone.² Despite the demonstrated safety and effectiveness of medications for opioid use disorder (MOUD), it is estimated that 86.6% of people with OUD eligible for treatment do not receive it.³

This substantial treatment gap persists, in part, due to structural (e.g., availability; financial cost; insurance coverage)^4–6 and health system barriers to access (e.g., prescribing policies, dispensing guidelines).^7–9 Gaps in MOUD treatment are particularly pronounced among racial and ethnic minority groups. Previous research has found Black and Latino/a patients were less likely to receive buprenorphine at effective doses and durations.^6,10–12 During the pandemic, of patients who were prescribed buprenorphine and extended release naltrexone, minoritized patients were less likely to fill prescriptions.¹³ To date, experts have hypothesized that most of these disparities are driven by structural differences in access to treatment.^6,14,15

Beyond these substantial structural barriers, there may be individual factors that shape uptake of MOUD. Previous research has highlighted gaps in awareness or knowledge of MOUD in clinical settings, resulting in providers who do not offer MOUD and patients who do not request the medication.^16,17 Lack of MOUD knowledge can exacerbate stigma around MOUD and a constellation of negative beliefs that call into question the legitimacy of medication-based treatment.^18–21 Negative perceptions of MOUD can also deter potential patients from accessing evidence-based treatment for OUD or lead to the discontinuation of treatment.^5,22,23

Although considerable research has explored MOUD beliefs among patients and providers,¹⁸ it is critical to understand MOUD knowledge and perceptions outside of these populations to increase treatment uptake. While healthcare providers can facilitate discussion around MOUD, friends and family have been increasingly recognized as influential in shaping decision-making around MOUD among potential patients.^24–27 Compassionate social networks can provide instrumental, emotional, and financial support as patients navigate substance use treatment.^28–30 However, it is unclear how knowledge of MOUD may vary by race and ethnicity and how perceived access and prior knowledge shapes treatment preferences across diverse communities. Further, few studies have compared perceptions of MOUD to non-medication treatments for OUD (e.g., behavioral therapies, detoxification, group therapies). In light of these knowledge gaps, we conducted an original survey with a diverse sample of U.S. adults to explore their knowledge, perceived access, and preferences toward MOUD and non-medication treatments for OUD.

METHODS

Study Cohort

We conducted a 1500-person web-based survey to assess perceptions of treatments for OUD between October 2020 and January 2021. To conduct this survey, we used Dynata, one of the world’s largest first-party data platforms utilizing innovative survey tools and advanced data collection techniques, and a tool that has been used in previous health and social science research, ^31–35 including in substance use-related studies.³⁶ Dynata aggregates potential participants from different sources to maintain sample consistency and utilizes random digit dialing, banner ads, and other permission-based techniques to recruit and maintain a large database of survey respondents for a non-probability-based sample. To recruit respondents for this survey, Dynata sent out email invitations to respondents, with the goal of a quota-based study sample stratified across race/ethnicity. Within each racial/ethnic subsample, respondents were further stratified by age, gender, and income to reflect the demographics of the US population. Respondents were compensated through Dynata’s credit-based incentive system, with payment rate commensurate with the estimated time required to complete the survey. The study was exempt from full review by the institutional review board at the University of Michigan.

Survey Design and Measures

The aim of the survey was to understand the association of perceived access and prior knowledge on perceptions of treatments for OUD, as well as personal experiences and beliefs around substance use and health-related measures. The survey used 60-second video vignettes to describe a hypothetical patient with OUD and provide basic information about treatments for OUD, mirroring basic information that would be found in an internet search on methadone, buprenorphine, naltrexone, or non-medication treatment options (see supplement for links to video vignettes). Vignettes were also chosen as a delivery format as they have been widely used in health research to simulate the brief interactions a person would hypothetically receive during a clinical encounter where treatment options are discussed.^37–39 Vignettes on MOUD included information about the medication mechanism, dispensation, efficacy, and medication risks.⁴⁰ Vignettes on non-medication treatment included information about different treatment options (e.g., counseling, narcotics anonymous), efficacy, and risks. The race/ethnicity of hypothetical patients were matched with respondents’ race/ethnicity to more closely mirror how respondents might choose treatments for a friend or family member.⁴¹ After the vignettes, the survey assessed respondents’ perceptions around treatment options.

Our primary variable of interest was respondents’ preference for a first choice of treatment for OUD for a hypothetical patient. Treatment options included methadone, buprenorphine, naltrexone, and non-medication options (e.g., Narcotics Anonymous, detoxification, counseling, etc.) assessed through the question: “Assume the cost of treatment is not an issue for [patient name], which treatment would you pick as the best choice for him?” Possible survey response options included 1) methadone treatment; 2) buprenorphine (Suboxone) treatment; 3) naltrexone (Vivitrol); and 4) non-medication treatment. We combined the three medications to compare to non-medication treatment.

Other survey measures included personal experience with substance use and treatments, using questions adapted from the Tobacco, Alcohol, Prescription Medicine and Substance Use (TAPS) tool,⁴² as well as the World Health Organization’s (WHO) alcohol, smoking and substance involvement screening test (ASSIST).^43,44 Measures on prescription opioid use were adapted from Screener and Opioid Assessment for Patients with Pain (SOAPP) alongside the current opioid misuse measure (COMM).⁴⁵ Respondents were also asked if they experienced chronic pain and if they ever “experienced opioid addiction.”

Survey measures assessed respondents’ self-rated knowledge of and perceived ease of access to OUD treatments (e.g., methadone, buprenorphine, naltrexone, non-medication options). treatment knowledge was assessed on a 5-point scale, in which lower values indicated less knowledge. In analysis, treatment knowledge was conceptualized as “a great deal,” “a lot,” or “a moderate amount,” versus “a little” or “none at all.” Perceived ease of access to treatment was assessed on a scale of one to seven, from “very difficult” to “very easy,” with an additional option to select “not sure.” In analysis, “very easy,” “moderately easy,” “somewhat easy,” “neither easy nor difficult” were combined to indicate relative ease of access, versus “somewhat difficult,” “moderately difficult,” “very difficult,” and “not sure” to reflect less perceived access. Lastly, respondents were asked additional demographic information, including age, gender identity, educational attainment, urbanicity (i.e., urban, suburban, rural/small town), and income.

Statistical Analysis

In total, 1530 respondents took the survey, among which 25 (1.63%) respondents had intentional or unintentional missing data in their questionnaires, with remaining total of 1505 (98.37%) respondents included in the analysis. In this exploratory analysis, we first assessed descriptive statistics for demographics and experience with opioid use and treatments for OUD, summarized by frequencies and proportions within White, Black, and Latino/a samples, using χ² test to assess differences across groups. We also assessed treatment knowledge and perceived access to treatment within racial/ethnic subsamples and χ² tests were used to assess differences across race/ethnicity.

Univariate and multivariable logistic regression were performed to evaluate the association between covariates of interest and preference for MOUD (versus non-medication treatment) as a first-choice treatment within race/ethnicity subsamples. In the multivariable logistic regression model, additional demographic including gender, age group and income category were included. Odds ratios (ORs) and the corresponding 95% confidence intervals (CIs) of key explanatory covariates were reported. Multicollinearity diagnostics were calculated by using variance of inflation factors (VIFs) to assure the fitness of all models. All statistical analyses were performed using R statistical software (version 4.1.3; R Foundation for Statistical Computing, Vienna, Austria). All tests were 2-sided with a significance level of 0.05 and p-values were adjusted to control false discovery rate (FDR).

RESULTS

There were 1505 total respondents in the analytic sample, including 523 White, 502 Black, and 480 Latino/a respondents. Across all groups, there were more women (56.4%) than men and the largest proportion of respondents was between 25 and 35 years old (32.4%). Across all groups 17.7% of respondents reported using opioids in the past year, and 8.5% had a lifetime history of self-reported OUD. In describing experiences with substance use treatment, 14.2% of all respondents had experience with MOUD, and 43.1% had experience with non-medication treatment. All demographics are displayed in Table 1, stratified by race/ethnicity.

Table 1:

Respondent Characteristics

	White (n=523)	Black (n=502)	Latino/a (n=480)	Adjusted p-value†
	N (%)	N (%)	N (%)
Demographics
Gender
Woman	286 (54.7)	263 (52.4)	300 (62.5)	0.004
Man	237 (45.3)	239 (47.6)	180 (37.5)
Age
<24 years old	59 (11.3)	104 (20.7)	117 (24.4)	<0.001
25–35 years old	153 (29.3)	160 (31.9)	175 (36.5)
35–44 years old	45 (8.6)	32 (6.4)	38 (7.9)
45–54 years old	30 (5.7)	29 (5.8)	23 (4.8)
55–64 years old	90 (17.2)	60 (12.0)	47 (9.8)
> 65 years old	146 (27.9)	117 (23.3)	80 (16.7)
Education
High school or less	164 (31.4)	172 (34.3)	128 (26.7)	<0.001
Some college	87 (16.6)	129 (25.7)	124 (25.8)
Associate	41 (7.8)	65 (13.0)	68 (14.2)
Bachelor’s or higher	231 (44.2)	136 (27.1)	160 (33.3)
Urbanicity
Rural/small town	224 (42.8)	134 (26.7)	131 (27.3)	<0.001
Suburban	201 (38.4)	173 (34.5)	178 (37.1)
Urban	98 (18.7)	195 (38.8)	171 (35.6)
Income
Less than $20,000	102 (19.5)	143 (28.5)	95 (19.8)	0.005
$20,000–49,999	180 (34.4)	175 (34.9)	179 (37.3)
$50,000–99,999	174 (33.3)	130 (25.9)	154 (32.1)
More than $100,000	67 (12.8)	54 (10.8)	52 (10.8)
Chronic pain and opioid use
Chronic pain	156 (29.8)	142 (28.3)	158 (32.9)	0.276
Opioid use (past year)	80 (15.3)	102 (20.3)	85 (17.7)	0.109
Experience with substance use disorders
Self-reported OUD (lifetime)	50 (9.6)	41 (8.2)	37 (7.7)	0.545
Close social exposure to OUD	143 (27.3)	128 (25.5)	145 (30.2)	0.252
Close social exposure to non-opioid SUD	126 (24.1)	129 (25.7)	149 (31.0)	0.036
Experience with treatment
MOUD	68 (13.0)	67 (13.4)	78 (16.3)	0.276
Non-medication treatment	197 (37.7)	219 (43.6)	233 (48.5)	0.002

OUD, Opioid Use Disorder; MOUD; Medications for Opioid Use Disorder

^†χ² tests were used. p-values were adjusted to control FDR < 0.05.

In assessing respondents’ treatment knowledge, 35% of respondents across all groups had knowledge of methadone, 24.4% had knowledge of buprenorphine, and 21.1% had knowledge of naltrexone. However, a majority (68.1%) of respondents had knowledge of non-medication treatment options, and Black (72.7%) and Latino/a (70.2%) respondents had significantly greater knowledge of non-medication treatment options than White (61.8%) respondents (Table 2). Respondents were also asked if they believed they would have access to each of the treatment options, and the majority perceived access to methadone (57.5%), buprenorphine (56.5%), naltrexone (54.3%), and non-medication treatment (60.8%) options. Perceived access was high overall.

Table 2:

Treatment Knowledge and Perceived Access to Treatment, by race/ethnicity

	White (n=523) n (%)	Black (n=502) n (%)	Latino/a (n=480) n (%)	Adjusted p-value†
Treatment knowledge ††
Methadone	161 (30.8)	191 (38.1)	175 (36.5)	0.099
Buprenorphine	112 (21.4)	133 (26.5)	122 (25.4)	0.218
Naltrexone	91 (17.4)	120 (23.9)	106 (22.1)	0.099
Non-medication treatment	323 (61.8)	365 (72.7)	337 (70.2)	0.003
Perceived ease of treatment access †††
Methadone	287 (54.9)	310 (61.8)	269 (56.0)	0.121
Buprenorphine	310 (59.3)	269 (53.6)	271 (56.5)	0.247
Naltrexone	293 (56.0)	268 (53.4)	256 (53.3)	0.614
Non-medication treatment	324 (62.0)	311 (62.0)	280 (58.3)	0.466

^†χ² test was used for both analyses. p-values were adjusted to control FDR < 0.05.

^††Defined as “A moderate amount,” “a lot,” or “a great deal” versus “none at all” or “a little”

^†††Defined as “very easy,” “moderately easy,” “somewhat easy,” “neither easy nor difficult”

When asked their first choice of treatment, respondents reported methadone (35.9%) and buprenorphine (34.9%) were the preferred choice across race/ethnicity groups, with no significant differences (Figure 1). A similar proportion of each race/ethnicity group (14.3%, 15.8% and 16.6% among Black, Latino/a and White respondents, respectively) chose naltrexone as their first choice of treatment, at lower rates than the other MOUD options. Notably, 17%, 13.4% and 11% respondents in Black, White and Latino/a race/ethnicity groups indicated that non-medication treatment was their first choice of treatment for OUD, with a significant higher proportion of Black respondents compared to White respondents (p-value: 0.005) (Figure 1).

FIGURE 1.

Respondents’ first-choice treatment for opioid use disorder, by race/ethnicity

Respondents were asked about their first choice of treatment for OUD, categorized as MOUD versus non-medication treatment in multivariable logistic regression analysis (Table 3). While overall, Black respondents were more likely to choose non-medication treatments (Figure 1), if they had higher knowledge of non-medication treatments, they were more likely to prefer a MOUD, compared to non-medication treatments (Odds Ratio (OR)=2.41, 95% Confidence Interval (CI)=1.34–4.34) (Table 3). While perceptions of access to treatment varied across treatment type and racial/ethnic sub-samples, there was no association between perceived ease of access and treatment choice.

Table 3:

Logistic Regression of MOUD as first-choice treatment (versus non-medication treatment), by race/ethnicity

	White		Black		Latino/a
	OR (95% CI)	Adjusted p-value†	OR (95% CI)	Adjusted p-value†	OR (95% CI)	Adjusted p-value†
Demographics
Gender (ref=woman)	0.59 (0.33 – 1.05)	0.211	0.89 (0.54 – 1.46)	0.788	0.75 (0.43 – 1.31)	0.668
Age	1.02 (0.86 – 1.20)	0.950	0.96 (0.84 – 1.1)	0.742	1.01 (0.86 – 1.19)	0.918
Income	1.25 (0.91 – 1.72)	0.383	1.34 (1.04 – 1.76)	0.153	1.22 (0.89 – 1.67)	0.595
Treatment Knowledge (ref = none at all + a little knowledge)
MOUD treatment	1.13 (0.48 – 2.72)	0.950	0.80 (0.44 – 1.43)	0.703	1.06 (0.56 – 2.01)	0.918
Non-medication treatment	2.03 (1.03 – 4.17)	0.211	2.41 (1.34 – 4.34)	0.035	1.30 (0.68 – 2.50)	0.671
Treatment Access (ref = not sure + have difficulty in access treatment)
Methadone	0.71 (0.26 – 1.91)	0.783	1.08 (0.53 – 2.2)	0.925	2.08 (0.82 – 5.25)	0.494
Buprenorphine	3.87 (1.04 – 15.02)	0.211	1.73 (0.71 – 4.19)	0.499	1.26 (0.44 – 3.64)	0.903
Naltrexone	0.96 (0.26 – 3.29)	0.950	0.96 (0.38 – 2.37)	0.929	0.51 (0.2 – 1.22)	0.494
Non-medication treatment	0.94 (0.44 – 2.04)	0.950	0.74 (0.4 – 1.35)	0.593	1.31 (0.7 – 2.47)	0.671
Opioid use
Self-reported OUD (lifetime)	1.96 (0.52 – 12.8)	0.710	2.15 (0.72 – 9.28)	0.499	1.21 (0.43 – 4.32)	0.903

Note: OUD, Opioid Use Disorder; OR, Odds Ratio; CI, Confidence Interval

^†p-values were adjusted to control FDR < 0.05.

DISCUSSION

This study evaluated treatment preferences for OUD across a racially diverse sample and assessed whether knowledge of and perceived access to these treatments was associated with treatment preferences. Results demonstrated that across race/ethnicity, despite having the greatest knowledge of non-medication treatment, after receiving basic video information about treatment options, respondents generally chose methadone and buprenorphine as a first-choice treatment. Further, most respondents believed they had access to all four treatment modalities, and in multivariable analysis, perceived access was not significantly associated with treatment preference. However, knowledge of non-medication treatments was associated with preference toward MOUD among Black respondents, although not in other groups.

This study is one of the first to assess individual MOUD preferences in a racially and ethnically diverse sample of the general U.S. population, who may or may not use opioids. Our findings found that by providing basic information on medications and efficacy using vignettes, overall preferences for MOUD were high irrespective of race/ethnicity. While Black respondents were more likely to prefer non-medication treatments (Figure 1), in multivariable analysis, Black respondents with greater knowledge of non-medication treatments were more likely to select MOUD as a first-choice treatment, consistent with previous research using patient populations.^46–48 This difference may highlight respondents’ knowledge of the lower efficacy or effectiveness of non-medication treatments.

Despite studies demonstrating structural barriers in access to MOUD,^6,14,15 a majority of respondents in our sample believed they could access treatments for OUD relatively easily. In addition, perceived ease of treatment access was not associated with their first choice of treatment. These perceptions are consistent with emerging literature, that despite recent policy efforts to improve treatment access, including both in-person and through telehealth, MOUD uptake remains relatively stagnant.^3,49 It may be that access is less of a consideration, at least for the general public, and that treatment perceptions (e.g., knowledge, beliefs, stigma) for family members and friends of people who use drugs should also be considered critical intervention points for improving uptake of evidence-based treatment strategies.

There are limitations to our study. First, we used an online working panel, collected via non-probability-based sampling methods, which may not be representative of the general U.S. population. In fact, our sample reported 8.5% with a history of self-assessed OUD which is higher than the national average.⁵⁰ However, prior studies using web-based surveys have shown similarly that this population may actually have more substance use and mental health comorbidities.^51,52 Despite these limitations, the Dynata panel allowed us to recruit a racially stratified sample which may not have been as feasible outside of this online setting. Second, we do not know the effect the vignettes may have had on treatment preference as we did not assess preferences, knowledge, or access preceding the vignettes. Further, we did not test the educational level of the vignettes or adjust the vignettes to varying literacy levels. However, we chose a vignette design to mirror the level of knowledge a person would acquire from a simple online search, or to jog their memory of previous treatment knowledge. We also chose a video vignette format to reduce respondent burden and address concerns about respondents’ literacy.⁵³ Third, due to the study methodology, we only measured perceived access to treatment in a general population. A population who does not have an OUD may not be aware of access barriers specific to SUD treatment and may respond based on perceived access to medical treatments in general. Future studies should triangulate associations between measured access (financial and geographic), perceived access, and preferences for treatment.

In this study, we explored associations between knowledge, access, and preferences toward treatments for OUD among a diverse sample of U.S. adults. Perceived medication access was less of a consideration than knowledge and preferences in shaping treatment choice, highlighting potential opportunities for public health interventions to focus on educational outreach and engagement around MOUD. Further, assessing treatment perceptions beyond the patient population provides critical insight into how friends and family may support their loved ones in discussing treatment options for OUD. Discrepancies in knowledge of MOUD versus non-medication options highlight the need for increasing education and improving perceptions of evidence-based treatment options among communities, to better support people with opioid use disorder in their treatment decision-making.