Rural-Urban Differences in Physician Bias Toward Patients With Opioid Use Disorder

The U.S. opioid epidemic has had considerable impacts on public health, local economies, and social life. On average, 130 Americans die each day from an opioid overdose, and 400,000 lives were lost between 1999 and 2017 (1). Nonmedical opioid use is further associated with a host of secondary health conditions, including infective endocarditis; skin, soft tissue, and bone infections; and transmission of HIV and the hepatitis C virus (2, 3). The result has been a vast increase in health care expenditures for opioid use disorder, which currently exceed $500 million annually (4). The health effects of the opioid crisis underscore the need for expanded health care services, especially in rural areas where opioid treatment and harm reduction services are scarce.

Distinctions between rural and urban contexts have been at the forefront of examinations of the opioid crisis because although the opioid epidemic is felt nationally, certain states and regions report disproportionate amounts of opioid use and associated deaths (5). Rurality and prescription drug misuse are correlated, and rural communities have correspondingly become the focus of news coverage of the opioid crisis (6). Opioid misuse also varies across rural America; states in the Midwest and Northeast have higher mortality rates than does the rest of the country (7). Together, these findings suggest that access to health care services related to opioid use disorder are critically needed in certain rural U.S. regions.

Opioid use disorder services are extremely limited in rural areas. For example, the waivers necessary to prescribe medications to patients who misuse opioids are unequally distributed, with the overwhelming majority located in urban areas (8). Harm reduction services reduce both morbidity and mortality rates related to opioid use disorder, but these services are also disproportionately found in urban areas (9). Some state regulations, for example, limit the ability of emergency medical technicians to administer the opioid antagonist naloxone, a restriction that disadvantages rural communities (9, 10). Although some states have successfully overcome barriers to expanding harm reduction services such as syringe exchanges, many rural communities still struggle to provide widespread access because of stigma surrounding opioid misuse (11, 12). In summary, the evidence on treatment and harm reduction programs for opioid use disorder suggests that health care resources are mainly concentrated in urban areas and may reflect broader social and cultural stigma toward opioid misuse in rural communities.

The goal of this study was to better understand the barriers to expanding opioid use disorder services in rural areas. We focused on physicians’ attitudes toward patients who misuse opioids, because biases against this patient population are known to be both widespread and a key obstacle to the provision of effective health care services (13). Indeed, physicians report negative attitudes toward patients who misuse opioids, experience negative emotional states when working with these patients, and are also less comfortable with and less prepared to treat patients with opioid misuse than patients with other psychological conditions (14). Patients with opioid use disorder also report that negative interactions with health care workers are commonplace (15, 16), and such biased interactions have been linked to less favorable treatment outcomes (14).

Despite this evidence, we still do not understand whether bias is a more acute barrier to health care access in rural settings. We hypothesized that physician attitudes differ between rural and urban physicians because of variation in the social, economic, and political factors that shape access to opioid use disorder treatment across rural and urban areas. It is plausible, therefore, that physician bias is an additional barrier to health care access in rural communities, where patients are already medically underserved. In this study, we investigated whether physician bias toward patients with opioid use disorder varies across urban and rural communities in Ohio, a state with one of the highest rates of drug overdose in the United States.

Methods

Study Population

Our sample included board-certified physicians licensed to practice in the state of Ohio who were recruited to participate in a study to better understand how the growing misuse of opioids has affected U.S. physicians. We identified possible participants with help from the state board of medical licensing, which provided a database of contact information for physicians who are eligible to practice in Ohio. We contacted all physicians who listed an Ohio address (N=34,397) by e-mail and invited them to participate in the study if their medical practice had been “reshaped by the opioid epidemic.” We sent an initial e-mail invitation and one reminder e-mail in October 2019. The anonymous online survey remained open through November 2019. At the end of the survey, participants could choose to click a link to a second survey where they could enter their e-mail address into a raffle to win one of three $200 Amazon gift cards as compensation for their time. The use of a separate survey to collect identifying information ensured that their contact information could not be paired with their survey responses. Our study was approved by the Ohio University Institutional Review Board, and all respondents provided electronic informed consent before participation.

Data and Measures

We used a survey composed of both open- and closed-ended questions to assess the extent to which physicians’ bias against patients who misuse opioids varied across rural and urban physicians and to identify the factors that might explain attitudinal differences among the physicians. Our dependent variable was physicians’ scores on a 10-item scale designed to assess bias toward patients who misuse opioids. The scale was initially created by Brener and von Hippel (13) to measure bias toward injecting drug users, and previous work has established the reliability and validity of this measure for use with both physician and nonphysician populations. We adapted the scale to assess beliefs about people and patients who misuse opioids more generally. Specifically, we replaced “injecting drug users” with “opioid misusers.” Sample items from the scale include, “Opioid misuse is immoral” and “I avoid opioid misusers whenever possible.” Respondents were asked to rate the extent to which they agreed with each item on a scale from 1, strongly disagree, to 5, strongly agree. Scores on the scale were averaged, and total scores ranged from 1 to 5, with higher scores indicating more bias. We computed an estimate of internal consistency and reliability, and our results indicated that the scale had adequate reliability (α=0.81).

In addition to asking physicians about their attitudes toward patients with opioid use disorder, we collected data on physicians’ primary area of practice. Specifically, we coded their responses (as yes or no) for whether they worked in professional settings, such as in addiction treatment or in the emergency department, where they may have had specific types of contact with patients who misuse opioids. Given previous studies of the correlates of bias toward individuals with substance use disorders (17, 18), we also assessed physicians’ level of contact with patients, measured as the percentage of their work hours dedicated to treating patients with opioid use disorder, and their levels of stress induced by working with such patients (19) and their general levels of burnout (20). Finally, respondents were asked to report demographic information, including their sex, age, average number of hours worked per week, and number of years in their current position, and whether they were currently prescribing or have previously prescribed opioids.

Our focal independent variable was whether the respondent lived in a rural or urban county, classified on the basis of the Office of Management and Budget’s definition of nonmetro counties. Ohio has 88 counties, 48 of which are classified as rural. Furthermore, approximately 20% of Ohio’s population resides in a rural county (21). We also accounted for the presence of opioid treatment and harm reduction resources in the counties where physicians practiced by merging our collected data with available data on relevant county characteristics. We included data on the county drug overdose rate from the 2019 County Health Rankings database (22). We further coded the number of naloxone dispensaries per 100,000 residents in a county, as reported by the State of Ohio Board of Pharmacy (23); the presence of syringe exchanges in a county, coded as a dichotomous variable sourced from Harm Reduction Ohio (24), a nonprofit organization that tracks syringe exchange programs across Ohio counties; and the number of buprenorphine prescribers per 100,000 residents in a county, sourced from the Substance Abuse and Mental Health Services Administration’s Buprenorphine Practitioner Locator (25).

Analysis

To assess differences in physicians’ attitudes between rural and urban settings, we first employed descriptive statistics to describe the sample and used t tests to calculate differences in bias and in burnout, stress, and contact levels with patients who misuse opioids between rural and urban physicians. We also used t tests to assess resource differences in the rural and urban counties where these physicians practiced. We then performed multivariable linear regression to assess whether observed differences in bias between rural and urban physicians persisted after accounting for several control variables and bias predictors. We report findings on the relationship between bias, rurality, and opioid use disorder services. All statistical analyses were conducted with Stata, version 15.

Results

In total, 408 physicians completed the online survey. Because some physicians did not report their county and small counties in Ohio do not report opioid-related statistics, we excluded 134 participants because of missing county-level data, leaving a total sample of 274. These physicians worked primarily within the state of Ohio; where available, we provide data on the extent to which our sample represented the state population of physicians (26).

Of the total sample, 61% identified as male (for comparison, 65% of all physicians in Ohio identify as male), and the mean participant age was 51 years (Table 1). Participants worked on average 46 hours per week and had been in their current position for on average 13 years. Approximately 11% of the sampled physicians worked in addiction treatment; another 14% worked in emergency medicine, compared with 5% of all physicians in Ohio who work in emergency medicine. On average, participants reported that 18% of their work hours were devoted to working with patients with opioid use disorder. Approximately 14% of physicians practiced in a rural county, and participants represented 61 of the 88 counties in Ohio.

On average, respondents scored 2.89 across the 10-item bias scale (Table 2), indicating a moderate level of bias. Of particular importance for our study, physicians from rural counties scored an average of 3.25 on this scale, compared with an average score of 2.83 among their urban counterparts, a difference that was statistically significant (p<0.01). No significant differences between rural and urban physicians were noted in the amount of contact with patients who misuse opioids, stress related to working with this patient population, and burnout levels. The percentage of physicians working in addiction or emergency medicine also did not differ between urban and rural counties.

We assessed the availability of treatment and harm reduction resources and observed differences between the rural and urban counties in which participants practiced. Buprenorphine waivers were more common, on average, in urban counties than in the rural Ohio counties represented in the sample (25.3 vs. 18.1 prescribers with waivers per 100,000 residents in a county, respectively, p<0.01). Of the 37 physicians in our sample practicing in rural counties, 19% (N=7) had a syringe exchange in their county, compared with the 68% (N=161) of the 237 physicians practicing in urban counties (p<0.01). Finally, physicians practicing in rural counties had fewer naloxone dispensaries nearby than had those practicing in urban counties (12.7 vs. 15.3 per 100,000 residents in a county, p<0.01).

In the multivariable linear regression model, rurality was positively associated with physician bias toward patients who misuse opioids, independently of key predictors of bias, access to specialty physicians and treatment, and the presence of harm reduction resources (i.e., syringe exchanges) (Table 3). In this analysis, stress also was positively associated with bias, as were the total number of hours a physician worked per week. Physicians who specialized in addiction medicine had lower bias than physicians without this specialization.

Discussion

The primary goal of this study was to examine patterns of physician bias toward patients with opioid use disorder across rural and urban contexts. Drawing on a sample of 274 physicians in Ohio, we found that their reported level of bias toward this patient population exceeded bias reported in previous studies toward injecting drug users (mean reported score on the 10-item scale=1.48), persons with hepatitis C virus (mean score=2.14), HIV-infected injecting drug users (mean score=1.88), and persons with same-sex attraction (mean score=1.39) (17, 27, 28). Furthermore, our results suggest that the reported level of bias in our sample was driven at least partially by comparatively high rates of bias in rural settings. This finding is important because these are the same communities that have been increasingly affected by the opioid epidemic and thus have the highest need to care for patients with opioid use disorder. Our results indicate that in addition to having greater bias than their urban counterparts, rural physicians also practice in counties with fewer syringe exchanges, naloxone dispensaries, and buprenorphine prescribers. In other words, physician bias appears to be a critical barrier in counties that are already facing shortages in treatment and harm reduction services and indeed may contribute to the lack of supportive services in these areas.

Bias was also associated with individual characteristics of physicians, but these factors could not fully explain the differences in bias between physicians in rural and urban counties. Physicians working in addiction medicine had lower rates of bias than those who did not work in this specialty, but they were equally as likely as their counterparts without this training to practice in rural or urban counties. Contact with patients with opioid use disorder, stress related to such contact, and burnout, which are three key predictors of bias and which have been the focus of interventions to reduce physician bias, did not vary systematically between physicians in urban and rural counties. Taken together, these findings suggest that typical interventions to reduce bias may not be appropriate or effective for overcoming the unique contributors to bias among rural health care providers.

Implications

Our findings have implications for patients with opioid use disorder and their access to adequate health care and other supportive services. First, we found relatively high levels of bias toward people who misuse opioids, and we also found that this bias was higher among rural physicians than among urban physicians. When paired with high rates of opioid use disorder in many rural communities, this finding suggests that bias may be an important barrier to accessing adequate health care in regions most affected by the opioid epidemic. Rural areas are already more likely to face resource and service shortages, and concerns may exist in this setting about rationing resources for individuals who are deemed most deserving. It is important to note that bias has been linked to a reluctance to provide care for patients, less effective health care delivery, and more negative treatment outcomes, among other deleterious effects (14). Thus, efforts to reduce bias among rural health care providers and facilities are clearly needed, particularly because rural areas may benefit most from such efforts.

Given the shortage of treatment and harm reduction services in rural counties, bias reduction efforts may also be important for garnering support to adopt and implement evidence-based services for opioid use disorder. In other words, bias among physicians may reflect stigma that is broadly embedded in medical and social institutions. Future studies should explore predictors of physician bias that are unique to rural areas. Drawing on previous research on the correlates of nonmedical opioid use in rural communities (29), we propose future investigation into the ways in which the social environment—including factors such as health care access, employment, and social cohesion—may shape the availability of substance use services and the beliefs and attitudes of physicians who treat patients with opioid use disorder. Moreover, previous research has suggested that some pharmacists also express negative attitudes toward dispensing naloxone and other medications that help treat patients with opioid use disorder, and future investigations could examine whether these negative attitudes are also more pronounced among pharmacists working in rural areas, which may compound existing treatment barriers (30).

The confluence of multiple factors, including nonmedical opioid use, chronic disease, and economic distress among patients, may contribute to greater stigma among physicians. In addition, political differences and beliefs about addiction treatment may account for the rural-urban differences in bias we found here. In rural Ohio, for example, elected officials have advocated limiting or eliminating naloxone administration in cases of opioid overdose (31, 32). Information on the specific determinants of bias in rural settings will be critical to developing interventions to reduce stigma and increase the availability of health care services for patients with opioid use disorder. Physicians also often play additional roles in the communities in which they practice and may be key change makers for lowering bias related to nonprescription opioid use. Thus, interventions to reduce physician bias may have implications beyond the clinic. Several rural counties in our sample had relatively low rates of physician bias and also had opioid services. These “bright spots” may provide additional insight into the physician- and community-level factors that can be addressed to reduce stigma and expand access to evidence-based opioid services (33).

Limitations

Although we report on bias from a larger sample than samples used in previous research studies, our participants represented a small percentage of physicians working in Ohio. For that reason, our findings may not generalize to all health care providers or providers working in other regional contexts. It is possible that physicians with higher levels of bias were more likely to take a survey related to working with patients who misuse opioids, thereby inflating our overall bias findings; however, a higher percentage of urban physicians, with lower average bias scores, completed the survey, which does not support this hypothesis. It is also possible that social desirability bias existed in this context, which would suggest that the levels of bias were higher than reported. We attempted to reduce this possibility by anonymizing the survey. We noted that our reliance on cross-sectional data integrated with existing county-level data did not allow us to establish causal relationships. Finally, in addition to county-level contextual variables, state-level policies and practices may be related to biases among physicians but could not be assessed here. Future researchers are encouraged to use additional research designs and samples from other regions to further explore the relationships found in this study.

Conclusions

The effects of the opioid epidemic have been felt nationwide but have been particularly pronounced in rural communities in the United States. Our findings revealed that physicians working in these same hard-hit rural communities also reported more negative attitudes toward persons and patients who misuse opioids. Given disparities in access to opioid use disorder and harm reduction services between rural and urban areas, the increased bias in counties lacking these services suggests that rural patients who misuse opioids face numerous challenges to finding effective treatment. Our findings underscore the importance of bias reduction efforts that target rural communities in which these efforts may have the most pronounced impact.