Mitigating the Risk of Infectious Diseases Among Rural Drug Users in Western North Carolina: Results of the Southern Appalachia Test, Link, Care (SA-TLC) Health Care Provider Survey

Abstract

Purpose:

To characterize how health care providers in western North Carolina (NC) manage patients with substance use disorders and to inform strategies for preventing injection drug use (IDU)-associated outbreaks of bloodborne infectious diseases.

Methods:

We collected data on practice characteristics, provider sociodemographics, and attitudes and beliefs about hepatitis C virus (HCV), human immunodeficiency virus (HIV), opioid use, and IDU via online survey. Providers in 8 counties of western NC were invited to participate by email. Results were analyzed using descriptive and bivariate statistics.

Findings:

Of 84 respondents participating between 30 July and 3 December 2018, 81% were practicing clinicians and 46% served a county identified as being vulnerable to IDU-associated outbreaks of HCV or HIV. A substantial proportion was unsure about injecting behaviors among patients. Scores reflected comfort working with opioid users, though this varied by medical specialty. One-quarter of respondents “never” discussed harm reduction or HCV treatment with patients known to inject drugs; 22% “never” discussed HIV screening with injectors; and 1 in 3 referred at-risk patients out for HCV or HIV testing rather than ordering a test themselves. Scores indicated low levels of stigma toward persons living with HCV or HIV. Respondents identified HIV treatment, HCV treatment, and liver disease management as training needs.

Conclusions:

Our findings provide insights to inform health infrastructure improvement, with the goal of preventing HCV or HIV outbreaks in southern Appalachia. Rural health care workers are willing to receive additional training if it can improve care for patients affected by substance use disorders.

Persons who inject drugs (PWID) are at markedly increased risk for infections from bloodborne pathogens, especially hepatitis B virus (HBV), hepatitis C virus (HCV), and human immunodeficiency virus (HIV).^1–5 Unlike earlier epidemics of these infections among PWID, new cases are accumulating among users outside urbanized areas.^6,7 Indeed, at least half of American PWID reside in rural areas with limited health care infrastructure, raising the risk for outbreaks among networks of rural injectors and other drug users.^8,9 Such a scenario occurred in late 2014 in Scott County, a rural community in southeastern Indiana,¹⁰ with over 188 persons diagnosed with HIV over the course of a months-long epidemiological investigation; 90% were co-infected with HCV. In response to the Scott County outbreak, the Centers for Disease Control and Prevention (CDC) identified 220 counties across the United States (US) that were vulnerable to similar transmission events and outbreaks. By far, the greatest density of at-risk counties were located in rural Appalachia.⁶

Multiple hotspots of acute HCV infection are occurring among PWID under age 30 and living in non-urban, resource-poor areas in central Appalachian states.⁹ In recent years, the number of new HIV infections among PWID in the US has leveled off after years of steady decline—raising alarm for a potential upswing driven by rural injection drug use (IDU).¹¹ Young, non-urban, white injectors represented the largest numbers of HIV infections attributable to IDU from 2010-2016.¹¹ In these non-urban areas, services to prevent (eg, medication-assisted therapy) and treat HIV and HCV infections are limited.¹²

In the CDC’s vulnerability assessment, 5 counties in North Carolina (NC) were identified as being at increased risk for an HIV or HCV outbreak.⁶ This led the NC Department of Health and Human Services (DHHS) to conduct its own assessment using more detailed, state-specific data, including emergency department visits for opioid overdoses, cases of acute and chronic HCV, and Medicaid records. Through this assessment and related stakeholder interviews, multiple additional counties were identified, with a majority located in the far west (part of southern Appalachia).¹³

The intersection of the opioid and HCV epidemics in NC is particularly sobering.¹⁴ Acute HCV in NC increased 3.5 fold from 2010 to 2017.¹⁵ Of 50 acute HCV reports in 2014 with complete case data, 42 (84%) indicated IDU as the most likely source of infection. From 2006-2015, the mean age of acute HCV cases decreased from 40 to 34 years, with an increasing proportion among white, non-Hispanic individuals. Five-year (2010-2014), county-specific acute HCV incidence rates were highest in western NC, with a top rate of 85.5 cases per 100,000 persons.^16,17 Considering the high frequency of asymptomatic illness, lack of a specific diagnostic test for acute HCV, and likely underreporting, acute HCV infection cases are almost certainly underestimated.¹⁸ Despite the HCV burden in and the assumed vulnerability of this area in NC, health care services and practice patterns related to persons who use injection and non-injection opioids in far western NC are not well defined. This information is necessary to develop health care and public health activities for maximum impact on these epidemics.

As part of efforts to mitigate the risk of an IDU-associated outbreak of infectious diseases in southern Appalachia, the CDC, National Institute on Drug Abuse, Appalachian Regional Commission, and Substance Abuse and Mental Health Services Administration partnered to fund innovative projects designed to assess service needs and address gaps identified. Funded under this initiative, our team of academic, public health, non-profit, and research institutions in NC conducted an in-depth assessment of the health care infrastructure serving PWID in 8 counties identified by NC-DHHS as being at risk for an outbreak. As part of this initiative, we developed and conducted an online survey of health care providers in those counties to characterize their approach to patients with substance use disorders and identify key areas for training and development to address infectious disease risk in this region of southern Appalachia.

Methods

Rurality of the Project Area

Eight counties of far western NC were selected for study based on NC-DHHS analyses¹³: Cherokee, Clay, Graham, Haywood, Jackson, Macon, Swain, and Transylvania. Based on Federal Office of Rural Health Policy definitions,¹⁹ the entire area of all but one of these counties (Haywood) is considered rural.

Selection and Adaptation of Specialized Instruments

We undertook a literature review to identify validated scales for characterizing provider attitudes and beliefs related to HCV, HIV, opioid use, and injection drug use. Because one’s personal beliefs may influence professional behaviors, we also sought an instrument to succinctly capture social conservatism. We ultimately selected 6 scales comprising 47 items (Supplementary Material, available online only):

1. Health Care Professional Attitudes Toward Caring for People with HCV.²⁰ We used 10 items from a larger instrument developed in Australia for gauging how health care workers felt about working with PWID and HCV-infected persons. Each item was scored on a 5-point scale (“strongly agree” to “strongly disagree,” with an “uncertain” midpoint), and we created average scores for each of 4 subscales: compassion toward people with HCV (3 items); willingness to provide care for people with HCV (3 items); fear of contracting HCV (2 items); and attitudes toward PWID (3 items). Lower scores indicate greater compassion, greater willingness to provide care, greater fear of HCV, and greater hostility toward PWID, respectively.

2. HIV Stigma and Discrimination Scale (SDS).²¹ This instrument (Cronbach’s α = 0.818) was developed among health care workers in Alabama and Mississippi to describe HIV-related stigma that might be directed toward patients. For this study, we used the “Attitudes Toward People Living with HIV” subscale, which was scored using a 4-point scale (“strongly disagree” to “strongly agree”; no midpoint). Higher scores indicate greater hostility toward people living with HIV.

3. Attitudes Toward Injection Drug Users Scale (ATIS).²² For this study, we used the 5 health care provider-specific items (Cronbach’s α = 0.57) from the larger, 10-item ATIS – an instrument designed to gauge bias against PWID. Each item was scored on a 5-point scale (“strongly disagree” to “strongly agree,” with a “neither agree nor disagree” midpoint), and scores were summed to create an overall ATIS score for each participant. Summary scores could range from 5 to 25, with higher scores indicating more favorable attitudes toward PWID.

4. Perceptions of Controllability of Drug Use.²² Four items assessing perceptions of the ability of PWID to manage their addiction were scored on a 5-point scale (“strongly disagree” to “strongly agree,” with a “neither agree nor disagree” midpoint; Cronbach’s α = 0.75). Scores were summed (range 4-20), with higher scores indicating more negative or hostile perceptions of drug use among PWID.

5. Short Alcohol Attitudes Problem Perception Questionnaire (SAAPPQ).^23,24 This 10-item instrument (Cronbach’s α = 0.7-0.9) uses a 7-point scale (“strongly disagree” to “strongly agree”) to capture provider sentiments related to working with alcohol abusers in 5 domains: role adequacy, role legitimacy, motivation, work satisfaction, and task-specific self-esteem. The role adequacy and legitimacy domains are then combined into a “role security” score, reflecting how comfortable a provider is in performing professional tasks related to alcohol use. The motivation, work satisfaction, and task-specific self-esteem domains are combined into a “therapeutic commitment” score, indicating how committed a provider is to caring for persons with alcohol use disorders. For this study, we adapted the SAAPPQ by replacing alcohol use with opioid use. For example, “Pessimism is the most realistic attitude to take toward drinkers” was replaced with “Pessimism is the most realistic attitude to take toward opioid users.” Scores for role security and therapeutic commitment could range from 4-28 and 6-42, respectively. Higher scores indicate greater role security or therapeutic commitment.

6. Social Conservatism.²⁵ We used a 7-item subscale (Cronbach’s α = 0.87) of the longer, 12-item Social & Economic Conservatism Scale, asking respondents to indicate how positively they feel toward specific topics such as “abortion,” “traditional marriage,” and “patriotism,” with scores ranging from 0 (very negative) to 100 (very positive). Higher values indicate greater social conservatism (range 0-700).

Assessment of Provider and Practice Characteristics

In addition to the instruments above, we developed items to capture provider age, gender, race/ethnicity, years in practice, professional certification, role in their practice (eg, administrator, counselor, clinician), and which county (or counties) of the 8 in the project area the provider served. We also asked respondents to describe their practice type (eg, hospital, public health department, office-based single-specialty), practice size (as determined by number of employees rather than number of patients), specialty (eg, family medicine, pediatrics, obstetrics), and whether they or anyone at their practice offered medication-assisted therapy (MAT) for opioid use disorders. Respondents were asked to estimate the proportion of their patients engaging in specific types of opioid and non-opioid drug use, including injection behaviors (“None,” “1%-25%,” “26%-50%,” “51%-75%,” “76%-100%,” or “I’m not sure”). Finally, we asked respondents to assess their knowledge of HCV and HIV screening and treatment guidelines and to identify training needs.

Online Survey

The above items were programmed into an online format using dedicated software (Qualtrics, Provo, UT) under a site license held by the University of North Carolina at Chapel Hill (UNC-CH). We then sent individualized survey invitations to providers serving the project area via email, using lists of licensees and practitioners provided by the NC Department of Health & Human Services, the NC Board of Nursing, the NC Medical Board, and the local health directors of the 8 counties of the study region. An online $20 Amazon gift card was offered as compensation for time spent taking the survey. The study was approved by the Institutional Review Board at RTI International.

Data Analysis

Data were de-identified by the administrative team at RTI International before transmitting to the project collaborator for analysis. Summaries of provider characteristics and their assessments of patient drug use behaviors were generated using descriptive statistics. Relationships between provider, practice, and patient characteristics and scales reflecting attitudes toward HCV, HIV, PWID, and opioid use behaviors were explored using bivariate analyses (Pearson’s χ² and Fisher’s exact tests for categorical variables; Wilcoxon rank sum and Kruskal-Wallis tests for comparisons with ordinal or continuous outcomes). Statistical significance was set at α = .05, and all analyses were performed using Stata version 11.2 (StataCorp, College Station, TX).

Results

In all, 389 providers were invited to participate by email. Between 30 July and 3 December 2018, 84 respondents (21.6%) answered at least part of the survey and 77 (19.8%) completed it entirely.

Provider and Practice Characteristics

Most respondents were female (n=51, 67%), at least 45 years old (n=49, 64%), and white (n=68, 88%) (Table 1). Eighty-one percent of participants were involved in direct patient care, comprising 39 physicians, 28 nurse practitioners, and 1 physician assistant. Eighteen persons had entered the workforce within the past 5 years (21%), and 51 respondents had been in health care for more than 20 years (61%). Just under half (46%) served one of the 5 counties identified by NC-DHHS as being highly vulnerable to an IDU-associated outbreak of HCV or HIV in its 2016 analysis (Cherokee, Clay, Graham, Haywood, and Swain), and about 1 in 5 (21%) practiced in a county (Cherokee, Clay, and Swain) with a high incidence of acute HCV in 2016 (>15 cases per 10,000 population; unpublished NC-DHHS data).

Table 1.

Characteristics of Providers Responding to Southern Appalachia Test, Link, Care (SA-TLC) Survey, 30 July-3 December 2018

Characteristic	N	%
Gender (n=76)
Female	51	67.11
Male	25	32.89
Age (n=77)
25-34	9	11.69
35-44	19	24.68
45-54	16	20.78
55-64	20	25.97
65+	13	16.88
Race/Ethnicity (n=77)
American Indian / Alaska Native	3	3.90
Black or African American	3	3.90
Hispanic / Latinx	2	2.60
Multiple	1	1.30
White / Caucasian	68	88.31
Years in Practice (n=84)
5 years or less	18	21.43
More than 5 years	66	78.57
20 years or less	33	39.29
More than 20 years	51	60.71
Practice Type (n=84)
Community / rural / federally qualified health center	15	17.86
Emergency department / urgent care	16	19.05
Hospital (inpatient)	4	4.76
Nursing facility / hospice	4	4.76
Office-based, multispecialty	7	8.33
Office-based, single specialty	29	34.52
Public health department	9	10.71
Practice size (n=84)
1-5	16	19.05
6-10	13	15.48
11-30	32	38.10
31-100	16	19.05
More than 100	7	8.33
Specialty (n=84)
Family medicine	21	25.00
Pediatrics & Obstretics-Gynecology	11	13.10
Medicine	20	23.81
Surgery	5	5.95
Mental health	11	13.10
Emergency medicine / urgent care	10	11.90
Provider offers medication-assisted therapy (n=84)
No	67	79.76
Yes	17	20.24
Practice offers medication-assisted therapy (n=84)
No	51	60.71
Yes	28	33.33
Not sure	5	5.95

Among all respondents, the most common practice venues were single-specialty, private offices (n=29, 34%), followed by emergency / urgent care settings (n=16, 19%), and community, rural, or federally qualified health centers (n=15, 18%). About a third (n=29; 34%) provided services in a practice with ≤10 employees; 27% (n=23) had 30 or more co-workers in their organization. Family medicine (n=21; 25%) and internal medicine (n=20; 24%) had the greatest representation in the sample; only 5 respondents (6%) were surgeons. A large majority of respondents (n=67; 80%) did not offer MAT, and 61% (n=51) did not have any coworkers offering MAT at their practice location.

Patients and Management Strategies

Generally, the majority of respondents felt that only a small proportion of their patients were using opioids (of any type and by any method; Supplementary Table 1, available online only). However, as they were asked to provide estimates for more specific subpopulations, such as the proportion of injectors using non-prescription opioids, the frequency of answering “not sure” increased.

With patients who were known to inject drugs, 26% of respondents “never” discussed harm reduction strategies (n=21), 25% “never” discussed screening or treatment for HCV (n=20), and 22% “never” discussed HIV screening (n=17). For a patient who injects drugs but is untested for HCV, just over half of respondents indicated they would order an HCV test themselves (n=44; 54%) and another third either encouraged testing or referred elsewhere for a test (n=28; 35%). Primary care providers (n=40; 49%) and health departments (n=29; 36%) were the most common venues to which injectors were directed for HCV testing. Nearly identical responses were noted for management of injectors without HIV testing: 54% of respondents would order a test themselves (n=43), 35% would send the patient elsewhere (n=28). Similarly, primary care offices (n=38; 49%) and health departments (n=29; 37%) were the most common locations for HIV testing of injectors. Only 7 respondents (9%) indicated they would treat an injector for HCV themselves; none indicated they would treat an injector for HIV. Preferred destinations for HCV- or HIV-infected injectors to be treated are shown in Supplementary Table 1 (available online only), with large academic referral centers, primary care practices, and local health departments or community health centers being the most popular options.

Attitudes Toward Working with Opioid Users

The role security and therapeutic commitment subscales of our opioid-adapted version of the SAAPPQ demonstrated acceptable internal reliability, with Cronbach’s α of 0.56 and 0.76, respectively. Among all respondents, role security was high (reflecting how comfortable providers feel performing professional tasks related to opioid use), with an overall median score of 23 (interquartile range [IQR] 21-26) on a scale with a maximum possible value of 28 (Table 2). There were no significant differences in median scores when assessed by gender, age, race, years in practice, or practice location. When compared dichotomously against all other provider types, family medicine providers had significantly greater role security (median 24 vs. 22; P = .015), while a combined category including pediatricians and obstetrician-gynecologists felt significantly less secure in their role when working with opioid users (median 21 vs. 24; P = .009). Role security tended to be lower among respondents who indicated uncertainty about the proportion of their patients using any opioids, compared with those who gave an actual estimate (median 20.5 vs. 24.0; P = .06). Scores were significantly lower among respondents unsure of the proportion of patients using opioid pills or patches (median 21 vs. 24; P = .018), as well as among those unsure of the proportion of injectors using opioids (median 21 vs. 24; P = .003), non-opioids (median 22 vs 24; P = .042), or prescription opioids (median 22 vs. 24; P = .034).

Table 2.

Scores on an Opioid-adapted Version of the Short Alcohol Attitudes Problem Perception Questionnaire (SAAPPQ), by Provider and Practice Characteristics, Southern Appalachia Test, Link, Care (SA-TLC) Survey, 30 July-3 December 2018

	Opioid-Adapted Short Alcohol Attitudes Problem Perception Questionnaire (SAAPPQ)
	Role security				Therapeutic commitment
	N	Median	IQR	P	N	Median	IQR	P
Overall	78	23	21-26	–	76	28	23-31	–
Gender
Female	51	23	20-25	.613	49	27	22-31	.238
Male	25	22	21-26		25	29	24-31
Age
25-34	9	22	18-24	.577	9	30	28-31	.105
35-44	19	23	21-25		19	25	21-28
45-54	16	23.5	21.5-26		15	29	24-35
55-64	20	22.5	19.5-25		20	27	22-30.5
65-74	12	24	2.5-26		11	30	26-38
75-84	1	21	21-21		1	26	26-26
Race
Non-white	9	23	22-25	.937	9	25	21-31	.613
White	68	23	20.5-26		66	28	23-31
Years in practice
More than 5 years	61	23	21-26	.362	59	27	23-30	.117
5 years or less	17	23	20-24		17	31	25-37
More than 20 years	47	23	21-26	.689	46	27	22-31	.710
20 years or less	31	23	20-26		30	28	23-31
County served
Highly vulnerable to IDU-associated HIV or HCV outbreak	36	24	21-26	.456	35	28	24-31	.407
High opioid overdose frequency	34	24	21-26	.540	33	29	24-31	.332
High HCV incidence	18	24	23-26	.567	18	29.5	23-31	.638
High HCV prevalence	19	24	22-26	.661	19	29	23-31	.737
Haywood County	19	24	20-26	.911	18	29	25-37	.152
All Other Counties	58	23	21-26		58	27.5	22-31
Specialty
Family medicine	21	24	23-26	.015	20	27.5	23.5-32	.981
Pediatrics / Obstretics-Gynecology	9	21	14-22	.009	9	26	22-30	.425
Medicine	18	23	22-26	.703	17	23	21-28	.0078
Surgery	5	21	20-21	.095	5	23	23-26	.197
Mental health	11	24	22-26	.270	11	38	35-41	.0003
Emergency medicine / urgent care	8	21.5	20-25.5	.637	8	28.5	23.5-30	.839
Unsure of proportion of patients…
Using any opioids	10	20.5	20-23	.058	10	26	22-28	.313
Using opioid pills/patches	2	21	19-23.5	.018	19	27	23-32	.923
Among patients injecting, unsure of proportion…
Injecting opioids	26	21	19-23	.003	26	25	22-30	.025
Injecting non-opioids	27	22	19-24	.042	26	26	22-30	.115
Injecting prescription opioids	40	22	19.5-25	.034	40	27	22-32	.567
Injecting non-prescription opioids	32	22.5	19.5-24.5	.070	30	26	22-30	.063

Therapeutic commitment to working with opioid users was not nearly as high as role security, with an overall median score of 28 (IQR 23-31) on a scale with a maximum possible score of 42. We did not observe any differences in median scores when compared by gender, age, race, years in practice, or practice location. Managers and administrators had significantly higher scores than non-managers (median 38.5 vs. 27.5; P = .046), but clinicians had significantly lower scores (median 27.0 vs. 30.5; P = .036). When compared against all other specialties using an indicator (dummy) variable, internal medicine providers had significantly lower therapeutic commitment scores (median 23 vs. 29; P = .008), while mental health providers had markedly higher scores (median 38 vs. 27; P = .0003). Therapeutic commitment was lower among respondents indicating uncertainty about the proportion of injectors using any opioids (median 25 vs. 29; P = .025) and trended lower among those uncertain of the proportion injecting non-prescription opioids (median 26 vs. 28; P = .063).

Social Conservatism

The median score among all respondents was 470 (IQR 390-570) on a scale with a possible range of 0-700, suggesting that the sample was slightly “right-of-center” in terms of socially conservative beliefs. Female respondents were more conservative than male respondents (median 510 vs. 420, P = .027), but there were no other statistically significant differences in median scores across any other comparison group.

Attitudes Toward Persons with HCV and HIV

Respondents generally had low scores indicating stigma or negative attitudes toward people living with HIV, with a median score of 9 (IQR 8-11) on a scale with a maximum of 24 (Supplementary Table 2, available online only). There were no characteristics associated with significantly higher or lower scores in any bivariate comparison. Scores reflecting compassion toward persons living with HCV (median 2, IQR 1.25-2.00) and willingness to provide care (median 2.66, IQR 2.33-3.00) suggested respondents did not harbor negative attitudes toward this population (Supplementary Table 3, available online only). Although there were no differences in compassion for persons living with HCV across all comparisons, respondents in practice for 5 years or less (median 2.33 vs. 2.66; P = .029), practicing in a high chronic HCV prevalence county (median 2.33 vs. 2.66; P = .031), non-clinicians (median 2.33 vs. 2.66; P = .049), and non-emergency medicine providers (median 2.66 vs. 3.00; P = .007) had significantly greater willingness to provide care. Overall the fear of acquiring HCV was low in the sample (median 4.5, IQR 4.0-5.0; higher scores reflect lower fear), and nurses were less fearful of HCV acquisition than non-nurses (median 4.5 vs. 4.0, P = .044). No other groups had significantly different scores.

Attitudes Toward Injection Drug Users

Respondents had greater equipoise with respect to attitudes and beliefs around drug use (Supplementary Table 4, available online only). On the ATIS scale for health care providers, the median score was 12 (IQR 11-13) on a scale with a range from 5-25; higher scores indicate more favorable attitudes toward PWID. Family medicine specialists had a significantly lower score (median 11 vs. 13; P = .014), suggesting less favorable attitudes toward PWID than other specialties. Hostility against PWID was low among respondents (median 4.33, IQR 4.00-4.66) on a scale with higher scores reflecting lower hostility (maximum 5), and no group had a significantly higher or lower score than any other. When asked about their perceptions of the controllability of drug use among users, the overall median 10 (IQR 8-11) was near the midpoint of the possible range of scores (4-20). No subgroups had significantly different scores in our analyses.

Self-assessment of HCV and HIV Knowledge and Related Training Needs

Fewer providers felt they had adequate knowledge of HCV screening and treatment than for HIV screening. One third of respondents indicated that they knew “nothing” or “a little bit” about HCV screening and treatment (n=28, 34%). Thirty-nine percent responded similarly for HIV screening knowledge (n=31). When offered specific training options on HCV and HIV, the most frequently selected responses suggested a desire to learn more about HIV treatment (n=51, 65%), HCV treatment (n=50, 64%), and the management of liver disease (n=31, 40%).

Evaluation of Haywood County Providers

Because Haywood was the only county in the project area that was not entirely rural, we compared demographics and attitudinal scores between providers working in Haywood County and those who did not work there. No significant differences were observed in provider characteristics (data not shown) or in attitudes with respect to the patient population (Table 2 and Supplemental Tables 3 and 4).

Discussion

To our knowledge, this is the first provider-level assessment of health care services and practice patterns related to persons who use injection and non-injection opioids in far western NC, a region of southern Appalachia at increased risk for an IDU-driven outbreak of infectious diseases.¹³ Our findings provide key insights to help guide efforts at strengthening health service infrastructure—with an ultimate goal of mitigating the risk of a Scott County-like outbreak of HCV or HIV in the Appalachian area of the state. Assuming that respondent attitudes reflect those of providers across the 8-county project area, it appears that health care workers in western NC are willing to receive additional support and training if it means that care can be improved for patients affected by the opioid epidemic.

We included scales assessing social conservatism and attitudes toward persons with HIV, HCV, or substance use disorders to discern whether providers in this remote, rural area might harbor more negative feelings toward patients with these conditions. It was reassuring to see very little antipathy toward persons living with HIV or HCV among respondents, and that conservative views did not seem to have any significant impact on practice patterns or interactions with patients. Attitudes toward PWID seemed less favorable but were far from hostile. Being “on the fence” about their personal feelings with opioid users was reflected in some providers’ lack of role security in their professional work with this population. This suggests that providers in this area of the state have yet to develop personal or practice-level strategies for dealing with an increasing number of opioid users in their practices.

Broadly speaking, health care workers tend to shy away from discussions about “private” concerns such as sexual behavior²⁶ or illicit drug use^27,28 unless it is the main reason for a visit. Indeed, approximately 1 in 4 never discussed harm reduction strategies or HCV screening with their patients, and a fifth never discussed HIV screening. Providers also commonly reported being unsure of the number of their patients using drugs, with the number of uncertain respondents increasing when it came to questions about injection behaviors. Studies of provider communication suggest that reticence to engage patients in discussions about sensitive subjects is rooted in a lack of education about communication strategies and best practices.^28,29 Thus, educational interventions focused on increasing provider self-efficacy in addressing sexual and drug use behaviors might be a particularly high-yield intervention for providers in areas dealing with increasing opioid use.

Current practice patterns seem to favor external referrals for key services such as screening and treatment for HCV and HIV, but providers preferred to send their patients to local colleagues such as primary care providers, community health centers, and health departments. Considering enthusiastic responses indicating a desire to learn about the treatment of HCV, HIV, and liver disease, it is clear that respondents to this survey are interested in improving the local capacity to provide services for patients with HCV and HIV infection who use drugs.

Limitations

Our study has several key limitations. Trying to identify short, validated instruments amenable to distribution in an online format was a challenge. We opted to adapt a well-characterized scale assessing provider interactions with alcohol users (SAAPPQ)^23,24 to gauge opioid-specific attitudes and beliefs. Though we do not have a validation of this opioid-adapted version in other settings, the responses proved especially useful for understanding the importance of role security and legitimacy in caring for this population. The survey was distributed through email lists provided by professional organizations, licensing boards and local health departments, and an incentive was offered for participation, but the response rate was lower than we had hoped. Anecdotally, our team was aware that multiple, similar surveys launched around the same time as ours, and we felt “survey fatigue” was the most likely explanation for this response rate. We were nonetheless able to get a cross-sectional view of health care providers serving these communities. It did not appear that those who participated did so to convey a specific viewpoint or attitude about the subject of opioid use, since responses were not especially polarized. Our survey findings resonate with an “all-hands-on-deck” sensibility among providers with whom we have engaged across NC. The severity of opioid-related adverse events they and their patients are experiencing is a strong motivator to learn more and improve the quality of services delivered.

Conclusion

In summary, this survey of providers in far western NC provides key, initial insights for our consortium of academic, public health, and non-profit organizations as we work with NC-DHHS to improve health care infrastructure in this region. Parallel surveys were conducted simultaneously among pharmacists and substance use disorder providers in the same 8 counties, and as we conclude analyses of those data we anticipate deepening our understanding of commonalities among providers across disciplines. Generally, survey participants seemed receptive to additional education and support, and their responses have enabled us to develop a roadmap for training on topics in the treatment and prevention of infectious diseases. It is our hope that strengthening the health care systems available to persons using opioids will have ripple effects throughout adjacent areas and provide a model for other areas of the state facing the challenge of the opioid epidemic.