Abstract
Introduction:
Opioid agonist treatment (OAT) is the only treatment for opioid use disorder (OUD) proven to reduce overdose mortality, yet access to this evidence-based treatment remains poor. The purpose of this cross-sectional audit study was to assess OAT availability at residential substance use services in North Carolina.
Methods:
We conducted a state-wide inventory of residential substance use service providers in North Carolina and subsequently called all providers identified, posing as uninsured persons who use heroin, seeking treatment services. Program characteristics, as reported in phone calls, were systematically recorded. We used Fisher’s exact tests to assess what program characteristics were associated with OAT availability and with staff making discouraging comments about OAT. We used unsupervised agglomerative clustering to identify facilities with similar characteristics.
Results:
Of the 94 treatment providers identified, we successfully contacted and collected data from 66. Of those, only 7 (10.6 %) provide OAT on site; an additional 9 (13.6 %) allow OAT through an outside or community-based prescriber. Only 8 (12.1 %) providers were licensed to provide residential substance use treatment. Staff from 33 (50.0 %) providers made negative, discouraging, or stigmatizing remarks about OAT—for example, that OAT substitutes one addiction for another or does not constitute “true recovery.” OAT availability was positively associated with a provider holding a state license for any substance use-related service (41.9 % vs 8.6 %, p =0.002) and offering 12-step programming (36.1 % vs. 10/0 %, p = 0.020). OAT availability was negatively associated with faith-based programming (6.1 % vs 42.4 %, p = 0.001), dress codes (5.3 % vs 50.0 %, p < 0.001), and mandates that residents work in a provider-owned and -operated commercial enterprise (5.0 % vs 32.6 %, p = 0.026). Cluster analysis revealed that the most common (n = 21) type of service provider in North Carolina is an unlicensed, faith-based organization that prohibits OAT, imposes a dress code, and mandates that residents work, often in provider-owned and -operated commercial enterprises.
Conclusion:
Evidence-based treatments for OUD are largely unavailable at providers of residential substance use services in North Carolina. The prohibition of OAT occurs most often among providers who are unlicensed and impose labor and/or 12-step mandates on residents. Changes to state licensure requirements and exemptions may help improve OAT availability.
Keywords: Addiction, Opioid use disorder, Treatment, Buprenorphine, Opioid agonist treatment
1. Introduction
In 2022, approximately 110,000 people in the United States died from drug overdose, >80,000 of which were opioid-involved (Ahmad et al., 2024). The treatment of opioid use disorder (OUD) with opioid agonist treatments (OAT) methadone or buprenorphine reduces the risk of fatal overdose by nearly 60 % and reduces all-cause mortality by nearly half (Larochelle et al., 2018). Other treatment modalities for OUD that do not include OAT medications, including treatment with the opioid antagonist medication naltrexone, have been associated with increased risk of fatal overdose (Ajazi et al., 2021; Heimer et al., 2024). Yet only one in eight people diagnosed with OUD receive OAT (Wakeman et al., 2020), with Black and Hispanic persons and criminal justice-involved persons living with OUD even less likely to receive OAT compared to their respective counterparts (Stahler et al., 2021, 2022).
The state of North Carolina recorded >4000 overdose fatalities in 2022, a 70 % increase since 2019 (U.S. Centers for Disease Control and Prevention, 2023; North Carolina Department of Health and Human Services, 2023b). That increase disproportionately affected Black and Hispanic populations, for whom overdose rates rose more than twofold during the same period (North Carolina Department of Health and Human Services, 2023a). Available data indicate that OAT remains widely unavailable to North Carolina residents. As recently as 2022, 30 % of North Carolina counties lacked a buprenorphine prescriber, and 75 % of counties lacked an opioid treatment program offering methadone (Donnelly-DeRoven, 2022).
The American Society of Addiction Medicine (ASAM) defines residential treatment for substance use disorders (SUDs) as “organized treatment that features a planned and structured regimen of care in a 24-hour residential setting” with supportive recovery services on-site and clinical services—including access to medications for OUD—made available on-site or in the immediate community (American Society of Addiction Medicine, 2023). Laypersons living with OUD generally perceive residential treatment as efficacious and desirable (Hay et al., 2019). However, many residential service providers do not ensure access to OAT or even prohibit OAT on the basis of false claims that these medications are unsafe (Healing Transitions, 2022) or constitute a “replacement” addiction (Yow & Washo, 2022). An audit of residential treatment programs in Ontario, Canada, found that approximately one in three did not allow the use of methadone for OAT, and approximately two in three did not allow OAT with buprenorphine (Ali et al., 2023). Similarly, a recent U.S.-based study found that fewer than one-third of residential treatment programs in a nationwide sample allow residents to access any form of OAT and that more than one-fifth of those programs discourage people inquiring about the program from seeking these medications (Beetham et al., 2020). In North Carolina, tens of millions of public dollars have been allocated to high-profile residential service providers (both licensed and unlicensed) that maintain policies barring participants from utilizing OAT (Goodloe-Murphy, 2023; Knopf, 2022; Pattani & Knopf, 2022), raising questions about the current and future availability of OAT for the state’s residents.
To assess OAT availability at providers of residential substance use services in North Carolina, we conducted a state-wide inventory of residential substance use service providers who advertise their services to members of the public seeking care for SUD. We subsequently carried out an audit study—sometimes called “secret shopper studies” (Rankin et al., 2022)—in which two trained staff persons made scripted calls to all known providers, posing as adults in their late twenties self-reporting daily heroin use, seeking residential services, and inquiring about the nature of services offered by each provider.
2. Methods
2.1. Study design
This audit study collected information on the services and program characteristics of residential substance use services providers in North Carolina. The study team collected data through a review of provider websites as well as through phone calls in which trained research staff posed as persons seeking treatment for OUD. The Institutional Review Board at North Carolina State University determined our protocol to be exempt from review.
2.2. Providers of residential SUD services in North Carolina
In this study, we defined a residential substance use services provider as any entity advertising residential, non-hospital SUD services for adults within the state of North Carolina. Programs not available to the general population (i.e. those only accepting court-ordered participants or operated by the Veterans Association) were excluded. Programs that only provided short-term detoxification services and/or transitional living opportunities for persons who have recently exited a residential treatment (i.e. “halfway houses”) were also excluded. Programs advertising residential services to potential clients as a first and primary step toward the resolution of substance use disorder (often immediately following detoxification) were included regardless of whether or not the program explicitly used the word “treatment” to describe available services in program materials. Discussion among the entire study team resolved any ambiguities about whether to include a program in the sample based on how a program advertised itself.
2.3. Identification of service providers
In January and February 2023, we identified eligible service providers using several public data sources. First, we utilized the treatment locator tool hosted by the U.S. Substance Abuse and Mental Health Services Administration (SAMHSA), including any programs located in North Carolina listed under one or more of the following categories: Residential/24-hour residential; Long-term residential; Short-term residential; and Hospital inpatient/24-hour hospital inpatient treatment (included to capture potentially misclassified providers). Second, we searched all facilities listed under the heading “Mental Health and Substance Use Disorder Services” in each county on the NC211.org website, a community resource locator operated by United Way. Third, we iteratively queried the Google search engine with one of the following search terms (adapted from those used by Beetham et al. (2021)) and the name of one North Carolina county until all county and search term combinations were exhausted: “rehabilitation drugs;” “drug rehab centers;” “drug rehab near me;” “addiction treatment center;” “substance abuse rehabilitation program;” “rehab;” “heroin treatment;” “opiate detox programs;” “detox for heroin;” and “suboxone detox.” Programs were included if they appeared within the first 10 pages of search results. Fourth, we included all programs registered (as of January 2023) with the North Carolina Division of Health Services Regulation under one or more of the following categories: “Residential Treatment – Individuals with Substance Abuse Dependency;” “Residential Recovery Programs for Individuals with Substance Abuse Disorders and their Children;” and “Supervised Living for Adults with Substance Abuse Dependency.” Fifth, we included all organizations listed in the online provider directory maintained by the North Carolina Association for the Treatment of Opioid Dependence. Sixth, we included all residential programs listed on the Bridge to 100 website, a privately-maintained directory of faith-based residential substance use service providers in North Carolina. Finally, we identified two additional organizations via program directories listed on the websites of providers already identified by the research team.
This initial search returned listings for 321 providers. Of these, 192 were identified as duplicates and removed. Another 12 were found to be operated by the Veteran’s Administration or state and regional hospital systems and removed. We reviewed the websites of the remaining 117 providers and removed an additional 28, which we found to be ineligible (serving children and adolescents only, offering outpatient treatment services and/or crisis stabilization only) or to have no website or other publicly available contact information that we could find. The remaining list of 89 providers was then circulated on a state-wide listserv for harm reduction, peer support, and recovery advocacy organizations. Listserv members identified five eligible providers not included on the list. The subsequent list of 94 providers constitutes our initial sample.
2.4. Data collection
Before undertaking the audit survey, study staff (JC, SD, and BM) reviewed treatment provider websites to collect information about facility location, demographics served, and whether the treatment provider represented themselves as faith-based. Our audit protocol, implemented between February and May 2023, was designed to mirror that utilized by Beetham et al. in their audit study of a nationwide sample of residential treatment providers (2021). In this study, trained research staff (SD and BM) called each of the 94 treatment providers included in our sample via publicly listed contact or intake numbers. When calling, research staff followed a standardized script (see Appendix A) adapted from that previously developed by Beetham et al. (2021). Specifically, research staff posed as a 27-year-old resident of North Carolina who was a daily heroin user, lacked health insurance, and was seeking residential SUD services.
Prior to data collection, a member of the study team (JC) trained the callers, a white man and a white woman in their late 20s, who speak American English as their first language. Training consisted of background education on OUD and various modalities of treatment (including OAT), pilot calls to treatment providers outside of North Carolina to test the script, debriefs from these calls with the entire research team, and regular discussion of questions or ambiguities in the script. This group (JC, SD, and BM) met weekly to debrief throughout data collection, and the entire study team debriefed at least monthly.
All providers identified as offering gender-specific services were assigned to the caller of that gender; the remaining providers were divided randomly between the callers until the call loads were approximately equal. If providers operated more than one residential site, callers contacted only one of those sites, selected according to the caller’s gender (i.e., men’s campus or women’s campus) or randomly. Providers were removed from the sample if: they were discovered during the call to be ineligible for inclusion in this study; the call revealed the program had permanently closed; or the program could not be reached after three separate call attempts between 9 am and 5 pm on three different business days. All data was captured with a standardized data-collection instrument designed on Qualtrics (Seattle, WA), which was piloted and refined with the call script.
2.5. Study measures
Key treatment variables captured during calls included: availability of at least one form of OAT (methadone or buprenorphine) (offered for maintenance/not offered by the program but allowed from an outside prescriber/not allowed for maintenance while at the program); whether the program offered naltrexone (yes/no/unsure or don’t know); and whether the provider offered detoxification services or made them available through a partnering agency (yes/no/unsure or don’t know).
Other variables captured included: program cost (open text field); whether the program offered 12-step programming, including an Evangelical Baptist adaptation of Alcoholics Anonymous’ original 12-step programming called Celebrate Recovery (available/not available/don’t know); whether 12-step programming, if available, was mandatory (yes/no/unsure or don’t know); what other therapeutic modalities the program offered (open text field); whether the program enforced a dress code (yes/no/unsure or don’t know); allowed smoking (yes/no/unsure or don’t know); provided withdrawal management services (yes, onsite/yes, through a partner organization/no or don’t know); and whether provider staff made negative, discouraging, or stigmatizing remarks about OAT (yes/no). Research staff also provided free text descriptions of and short verbatim quotes about these program elements as offered by the provider staff on the call. Finally, callers asked providers whether they imposed work obligations (yes/no/unsure or don’t know) and prompted them to describe those obligations. Callers noted work obligations, consisting of labor typically undertaken in paid, professional settings (e.g. retail service, lawn care, moving services, call center services, auto work), in open text fields and subsequently coded them as “job-like work obligations” (yes/no, unsure, or not applicable). When program staff provided the information, callers noted whether participants fulfilled job-like work requirements in commercial enterprises owned and operated by the service provider (e.g. in provider-run thrift stores, moving services, etc.).
Finally, using publicly available information posted to the North Carolina Department of Health and Human Services website, we determined what licenses for mental health services the North Carolina Division of Health Services Regulation had issued (as of January 2023) to providers in our sample.
2.6. Analysis of program characteristics
The study generated descriptive statistics for all variables. We used Fisher’s exact tests to identify associations between binary-coded (yes/combined no or unsure) provider characteristics and our two primary outcomes of interest: availability of OAT (prescribed by the provider or enabled through a community-based prescriber) and negative, discouraging, or stigmatizing remarks that provider staff made about OAT. For this analysis, licensure categories were collapsed into a binary variable indicating whether a provider was operating under a state license covering residential SUD treatment, other SUD-related residential services (including supervised living for adults with SUD and therapeutic communities1), or outpatient treatment for adults with SUD—or whether the provider was operating under none of these licenses.2 Cost was a binary variable for this analysis, reflecting a monthly cost to participants of $0–500 or a monthly cost of more than $500. We utilized the Benjamini-Hochberg procedure, with α = 0.05, to control the false discovery rate for these independent tests (Benjamini & Hochberg, 1995). These statistical functions were performed with STATA 18 (StataCorp, LLC. College Station, Texas, USA).
We used unsupervised agglomerative clustering to identify facilities with similar characteristics (Nielsen, 2016). These models identify natural categories existing within the data based on facility characteristics and can do so without imposing investigator bias. Facilities were clustered according to similarity based on Jaccard distance across the following binary (yes/no or unsure) program characteristics; allows OAT; allows naltrexone only; is faith-based; mandates 12-step programming; imposes a dress code; allows smoking; and imposes job-like work requirements. The study team selected these variables based on the aforementioned descriptive analyses, correlation with OAT availability, and consensus among subject matter experts on the research team (JC, BO, and RS) about the relative salience of program characteristics in shaping client interest and experience. The analysis assessed clusters using three measures of goodness of fit—Silhouette index (Rousseeuw, 1987); Caliński-Harabasz index (Caliński & Harabasz, 1974); and Davies-Bouldin index (Davies & Bouldin, 1979). Patterns of two to six clusters were considered and compared using the three indices of goodness of fit and an elbow graph of within-cluster sum of squares as a function of the number of clusters. This clustering analysis was completed in Pyt hon 3.9 using the unsupervised sklearn.cluster package “AgglomerativeClustering” within a Deepnote distributed environment (Deepnote, Prague, CZ).
We subsequently used a random forest algorithm to empirically recreate those clusters and assess which variables were most predictive of cluster classifications (i.e., feature importance). The unsupervised machine learning algorithm ran 500 iterations with a maximal depth (i. e., maximum number of splits in the decision tree) of three to avoid overfitting. Finally, the study visualized program characteristics of validated clusters in a heat map showing the proportion of service providers representing each of the eight program characteristics. Random forest classification and heat map visualization were completed in Python 3.9 using the sklearn.ensemble package “Random-ForestClassifier” within a Deepnote distributed environment (Deepnote, Prague, CZ).
We further assessed cluster construct validity using two-dimensional contour plots of t-distributed Stochastic Neighbor Embedding (t-SNE) plots of the entire vector space. To provide an additional measure of internal validity on the unsupervised machine learning algorithms, we compared the facility licensure status of the empirically derived clusters. All code used and output generated for agglomerative clustering, random forest classification, and cluster visualization in the Deepnote environment are in Appendix B.
2.7. Adherence to Open Science principles
This study was not pre-registered. All code used in agglomerative clustering, random forest classification, and cluster visualization are provided in the form of a Jupyter Notebook.3 Data that support the findings of this study are openly available in the Carolina Digital Repository (DOI: doi:10.17615/j42c-pf21).
3. Results
Of the 94 treatment providers included in our original sample, 66 were included in this analysis. Callers were unable to reach 10 providers after three separate contact attempts; determined 10 providers to be ineligible for inclusion in the study based on information obtained in the call; and confirmed two to be permanently closed via recorded message or live conversation. Callers successfully contacted the remaining 72 treatment providers; however, staff from six of those 72 providers refused to answer questions about their treatment program over the phone. As no program data was obtained from these treatment providers, they were not included in this analysis. This left 66 treatment providers in our sample.
3.1. Program characteristics
Table 1 presents all descriptive statistics. Eight (12.1 %) providers were licensed as “Residential Treatment – Individuals with Substance Abuse Disorder.” An additional 15 were licensed to provide other residential substance use services. Eight programs (12.1 %) were only licensed for outpatient SUD treatment. More than half (n = 35, 53 %) of all providers were not operating under any of these licenses.
Table 1.
Characteristics of substance use service providers in North Carolina (N = 66).
| Provider characteristics | n (%) |
|---|---|
| State licensing | |
| Has a license for residential treatment | 8 (12.1) |
| Has license for SAIOP, SACOT, or day tx, but not residential | 8 (12.1) |
| Has a license for supervised living (or tc, not tx) | 15 (22.7) |
| None of these | 35 (53.0) |
| Medications for opioid use disorder | |
| At least one form of OAT is prescribed | 7 (10.6) |
| At least one form of OAT is allowed from another prescriber | 9 (13.9) |
| OAT is prohibited | 49 (74.2) |
| Naltrexone is prescribed and at least one kind of OAT is allowed | 11 (16.7) |
| Naltrexone is prescribed and OAT is prohibited | 6 (9.2) |
| Anti-OAT statements | |
| Provider made negative or stigmatizing comments about OAT | 33 (50.0) |
| Detox services | |
| Detox onsite | 10 (15.2) |
| Partners with a detox facility | 3 (4.6) |
| 12 step programming (including Celebrate Recovery) | |
| Offered | 36 (54.6) |
| Offered and mandatory | 25 (37.9) |
| Religion | |
| Provider is a faith-based organization | 33 (50.0) |
| The program has an explicit focus on biblical study | 11 (16.7) |
| Work requirements | |
| Job-like work requirements | 28 (42.4) |
| Work obligations in a provider-owned and -operated enterprise | 20 (30.3) |
| Program rules | |
| Dress code | 38 (57.6) |
| Allows smoking | 26 (43.3) |
| Program cost | |
| Free for some or all persons | 17 (25.8) |
| $500 or less per month | 15 (22.7) |
| $501–$10,000 per month | 17 (25.8) |
| More than $20,000 per month | 8 (12.2) |
| Unable or unwilling to say over the phone | 9 (13.6) |
Ten (15.2 %) providers offered detoxification services on-site, and an additional three (4.6 %) partnered with a separate detoxification facility. More than half of the providers (n = 36, 54.6 %) offer some form of 12-step programming, and more than one third of providers (n = 25, 37.9 %) mandate 12-step participation. Half of all providers (n = 33, 50.0 %) described their services as faith-based on their website; of these, 11 (16.7 %) described their services to callers as consisting primarily of biblical study. Two (6.1 %) of the 33 faith-based service providers were licensed: one for residential SUD treatment, the other for outpatient SUD treatment. All programs endorsed some other kind of service as a core element of their program, such as: individual and group counseling, seminars, parenting classes, addiction classes, exercise classes, and life skills classes.
Nearly half (n = 28, 42.4 %) of all providers impose some kind of work requirement. Of these, 20 (71.4 % of the 28 programs that impose work obligations and 30.3 % of the total sample) required clients to work in provider-owned and -operated enterprises (e.g. a retail store or moving service). Seventeen (25.8 %) providers reported that their services were free of charge to some or all clients. Of the remaining providers, 15 (22.7 %) reported a cost of $500 or less per month, 17 (25.8 %) reported a cost of $501–10,000 per month, 8 (12.2 %) reported that their services cost more than $20,000 per month for persons without insurance. The remaining nine providers were unwilling or unable to discuss the cost of services over the phone.
3.2. Prevalence and correlates of OAT availability and discouraging remarks about OAT
Seven (10.6 %) treatment providers informed callers that they provided OAT (typically buprenorphine) on-site; an additional nine (13.6 %) described OAT as available through an outside or community-based prescriber. Staff at one service provider could not say whether OAT was available. The remaining 49 (74.2 %) providers told callers that residents were barred from receiving OAT. Eleven (16.7 %) providers reported offering naltrexone, of which five (7.5 %) also provide at least one form of OAT. The remaining six (9.2 %) offered naltrexone only.
Representatives from 33 (50.0 %) providers made negative, discouraging, or stigmatizing remarks about OAT. Table 2 presents direct quotes from program staff who made discouraging remarks to callers. Sentiments expressed included that OAT simply replaces one addiction for another, that the OAT is more addictive than illicit drugs, that someone taking OAT is not truly “abstinent” or “clean,” that OAT does not constitute “true recovery,” and that OAT medications “will kill you.”
Table 2.
Negative or stigmatizing statements made about OAT by provider staff.
| Statements made by provider staff |
|---|
| “[OAT is] trading one addiction for another.” |
| “[OAT] still feeds the addict brain.” |
| “[OAT is not allowed] because the goal is long-term abstinence.” |
| “[OAT is] the lesser of two evils.” |
| “Personally, think it’s better to try without [OAT] first.” |
| “[OAT] tends to be just as addictive or more addictive than heroin.” |
| “[OAT is a] chemical dependency.” |
| “[OAT is not allowed because] we don’t allow anything that can be abused.” |
| “We don’t do [OAT, because] you have to be clean before you get here.” |
| “[OAT is] addictive.” |
| “Some people in treatment come because of Suboxone [buprenorphine] addiction.” |
| “We discourage [OAT].” |
| “[OAT] can sometimes be detrimental in the recovery process.” |
| “[On OAT], your body just transfers the addiction.” |
| “[OAT is] harder to get off of.” |
| “My personal opinion is that suboxone [buprenorphine] is only for detox, and even that’s not true detox.” |
| “When I had my own issues, I was prescribed Suboxone [buprenorphine], but only used it when I couldn’t get other opioids.” |
| “I’ve never heard of any successful recovery stories on [OAT].” |
| “Suboxone [buprenorphine] enslaved me, entrapped me in a prison. It prohibited me from getting to the root of my substance use.” |
| “[OAT is] just another addiction” |
| “All [that OAT] is, is a legal morphine.” |
| “[OAT is] still bad for you.” |
| “[OAT] will kill you.” |
| “These guys have tried it all, including Suboxone [buprenorphine], and nothing has worked. You know why? Because they haven’t tried Christ.” |
| “[OAT is not allowed because] we don’t allow anything addictive here.” |
| “[OAT] is legal heroin.” |
| “If you really want to be clean you have to start at ground zero.” |
Correlates of OAT availability and discouraging remarks about OAT with Benjamini-Hochberg critical values for determining statistical significance are listed in Table 3. Providers were significantly more likely to allow participants to receive OAT if they were operating under any state license for substance use services (41.9 % vs 8.6 %, p = 0.002) or if they offered 12-step programming (36.1 % vs. 10.0 %, p = 0.020). Providers were significantly less likely to allow access to OAT if they were faith-based (6.1 % vs 42.4 %, p = 0.001), imposed a dress code (5.3 % vs 50.0 %, p < 0.001), or required residents to work in a provider-owned and -operated commercial enterprise (5.0 % vs 32.6 %, p = 0.026).
Table 3.
Association of OAT availability and anti-OAT statements with program characteristics.
| Program characteristics | OAT allowed (from any prescriber) | Fisher’s 2-sided p-value | Benjamini-Hochberg critical value |
|---|---|---|---|
| n (%) | (i/m)Qa | ||
| Dress code imposed | |||
| Yes | 2 (5.3) | <0.001b | 0.006 |
| No or unsure | 14 (50.0) | ||
| Faith-based | |||
| Yes | 2 (6.1) | 0.001b | 0.013 |
| No or unsure | 14 (42.4) | ||
| State licensure | |||
| Any residential or treatment licensec | 13 (41.9) | 0.003b | 0.019 |
| None of these licenses | 3 (8.6) | ||
| Offers 12-step programming | |||
| Yes | 13 (36.1) | 0.020b | 0.025 |
| No or unsure | 3 (10.0) | ||
| Work in provider-owned enterprise | |||
| Yes | 1 (5.0) | 0.026b | 0.031 |
| No or unsure | 15 (32.6) | ||
| Smoking allowed | |||
| Yes | 10 (38.5) | 0.041 | 0.038 |
| No or unsure | 6 (15.0) | ||
| Job-like work requirements | |||
| Yes | 4 (14.3) | 0.092 | 0.044 |
| No or unsure | 12 (31.6) | ||
| Cost of services | |||
| $500 or less per month | 6 (18.8) | 1.000 | 0.050 |
| more than $500 per month | 4 (16.0) | ||
| Program characteristics | Anti-OAT statements | Fisher’s 2-sided p-value | Benjamini-Hochberg critical value |
| n (%) | (i/m)Qa | ||
| Work in provider-owned enterprise | |||
| Yes | 15 (75.0) | 0.015 | 0.006 |
| No or unsure | 18 (39.1) | ||
| Dress code imposed | |||
| Yes | 24 (63.2) | 0.024 | 0.013 |
| No or unsure | 9 (32.1) | ||
| Job-like work requirements | |||
| Yes | 18 (64.3) | 0.136 | 0.019 |
| No or unsure | 15 (39.5) | ||
| Faith-based | |||
| Yes | 19 (57.6) | 0.325 | 0.025 |
| No or unsure | 14 (42.4) | ||
| Cost of services | |||
| $500 or less per month | 16 (50.0) | 0.593 | 0.031 |
| more than $500 per month | 15 (60.0) | ||
| State licensure | |||
| Any residential or treatment licensec | 14 (45.2) | 0.622 | 0.038 |
| None of these licenses | 19 (54.3) | ||
| Offers 12-step programming | |||
| Yes | 19 (52.8) | 0.805 | 0.044 |
| No or unsure | 14 (46.7) | ||
| Smoking allowed | |||
| Yes | 13 (50.0) | 1.000 | 0.050 |
| No or unsure | 20 (50.0) |
The Benjamini-Hochberg Correction is a method for adjusting calculated p-values to reduce the likelihood of a false positive when many comparisons are calculated. Tests are considered statistically significant if the calculated p-value is less than the Benjamini-Hochberg critical value, for which i = is the p-value rank of each test, m = the number of tests, and Q = the chosen maximum allowable false discovery rate (here, Q = 0.05).
Statistically significant [p < (i/m)Q].
“Residential Treatment – Individuals with Substance Abuse Disorder;” “Residential Recovery Programs for Individuals with Substance Abuse Disorders and their Children;” “Supervised Living for Adults with Substance Abuse Dependency;” “Therapeutic Community;” “Substance Abuse Intensive Outpatient Treatment (SAIOP);” “Substance Abuse Comprehensive Outpatient Treatment (SACOT);” or “Day Treatment Facilities for Individuals with Substance Abuse Disorders.”
No program characteristics were found to be significantly associated with negative, discouraging, or stigmatizing remarks about OAT. Descriptively, the largest difference in rates of negative, discouraging, or stigmatizing remarks about OAT was seen between those organizations that require residents to work in provider-owned and -operated commercial enterprises (75 %, n = 15) and those that do not (39.1 %, n = 18).
3.3. Cluster analysis
Visual inspection of the elbow curve showed the optimal number of clusters to be either three or four. We opted for the four-cluster solution over the three-cluster solution, as this solution had a better Silhouette index (0.44 vs. 0.39) and a much better Caliński-Harabasz index (57.35 vs. 31.77), which we determined to outweigh its slightly poorer Davies-Bouldin index (1.12 vs. 1.01). For the sake of completeness, an alternative three-cluster solution is presented in Appendix B. Random forest classification identified rules about tobacco smoking and faith-based programming as the two most predictive features of program clustering, followed by OAT availability and mandatory 12-step programming. Dress code enforcement, naltrexone availability only (no OAT), and job-like work obligations were the least predictive features. We examined the commonality of program characteristics in each of the four clusters that had been defined empirically. Descriptive statistics of program characteristics are visualized in Fig. 1.
Fig. 1.
Percent prevalence of provider characteristics by cluster.
The largest cluster of providers (“unlicensed abstinence-only work programs”, n = 21) consists entirely of faith-based organizations (100 %) that prohibit OAT (0 % availability), enforce strict dress codes (100 %), and (with one program being the exception) impose job-like work requirements (95 %). Most also prohibit smoking (76 %). Cross-referencing against state licensure data revealed all providers in this cluster to be unlicensed; this provides a strong level of internal validity since licensure status was not included in agglomerative clustering or random forest classification.
“Mixed OAT/12-step programs” (n = 17) and “programs without work requirements” (n = 18) are the only clusters where OAT is available, but not all programs in these clusters offer OAT. The “mixed OAT/12-step programs” cluster has the highest proportion of OAT availability (65 %). Most providers in this cluster also allow smoking (82 %), mandate 12-step programming (59 %), and are not faith-based (only 18 % faith-based). By contrast, the “programs without work requirements” cluster had a lower proportion of OAT availability (28 %). Half of providers in this cluster are faith-based (50 %), and half impose a dress code (50 %); none allow smoking (0 %). Importantly, this cluster contains only one provider (6 %) who imposes job-like work requirements upon residents.
In the smallest cluster (“licensed abstinence-only 12-step programs”, n = 10), all providers prohibit OAT (0 % availability) but are not faith-based (0 % faith-based) and do mandate 12-step programming (100 %). This cluster also contains the highest proportion of providers who offer naltrexone but prohibit the use of OAT (40 %). Again, providing support for the internal validity of cluster structure, cross-referencing against state data revealed all but one provider in this cluster to be operating under a state license for some form of residential or outpatient substance use services.
4. Discussion
This audit study found that evidence-based treatments for OUD are largely unavailable at residential substance use services in North Carolina. Fewer than one-quarter of these providers in the state allow access to OAT, and half actively discourage treatment-seekers from utilizing these medications. These findings mirror previous research, which found that less than one-third of residential service providers in a nationwide sample allow residents to access OAT (Beetham et al., 2020). Though narrower in geographic scope, the study presented here builds upon that conducted by Beetham et al. (2020) by expanding the identification of residential substance use services providers beyond the SAMHSA treatment locator to include state and local listings, internet search results, available state licensure information, and knowledge obtained directly from persons who work in peer or recovery support services in North Carolina—a sample likely more representative of the providers that members of the public would identify and approach for support.
A major finding to emerge from our more expansive search strategy is the sheer number of faith-based service providers who bar access to evidence-based care while operating beyond the purview of regulatory control. The most common type of residential substance use service provider—representing nearly one third of all such providers in the state (n = 21)—are unlicensed, faith-based organizations that prohibit OAT and impose job-like work requirements upon residents. It may be tempting, therefore, to suggest that faith-orientation in residential substance use services is, itself, a barrier to evidence-based treatment. However, OAT was widely unavailable among non-faith-based providers in our sample as well, and two (6 %) of the 16 providers in our sample allowing access to OAT are faith-based—one of which also operates under a state-issued license for residential substance use services. Therefore, our data more likely suggest that a major barrier to OAT in North Carolina is not faith-orientation, per se, but a specific kind of faith-based programming that is resistant to evidence-based treatments, prioritizes unpaid labor as a supposedly therapeutic activity, and operates (often legally, per state exemption rules) without any licensure or supervision from state health authorities. Of considerable policy importance, this type of program outnumbers all other varieties of residential services in the state.
In our sample, state licensing was associated with higher rates of OAT access; still, fewer than half of state-licensed providers included in this study allow OAT, and licensed providers were no less likely than unlicensed providers to make discouraging comments about OAT. Current regulations for licensable substance use services (10A N.C. Admin. Code 27G) do not require licensed service providers to provide access to any evidence-based medications. Updating state regulations to tie obligatory OAT access to the receipt of state funding while simultaneously closing exemption loopholes that allow unlicensed providers to proliferate without scrutiny may increase access to evidence-based treatments across the state. In other states, the prohibition of OAT in residential settings has been remedied through lawsuits filed under the Americans with Disabilities Act, by persons subject to discrimination or by the U.S. Department of Justice (Legal Action Center, 2023).
The evidence that OAT improves outcomes and increases the survivability of OUD is overwhelming (Heimer et al., 2024; Larochelle et al., 2018; Mattick et al., 2009; Sordo et al., 2017; Wakeman et al., 2020). In fact, one recent study suggests that, with respect to overdose risk, treatment without OAT is worse than no treatment at all (Heimer et al., 2024). There is also no evidence that residential treatment produces better outcomes for persons with OUD (Wakeman et al., 2020). Therefore, residential services settings without access to OAT are unlikely to result in improved outcomes for persons with OUD; however, service providers usually receive direct payment for services, insurance reimbursements for services, and/or income from residents’ unpaid commercial labor—all significant sources of income for the service provider—so long as residents are retained in their programs. Thus, the financial well-being of the service provider and the behavioral health of residents they serve may be at odds with one another in very consequential ways.
These problems are not specific to North Carolina. Access to medications for OUD is poor across the United States, and the majority of U.S. residents living with OUD do not receive evidence-based care (Krawczyk et al., 2022). North Carolina is set to receive approximately $1.5 billion in opioid settlement funds in the coming years (North Carolina Department of Justice, 2023), and the expansion of evidence-based medication treatment for OUD, including OAT, is a core opioid remediation strategy allowable under the settlement terms (Stein & North Carolina Association of County Commissioners, 2021). Nevertheless, state and county leaders have allocated tens of millions of public dollars to residential substance use service organizations that prohibit OAT, including providers that are not licensed to offer treatment services (Pattani & Knopf, 2022), are not licensed to offer any services (Goodloe-Murphy, 2023), or whose services are merely hypothetical—still in the design stages with no intention to include addiction medicine experts in their planning or implementation at the time of funding allocation (Knopf, 2021).
This widespread failure to support evidence-based treatment illustrates the fact that stigma against OAT remains pervasive both within (Healing Transitions, 2022; Yow & Washo, 2022) and beyond North Carolina (Goodnough, 2018; Lopez, 2017; Roelofs, 2019). Further, public misperceptions of what constitutes effective, evidence-based treatment for OUD are widespread (Walters et al., 2023). Public awareness campaigns designed to increase understanding that OAT is an evidence-based treatment for OUD may also be merited. In North Carolina recently passed new “truth in marketing” legislation (NC § 90–113.151) requiring substance use service providers to include complete, plain language descriptions of their services, including what ASAM level of care those services constitute, in all marketing materials. Future research should evaluate the impact of this legislation.
These findings are subject to specific limitations. The findings of this audit study may not be representative of residential service providers beyond North Carolina. These findings may also be subject to sample bias resulting from our inability to reach or gain information from all identified providers. It is also possible that the statements made by provider staff do not accurately reflect official policies and practices employed in the program they were representing; however, given the frequency with which staff willingly told callers that they lacked information to answer some or all of our questions, the risk of outright mischaracterization is likely low. There may also be variation between staff that would have required serial resampling to capture. Finally, there may be other categorizations that are relevant and valid ways in which to group facilities that we did not quantify, such as rurality, catchment population, and which insurance plans are accepted, among others. These factors could be explored in subsequent research, especially those factors which drive patients’ choices.
5. Conclusion
Fewer than 25 % of residential substance use services providers in North Carolina—and fewer than half of providers licensed by the state—allow residents to access OAT. Half of all providers made negative or discouraging comments about OAT, regardless of licensure, faith-orientation, or cost. State regulators may find opportunities to improve OAT availability through changes to licensure or funding requirements or through consumer protection regulations that limit the ability of service providers who do not offer evidence-based treatments to give members of the public the false impression that they do.
Supplementary Material
Funding
This work was supported by the Open Society Foundations (grant #87079; Carroll) and the National Institutes of Health, National institute on Drug Abuse (K01DA057414; El-Sabawi).
Footnotes
CRediT authorship contribution statement
Jennifer J. Carroll: Writing – original draft, Supervision, Project administration, Methodology, Funding acquisition, Formal analysis, Conceptualization. Nabarun Dasgupta: Writing – review & editing, Software, Methodology, Formal analysis. Bayla Ostrach: Writing – review & editing, Methodology. Taleed El-Sabawi: Writing – review & editing, Methodology. Sarah Dixon: Writing – review & editing, Investigation. Brandon Morrissey: Writing – review & editing, Investigation. Roxanne Saucier: Writing – review & editing, Methodology, Conceptualization.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.josat.2024.209474.
Therapeutic communities are structured residential programs that typically follow an abstinence-only approach to recovery and often operate in the United States according to recovery principles originally developed by the Church of Synanon (Yablonsky, 2002).
In most cases, facilities operating without a state license fall into two categories: (1) those whom the state has not determined to be offering a licensable service (i.e. the program does not conform to the definition of any of the 5 statutorily defined treatment and recovery services considered in this study) or (2) are exempt from licensure under NCAC § 122C-22 as “a charitable, nonprofit, faith-based, adult residential treatment facility that does not receive any federal or State funding and is a religious organization exempt from federal income tax under section 501(a) of the Internal Revenue Code.”
That notebook is visualized in Appendix B and available in .ipynb format through the Carolina Digital Repository.
References
- Ahmad F, Cisewski J, Rossen L, & Sutton P (2024). Provisional drug overdose death counts. National Center for Health Statistics. https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm. [Google Scholar]
- Ajazi EM, Dasgupta N, Marshall SW, Monaco J, Howard AG, Preisser JS, & Schwartz TA (2021). Revisiting the X: BOT naltrexone clinical trial using a comprehensive survival analysis. Journal of Addiction Medicine.. 10.1097/ADM.0000000000000931 [DOI] [PubMed] [Google Scholar]
- Ali F, Law J, Russell C, Bozinoff N, & Rush B (2023). An environmental scan of residential treatment service provision in Ontario. Substance Abuse Treatment, Prevention, and Policy, 18(1), 73. 10.1186/s13011-023-00586-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- American Society of Addiction Medicine. (2023). The ASAM criteria: Treatment criteria for adictive, substance-related, and co-occurring conditions, 4th edition: vol. volume 1-adults (4th ed.). Hazelden Publishing. [Google Scholar]
- Beetham T, Saloner B, Gaye M, Wakeman SE, Frank RG, & Barnett ML (2020). Therapies offered at residential addiction treatment programs in the United States. JAMA, 324(8), 804–806. 10.1001/jama.2020.8969 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Beetham T, Saloner B, Gaye M, Wakeman SE, Frank RG, & Barnett ML (2021). Admission practices and cost of care for opioid use disorder at residential addiction treatment programs in the US. Health Affairs, 40(2), 317–325. 10.1377/hlthaff.2020.00378 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Benjamini Y, & Hochberg Y (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological), 57(1), 289–300. 10.1111/j.2517-6161.1995.tb02031.x [DOI] [Google Scholar]
- Caliński T, & Harabasz J (1974). A dendrite method for cluster analysis. Communications in Statistics, 3(1), 1–27. 10.1080/03610927408827101 [DOI] [Google Scholar]
- Davies DL, & Bouldin DW (1979). A cluster separation measure. IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-1(2), 224–227. 10.1109/TPAMI.1979.4766909 [DOI] [PubMed] [Google Scholar]
- Donnelly-DeRoven C (2022, June 6). Gaps in NC addiction treatment disproportionately hit rural residents of color. North Carolina Health News. http://www.northcarolinahealthnews.org/2022/06/06/gaps-in-nc-addiction-treatment-disproportionately-hit-rural-residents-of-color/. [Google Scholar]
- Goodloe-Murphy MH (2023, June 26). Opioid settlement funds budgeted in Dare. The Coastland Times. https://www.thecoastlandtimes.com/2023/06/25/opioid-settlement-funds-budgeted-in-dare/. [Google Scholar]
- Goodnough A (2018, December 29). In rehab, ‘two warring factions’: Abstinence vs. medication. The New York Times. https://www.nytimes.com/2018/12/29/health/opioid-rehab-abstinence-medication.html. [Google Scholar]
- Hay KR, Huhn AS, Tompkins DA, & Dunn KE (2019). Recovery goals and long-term treatment preference in persons who engage in nonmedical opioid use. Journal of Addiction Medicine, 13(4), 300–305. 10.1097/ADM.0000000000000498 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Healing Transitions. (2022). Healing transitions and agonist and partial agonist based medication assisted treatment for opioid use disorder. https://www.northcarolinahealthnews.org/wp-content/uploads/2022/04/Healing-Transitions-and-Medication-Assisted-Treatment.pdf.
- Heimer R, Black AC, Lin H, Grau LE, Fiellin DA, Howell BA, … Becker WC (2024). Receipt of opioid use disorder treatments prior to fatal overdoses and comparison to no treatment in Connecticut, 2016–17. Drug and Alcohol Dependence, 254, Article 111040. 10.1016/j.drugalcdep.2023.111040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Knopf T (2021, December 7). Targeting the opioid crisis, NC lawmakers give $10 million to a new church ministry. WFAE 90.7 - Charlotte’s NPR News Source. https://www.wfae.org/health/2021-12-07/targeting-the-opioid-crisis-nc-lawmakers-give-10-million-to-a-new-church-ministry.
- Knopf T (2022, April 7). Tension builds around best ways to spend NC’s opioid settlement money. North Carolina Health News. http://www.northcarolinahealthnews.org/2022/04/07/tension-builds-around-ncs-opioid-settlement-money/. [Google Scholar]
- Krawczyk N, Rivera BD, Jent V, Keyes KM, Jones CM, & Cerdá M (2022). Has the treatment gap for opioid use disorder narrowed in the U.S.?: A yearly assessment from 2010 to 2019. International Journal of Drug Policy, 110, Article 103786. 10.1016/j.drugpo.2022.103786 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Larochelle MR, Bernson D, Land T, Stopka TJ, Wang N, Xuan Z, … Walley AY (2018). Medication for opioid use disorder after nonfatal opioid overdose and association with mortality: A cohort study. Annals of Internal Medicine, 169(3), 137–145. 10.7326/M17-3107 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Legal Action Center. (2023). Cases involving discrimination based on treatment with medication for opioid use disorder (MOUD). https://www.lac.org/assets/files/Cases-involving-denial-of-access-to-MOUD.pdf.
- Lopez G (2017, July 20). There’s a highly successful treatment for opioid addiction. But stigma is holding it back. Vox. https://www.vox.com/science-and-health/2017/7/20/15937896/medication-assisted-treatment-methadone-buprenorphine-naltrexone.
- Mattick RP, Breen C, Kimber J, & Davoli M (2009). Methadone maintenance therapy versus no opioid replacement therapy for opioid dependence. In Cochrane database of systematic reviews. John Wiley & Sons, Ltd.. http://onlinelibrary.wiley.com/doi/10.1002/14651858.CD002209.pub2/abstract [Google Scholar]
- Nielsen F (2016). Hierarchical clustering. In Introduction to HPC with MPI for data science (pp. 195–211). Springer International Publishing. 10.1007/978-3-319-21903-5_8. [DOI] [Google Scholar]
- North Carolina Department of Health and Human Services. (2023a). North Carolina reports 22% increase in overdose deaths | NCDHHS. NCDHHS. https://www.ncdhhs.gov/news/press-releases/2023/02/21/north-carolina-reports-22-increase-overdose-deaths. [Google Scholar]
- North Carolina Department of Health and Human Services. (2023b). Opioid action plan data dashboard. NCDHHS. https://www.ncdhhs.gov/about/department-initiatives/opioid-epidemic/opioid-action-plan-data-dashboard. [Google Scholar]
- North Carolina Department of Justice. (2023, June 9). Teva, Allergan, CVS, and Walgreens finalize opioid settlement agreements. Attorney General Josh Stein. https://ncdoj.gov/teva-allergan-cvs-and-walgreens-finalize-opioid-settlement-agreements/. [Google Scholar]
- Pattani A, & Knopf T (2022, April 6). $11M for North Carolina work-based rehab raises concerns. KFF Health News. https://kffhealthnews.org/news/article/trosa-north-carolina-drug-opioid-work-rehab-funding-concerns/. [Google Scholar]
- Rankin KA, Mosier-Mills A, Hsiang W, & Wiznia DH (2022). Secret shopper studies: An unorthodox design that measures inequities in healthcare access. Archives of Public Health, 80(1), 226. 10.1186/s13690-022-00979-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- Roelofs T (2019, January 17). In opioid war, studies urge medication. Michigan clinics push abstinence. Bridge Michigan. https://www.bridgemi.com/children-families/opioid-war-studies-urge-medication-michigan-clinics-push-abstinence. [Google Scholar]
- Rousseeuw PJ (1987). Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics, 20, 53–65. 10.1016/0377-0427(87)90125-7 [DOI] [Google Scholar]
- Sordo L, Barrio G, Bravo MJ, Indave BI, Degenhardt L, Wiessing L, … Pastor-Barriuso R (2017). Mortality risk during and after opioid substitution treatment: Systematic review and meta-analysis of cohort studies. BMJ (Clinical Research Ed.), 357, Article j1550. 10.1136/bmj.j1550 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Stahler GJ, Mennis J, & Baron DA (2021). Racial/ethnic disparities in the use of medications for opioid use disorder (MOUD) and their effects on residential drug treatment outcomes in the US. Drug and Alcohol Dependence, 226, Article 108849. 10.1016/j.drugalcdep.2021.108849 [DOI] [PubMed] [Google Scholar]
- Stahler GJ, Mennis J, Stein L.a. R., Belenko S, Rohsenow DJ, Grunwald HE, … Martin RA (2022). Treatment outcomes associated with medications for opioid use disorder (MOUD) among criminal justice-referred admissions to residential treatment in the U.S., 2015–2018. Drug and Alcohol Dependence, 236, Article 109498. 10.1016/j.drugalcdep.2022.109498 [DOI] [PubMed] [Google Scholar]
- Stein J, & North Carolina Association of County Commissioners. (2021). Memorandum of agreement between the state of North Carolina and local governments on proceeds relating to the settlement of opioid litigation. https://www.morepowerfulnc.org/wp-content/uploads/2023/07/Final-Opioid-MOA-rev-July-2023-re-Ex-E-and-F.pdf.
- U.S. Centers for Disease Control and Prevention. (2023). State unintentional drug overdose reporting system (SUDORS). Final data. U.S. Department of Health and Human Services, CDC. https://www.cdc.gov/drugoverdose/fatal/dashboard/index.html. [Google Scholar]
- Wakeman SE, Larochelle MR, Ameli O, Chaisson CE, McPheeters JT, Crown WH, … Sanghavi DM (2020). Comparative effectiveness of different treatment pathways for opioid use disorder. JAMA Network Open, 3(2), e1920622. 10.1001/jamanetworkopen.2019.20622 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Walters SM, Liu W, Lamuda P, Huh J, Brewer R, Johnson O, … Schneider JA (2023). A national portrait of public attitudes toward opioid use in the US: A latent class analysis. International Journal of Environmental Research and Public Health, 20 (5), 4455. 10.3390/ijerph20054455 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yablonsky L (2002). Whatever happened to Synanon? The birth of the anticriminal therapeutic community methodology. Criminal Justice Policy Review, 13(4), 329–336. 10.1177/088740302237802 [DOI] [Google Scholar]
- Yow M, & Washo M (2022, April 6). Letter from fellowship hall drug + alcohol rehab center to the wake county drug overdose prevention coalition. https://www.northcarolinahealthnews.org/wp-content/uploads/2022/04/Fellowship-Hall-statement.pdf.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.