Stigma and quality of co‐located care for HIV‐positive people in addiction treatment in Ukraine: a cross‐sectional study

Abstract Introduction Co‐located treatment for HIV and opioid use disorder has been shown to improve care outcomes for HIV‐positive people who inject drugs (PWID) in Ukraine. However, patients continue to be stigmatized for both HIV and substance use. This study aimed to assess whether co‐located care for HIV‐positive PWID receiving opioid agonist treatment (OAT) services in Ukraine is associated with less stigma and better perceived quality of HIV services. Methods This cross‐sectional study enrolled 191 HIV‐positive PWID who received OAT services at three healthcare facilities providing substance use treatment (OAT only) and at four facilities that provided co‐located care (both OAT and HIV treatment) in six regions in Ukraine during July‐September, 2017. Primary outcomes were HIV stigma (Berger scale), substance use stigma (Substance Abuse Stigma Scale) and intersectional stigma (both stigma forms above 75th percentile). Secondary outcome was quality of HIV care, a composite score based on a package of received services. Linear and ordinal regressions were used to assess the predictors of selected outcomes. Results Study participants were 75% male, mean age 40 ± 7 years; 47% received co‐located care, and 10.5% had both high HIV and substance use stigma. Co‐located care was neither associated with HIV nor substance use stigma but it was linked to better quality of HIV care (adjusted odds ratio: 4.13; 95% CI: 2.31, 7.54). HIV stigma was associated with suicide attempts (adjusted beta (aβ): 5.90; 95% CI: 2.05, 9.75), and substance use stigma was linked to poor mental health (aβ: −0.26; 95% CI: −0.44, −0.08) and lower likelihood of receipt of services from non‐governmental organization (NGO; aβ: −6.40; 95% CI: −10.23, −2.57). Conclusion One in ten people with HIV in this cohort who received OAT services experienced high levels of both HIV and substance use stigma, which was associated with poorer mental health and less NGO support. Co‐located HIV and OAT services were linked to better perceived quality of HIV care, but did not seem to reduce stigma for this key population. Stigma interventions for PWID, possibly delivered involving NGOs, may be an approach to mitigate this challenge.


| INTRODUCTION
Stigma, the social exclusion of individuals labelled with an undesirable trait, remains one of the key barriers discouraging people from seeking and receiving services for HIV and associated diseases [1]. About one in eight people living with HIV (PLHIV) reported being denied health services because of stigma and discrimination [2]. HIV-related stigma has been linked to numerous poor health outcomes: negative effects on mental health [3][4][5]; unhealthy alcohol use [3,6,7]; poor antiretroviral treatment (ART) adherence [3,8]; and lower overall quality of life [3,9]. In addition to HIV-related stigma, key populations are also affected by stigma relating to people with other identities and behaviours, such as people who use substances [10], people engaged in sex work [11], sexual minorities [12] and people under 25 years of age [13].
Stigma research among PLHIV rarely utilized an intersectionality framework incorporating multiple levels of stigma in their study [14]. Intersectional stigma can results from the layering of multiple marginalized statuses [15]. The few intersectional stigma studies conducted on HIV and substance use found that HIV-positive people who inject drugs (PWID) endorse internalized stigma (experienced or expected from others by affected people) and public stigma (created by public attitudes) related to substance use, in addition to HIV stigma; both forms were linked to poor mental health and less engagement in treatment [16,17].
PWID are a highly marginalized and stigmatized group, and even more so are those living with HIV [16][17][18][19]. Due to stigma, health providers may perceive patients with substance use disorders as problematic and therefore be less motivated to initiate their HIV treatment, an assumption shown to be unfounded [17,19].
In response to the high prevalence of HIV among PWID in Ukraine (23%) [20,21], some healthcare facilities have started offering co-located HIV and opioid agonist treatment (OAT) for opioid use disorders. Provision of HIV, OAT, and other health services at one location can increase access to services for HIV-positive PWID by improving accessibility, user-friendliness, and comprehensiveness of care [20][21][22][23]. However, negative attitudes about methadone or buprenorphine (OAT medications) can stigmatize patients using OAT services [24][25][26]. Notably, little is known about the effect of care co-location on patients' perceived stigma. In particular, it is unknown whether co-located care reduces patients' perception or expectation of HIV and substance use stigma.
This study aimed to assess whether HIV and substance use treatment co-location for HIV-positive PWID receiving OAT services in Ukraine is associated with lower stigma levels, compared to OAT only sites. We also explored association between treatment co-location and perceived quality of care. We hypothesized that co-located care is associated with lower stigma levels and better quality of care, compared to OAT only sites.

| Study design
We conducted a cross-sectional study at seven healthcare facilities in six regions of Ukraine. We collected data at three sites providing OAT only (two branches of narcological clinic in Kyiv and narcological dispensary in Mykolaiv) and at four sites that provided co-located OAT and HIV treatment services (narcological dispensaries in Dnipro, Lviv and Cherkasy and AIDS Center in Odesa).

| Country settings
Ukraine has the second-largest HIV epidemic in Eastern Europe and Central Asia [20,21]. HIV prevalence is the highest among PWID (23%) followed by males who have sex with males (7.5%) and sex workers (5%) [20,21].
In 2017, the estimated PWID population in Ukraine was 350,300 with over 80,600 of them being HIV-positive [20,21]. Among HIV-positive PWID, 58% are aware of their status, and 38% receive ART [20,21]. This epidemiologic situation is further complicated by the reality that tuberculosis (TB) coinfection is very common in Ukraine [27].
One study site (AIDS Center in Odesa) was originally an HIV clinic founded in 1999 that started to prescribe OAT in 2014. Other study sites were substance use treatment clinics which introduced OAT between 2008 and 2012 (narcological clinics in Cherkasy, Dnipro, Kyiv, Lviv and Mykolaiv). During 2012-2015, HIV treatment sites were created at narcological clinics in Cherkasy, Dnipro and Lviv.
In Ukraine, co-located OAT and HIV care is delivered by two distinct service providers, i.e. at narcological clinics (OAT sites) and at HIV clinics (ART sites), that are located in the same building. Providers at co-located care sites routinely communicate to discuss patient care, while providers at OAT sites without co-located HIV services usually do not communicate with HIV specialists.
HIV testing is available at both narcological and HIV clinics. HIV-positive PWID identified at narcological clinics are referred to HIV clinics for diagnosis, ART prescription and monitoring. PWID identified at HIV clinics are referred to narcological clinics for addiction diagnosis and treatment, including OAT. OAT is supervised by Addiction Treatment Specialists, while HIV treatment is supervised by Infectiouos Disease Specialists.
Only 20% of OAT patients in Ukraine are on self-administered treatment and receive take-home doses for 10 days maximum [29]. Most patients visit OAT sites every day and receive methadone or buprenorphine under direct supervision of a health worker. Patients may be allowed to take OAT at home if they have proven compliance with treatment during six months [30]. OAT treatment monitoring, including urine tests for the presence of opioids, is conducted at least once per three months [30]. Frequency of visits is different for HIV treatment. Typically, HIV patients receive take-home ART doses for 30 days immediately after treatment initiation, and this period may be extended up to 90 days in case of consistent treatment compliance [31]. National HIV treatment guidelines recommend at least one visit per six months for HIV treatment monitoring (i.e. CD4 and viral load testing) [31].
According to the national legislation, OAT and ART services can be provided only by physicians and nurses, and therefore they are not available in non-governmental organizations (NGOs) working with PWID. NGOs provide case-management, social support and counselling, information and education, condom distribution and sterile syringes programmes. Four of the seven selected sites (Cherkasy, Dnipro, Mykolaiv, and Odesa) had a co-located NGO providing services to PWID at the healthcare facility.

| Participants
Eligibility criteria for the study participants were the following: age ≥18 years; lifetime history of drug injection (by self-report); HIV-positive status (by self-report); receiving OAT; fluent in Russian or Ukrainian. The exclusion criterion was cognitive impairment resulting in inability to provide informed consent based on an assessment during the consent process.

| Sample
We recruited a consecutive sample of HIV-positive individuals receiving OAT, who were referred to the study by their healthcare providers. In total, 198 participants were screened for the study. Of them, 191 met inclusion criteria, participated in the study and were included in the data analysis.

| Study procedures and period
The Ukrainian Institute on Public Health Policy collected data from July through September 2017. An on-site research assessor screened patients referred by their healthcare provider at the clinic for eligibility, offered eligible individuals participation in the study, obtained informed consent, and administered the survey in a private location at the clinic. Survey measures were pilot-tested and administered in Russian or Ukrainian, as to the participant's preference. All study staff was trained on the study protocol, human subjects protection and participant assessment prior to recruitment.
We collected all study data via Research Electronic Data Capture tools [32]. Participants received 200 Ukrainian Hryvnia (the equivalent of US $8 at the time of study) in cash as compensation for their time and transportation expenses.

| Variables
Primary outcomes included HIV and substance use stigma scores, and intersectional stigma (i.e. those who reported high scores for both stigma forms). We measured HIV stigma with the abbreviated Berger Scale [33]. Substance use stigma was assessed with an abbreviated Substance Abuse Stigma Scale [34]. Minimum-maximum standardization with a range from 0 to 100 was used to make total HIV and substance use stigma scales comparable. The intersectional stigma variable was categorized as "both HIV and substance use stigma score high"; "high HIV stigma only, " "high substance use stigma only, " "medium or low scores for both HIV and substance use stigma. " Scores ≥75th percentile were considered as high stigma level.
Secondary outcome was quality of HIV care measured as HIV Care Quality Index (HCQI) [22]. The scale includes four items related to coverage of HIV services recommended per national guidelines at the time of study (at least one regular HIV check-up within the last six months; at least one CD4 test within the last six months; receipt of ART; undetectable viral load); two items related to TB services (at least one TB screening within the last 12 months; receipt of isoniazid preventive therapy for TB); two items related to OAT (receipt of World Health Organization recommended dosage of ≥80mg of methadone or ≥12mg of buprenorphine; and no injection drug use during last 30 days); and two additional items on the receipt of hepatitis C virus (HCV) screening and receipt of referrals for other services or support. The HCQI scores ranged from 0 to 100, indicating the percentage met of the ten indicators for care among HIV-positive OAT patients.
Covariates included demographic, behavioural and clinical characteristics of the study participants as well as perceptions on health care services and social support. Demographic variables were age, gender, marital status, household size, education, employment, personal monthly income in the national currency (Ukrainian Hryvnias) dichotomized at median, and history of incarceration. Behavioural variables included illicit drug use during the past 30 days, number of sexual partners in the last year, condomless sex in the last year, history of selling sex, smoking and unhealthy alcohol use per Alcohol Use Disorders Identification Test, short version (AUDIT-C) score [35]. Clinical characteristics were time since first HIVpositive test, CD4 cell count, history of TB and HCV infection, perceived mental and physical health measured by short form health survey version 2.0 (SF-12v2, higher scores indicating better health) [36], past suicide attempts, OAT drug (methadone or buprenorphine) and duration. All clinical variables were based on self-report. Accessibility and user-friendliness scales were constructed based on the approach used in a previous Ukrainian study of client satisfaction with HIV services [37]. Social support scale was measured using modified version of the Duke University-University of North Carolina Functional Support Questionnaire [38].
We pilot tested the questions for the main variables among five participants at one site in Kyiv prior to the study. To assess internal consistency of the main variables, we conducted confirmatory factor analysis and calculated Cronbach's alphas (a) for the following scales: (1)  Applying thresholds of a > 0.7 for satisfactory internal consistency of scales, CFI, TLI > 0.9 and SRMR < 0.1 for factor validity, we found all items except HCQI sufficiently valid and internally consistent. We used the HCQI for comparison with a previous study conducted in Ukraine [22].

| Data analysis
Characteristics of the study participants stratified by the type of site were summarized with descriptive statistics and compared by chi-squared tests for categorical variables, T-tests for continuous variables with approximately normal distribution and Kruskal-Wallis tests for continuous variables with skewed distribution. HIV and substance use stigma scores were normally distributed, and we assessed their predictors using linear regression. We estimated Intersection between HIV and substance use stigma, depicted as number of people with a given substance use stigma score quartile within each HIV stigma quartile. The HCQI score distribution was rightskewed. To assess the HCQI predictors, the score was categorized by quartiles (<70, 70-79, 80-89, ≥90 out of 100 points), and ordinal regression analysis was applied. Multivariable regressions for stigma outcomes and HCQI included variables that were statistically significantly (p < 0.05) in simple regressions with one predictor in addition to exposure of interest (co-located care for stigma outcomes, co-located care and intersectional stigma for HCQI). Presence of multicollinearity was tested with the help of variance inflation factor (VIF). A VIF of 10 was considered as evidence of collinearity. Missing data were excluded from the analysis in the following variables: number of sexual partners (n = 12), history of selling sex (n = 1), smoking (n = 1), AUDIT-C (n = 1), history of TB (n = 1), diagnosed with HCV (n = 9), history of suicide (n = 1), OAT drug (n = 1), receiving regular HIV check-up (n = 1), CD4 testing (n = 7), receipt of HCV screening (n = 1), receipt of referrals (n = 1), HIV stigma (n = 1) and substance use stigma (n = 1). We conducted all analyses in R version 3.5.2 (The R Foundation for Statistical Computing, Vienna, Austria).

| Ethics
The study protocol and instruments received ethical approval from the Institutional Review Boards at Boston University Medical Campus, The Miriam Hospital, and the Ukrainian Institute on Public Health Policy. All study participants provided informed consent.

| Characteristics of the study participants
Of the 191 participants, close to half (47%) were recruited at healthcare facilities with co-located care ( Table 1). The sample was predominantly male (75%), with a mean age of 40 years (SD: 7 years).
Patients at OAT only sites reported more substance use (53% compared to 38% at co-located care) and more unhealthy alcohol use (30% vs. 18% at co-located care), while condomless sex was more frequent among patients in co-located care (46% compared to 32% at OAT only sites).
Most study participants (77%) received methadone, all the rest buprenorphine. On average, participants were satisfied with the health services they received and with their social support. Mean scores were 76 (SD: 22) out of 100 for accessibility; 73 (SD: 24) for user-friendliness, and 73 (SD: 22) for social support.

| HIV stigma
The mean HIV stigma score of 49 out of 100 (SD: 12) was identical at both co-located OAT and ART sites and OAT only sites. In adjusted analyses, a history of suicide attempt was associated with greater HIV stigma (adjusted beta (ab): 5.90; 95% CI: 2.05, 9.75, Table 2).

| Substance use stigma
Similarly to HIV stigma, substance use stigma scores did not differ at co-located OAT and ART sites or OAT only facilities, with a mean score of 53 out of 100 (SD: 13). On average, when comparing standardized scales, the level of substance use stigma was higher among study participants compared to HIV stigma (mean difference: 3.91; 95% CI: 1.39, 6.42).

| Intersectional stigma
HIV and substance use stigma scores moderately correlated with each other (Spearman q = 0.415, p < 0.001). Of 190 study participants who replied to both HIV and substance use stigma questions, 20 (10.5%) had both HIV and substance use stigma scores above the 75th percentile, while 10.5% (n = 20) had only HIV stigma score above the 75th percentile and 14% (n = 26) had only substance use stigma score above the 75th percentile (Figure 1). The remaining 65% (n = 124) had medium or low scores (below the 75th percentile) for both forms of stigma.

| DISCUSSION
In our study conducted at seven sites in Ukraine, receipt of co-located HIV and OAT services was associated with better perceived quality of HIV care but was not linked to stigma levels. We found that regardless of the model of care (co-located OAT and ART sites or OAT only services), one in ten people with HIV who injected drugs experienced high levels of both HIV and substance use stigma, which was associated with poorer mental health and less social support from NGOs. Intersection theory stipulates that multiple stigmas, rather than having additive effects, interact to create "intersectional" stigma [14]. Qualitative studies will be necessary to explore how both stigma forms intersect in this population, and to explain substance use stigma's role in people who are also affected by HIV stigma.
While co-located OAT and ART services care model might aim at reducing intersectional stigma related to HIV and substance use, other structural factors (e.g. social positions, processes, policies and environmental issues) may influence one's stigmatization in the health care setting [39]. Thus, despite its potential to improve the HIV care cascade, care co-location alone might not mitigate stigma. Unless privacy and confidentiality are protected for those who have not only addictions but also HIV, patients of co-located OAT and ART sites may have concerns with their HIV status disclosure within the care setting [20,21,40]. In addition, other barriers such as concerns  over HIV treatment efficacy, safety and tolerability might add to stigma [40].
According to this study's results, patients attending co-located OAT and ART sites more frequently reported receipt of ART, uptake of isoniazid preventive TB treatment and receiving referrals for additional services, including psychosocial support. These findings on care quality are consistent with previous studies showing that co-located HIV and addiction substance use treatment services improve accessibility of care, and therefore increase uptake of services [20][21][22].
In our study, three of four co-located OAT and ART sites had also an NGO providing social support and harm reduction services for PWID, while the respective proportion was one in three at OAT only sites. In Ukraine, NGOs take an active role in advocating for user-friendliness of services for key populations, such as attentiveness, respect and clarity of explanation  = 12), history of selling sex (n = 1), smoking (n = 1), AUDIT-C (n = 1), history of TB (n = 1), diagnosed with HCV (n = 9), history of suicide (n = 1), OAT drug (n = 1), receiving regular HIV check-up (n = 1), CD4 testing (n = 7), receipt of HCV screening (n = 1), receipt of referrals (n = 1), HIV stigma (n = 1) and substance use stigma (n = 1).  in interaction between health providers and clients [41]. According to our findings, receipt of NGO services was associated with lower substance use stigma. However, it was not associated with better perceived quality of care. There are several limitations to this study. The HIV and substance use stigma scales have not been validated in Ukraine, but we conducted factor analysis to assess internal consistency and validity. Given the cross-sectional design, we cannot imply direction of the observed associations. We recruited participants at OAT sites, while participants' stigma scores might primarily reflect previous experiences in other facilities and communities outside of current treatment settings. The accuracy of clinical information might have been limited by recall error. Finally, study results cannot be generalized to the entire population of HIV-positive PWID in Ukraine, as only 6% of the estimated HIV-positive PWID population receive OAT [29].
These limitations notwithstanding, this study's results have several implications. First, treatment integration alone does not seem sufficient to address stigma among people with both substance use disorders and HIV. Stigma interventions, such as Acceptance and Commitment Therapy [42], targeted to HIV-positive people with a history of substance use should therefore be designed and tested in both co-located and not co-located healthcare settings. Second, given the high intersectional stigma we found in this study, stigma interventions might benefit from incorporating an intersectionality framework to reduce shame and devaluation related to both substance use and HIV. Third, stigma's association with suicide attempts and poor mental health might need particular   High stigma score was defined as a score above the 75th percentile and low/medium stigma as a score below the 75th percentile.; c 1U.S. dollar = 26 UAH.; d Due to the limited sample size, history of suicide and type of OAT drug were not included in the multivariable model despite their significant association with HCQI in unadjusted analysis. attention. Stigma interventions need to consider mental health in their approach. Finally, NGOs with direct reach to stigmatized populations might be well suited to develop and implement stigma reduction interventions to empower affected people to cope effectively with stigma and its implications.

| CONCLUSION
Patients receiving substance use treatment in Ukraine report high levels of both HIV and substance use stigma linked to poorer mental health and less social support by NGO. Co-located treatment did not seem to provide a stigma-free environment for this key population. Stigma interventions for HIVpositive PWID, possibly delivered involving peers at NGOs, are required to help affected people cope with negative emotions and experiences associated with stigmatization.

C O M P E T I N G I N T E R E S T S
The authors declare that they have no conflict of interest. YS and TK drafted the manuscript. YS created data collection tools and conducted the data analyses, with substantial contributions to the analytic plan from KL and TK. KL and TK conceived of and designed the study, with substantial contributions from JS and TF. OM and TK led the data acquisition, with substantial contributions from JJC, NR, AC, SB, BI, AD and FSG. OM, JJC, NR, AC, SB, BI, AD, FSG, JHS, TF and KL substantively revised the manuscript for important intellectual content.