“I refused to drink but they still forced me”: A mixed-methods approach to understanding the pathways to reduce alcohol use among Vietnamese people with HIV

Abstract

Aims.

This study explores the effects of two evidence-based alcohol reduction counseling interventions on readiness to change, alcohol abstinence self-efficacy, social support, and alcohol abstinence stigma among people with HIV (PWH) who have hazardous alcohol use in Vietnam.

Methods.

PWH receiving antiretroviral therapy (ART) were screened for hazardous drinking and randomized to one of three study arms: combined intervention (CoI), brief intervention (BI), and standard of care (SOC). A quantitative survey was conducted at baseline (N=440) and 3-month post-intervention (N=405), while in-depth interviews were conducted with a subset of BI and CoI participants at baseline (N=14) and 3 months (N=14). Data was collected from March 2016 to August 2017. A concurrent mixed-methods model was used to triangulate quantitative and qualitative data to cross-validate findings.

Results.

At 3 months, receiving the BI and CoI arms was associated with 2.64 and 3.50 points higher in mean readiness to change scores, respectively, compared to the SOC group (BI: β=2.64, 95% CI: 1.17–4.12; CoI: β=3.50, 95% CI 2.02–4.98). Mean alcohol abstinence self-efficacy scores were 4.03 and 3.93 points higher among the BI and CoI arm at 3 months, compared to SOC (BI: β=4.03, 95% CI: 0.17–7.89; CoI: β=3.93, 95% CI: 0.05–7.81). The impacts of the interventions on social support and alcohol abstinence stigma were not significant. Perceived challenges to refusing drinks at social events remained due to strong alcohol abstinence stigma and perceived negative support from family and friends who encouraged participants to drink posed additional barriers to reducing alcohol use.

Conclusions.

Both the CoI and BI were effective in improving readiness to change and alcohol abstinence self-efficacy among PWH. Yet, participants still faced significant barriers to reducing their drinking due to social influences and pressure to drink. Interventions at different levels addressing social support and alcohol abstinence stigma are warranted.

Introduction

Hazardous alcohol use, defined as a quantity or pattern of alcohol consumption that increases adverse health outcomes for an individual (Reid et al., 1999), is prevalent among people with HIV (PWH). PWH are two to four times more likely to report hazardous drinking compared to the general population (Park et al., 2016), and global data show that a quarter to almost half of PWH report hazardous drinking (Silverberg et al., 2018, Veld et al., 2017, da Silva et al., 2017). Hazardous alcohol use among PWH is linked to increased sexual risk behaviors (Reis et al., 2016) and negatively affects HIV treatment outcomes by reducing antiretroviral treatment (ART) adherence and increasing the risk of treatment failure and mortality (Chander et al., 2006, Cook et al., 2001, Gross et al., 2017, Hendershot et al., 2009, McCance-Katz et al., 2013).

Motivational Enhancement Therapy (MET) and Cognitive Behavioral Therapy (CBT) are effective interventions for reducing hazardous alcohol use among PWH (Brown et al., 2013, Chander et al., 2015, Parsons et al., 2007). MET is based on Motivational Interviewing (MI) — a directive, client-centered counseling style that provides personalized feedback and facilitates behavioral change by exploring, resolving ambivalence, and eliciting clients’ personal motivations for change (Miller and Rollnick, 2012). CBT consists of a group of psychological therapies that aim to address negative emotions and dysfunctional behaviors by acknowledging the effects of the environment, cognition, language and human learning on behaviors (Vernon and Doyle, 2017).

In Vietnam, high levels of alcohol use are normalized during social gatherings, where individuals encourage each other to drink to intoxication (Lincoln, 2016). According to the World Health Organization (WHO), it is estimated that 4.7% (8.9% of men, 0.9% of women) of the population in Vietnam have an alcohol use disorder, and 2.9% (5.9% of men and 0.1% of women) are alcohol dependent (WHO, 2014). A study by Tran et al. found that 30% of PWH on ART reported hazardous drinking in Vietnam (Tran et al., 2013). Moreover, Vietnamese men are disproportionately affected by drinking problems, as compared to women (Lincoln, 2016). A study in rural of Vietnam found that binge drinking only existed among men, and the prevalence of alcohol consumption-related problems was more than 30 times higher among men (Giang et al., 2008).

In 2018, two alcohol interventions were culturally adapted for ART clients in Vietnam: a combined intervention (CoI) — which combined aspects of both MET and CBT, and a brief intervention (BI) — a more compact version of the combined intervention (Go et al., 2020). The combined and brief interventions were associated with significant increases in the percentage of days abstinent from alcohol, compared to a standard of care (SOC) group at 12 months; viral suppression (ie, <20 copies of HIV-1 RNA per milliliter) at 12 months was higher for the BI arm than the SOC, but no significant difference between the CoI and SOC was observed (Go et al., 2020). Even though the interventions were effective in reducing alcohol use, little is known about their effects on proximal factors that may have affected drinking avoidance. According to the Information-Motivation-Behavioral Skills (IMB) model developed and tested by Fisher and Fisher (Amico, 2011, Fisher and Fisher, 1992), people must be informed, motivated, and behaviorally skilled to change behavior before they can initiate and maintain the changes. Readiness to change and alcohol abstinence self-efficacy may be predictors of alcohol consumption in various populations (Gaume et al., 2016, DiClemente et al., 2009, Britton, 2004). In addition, social support is a key factor that strongly influences the abilities of people with alcohol use disorders to reduce their alcohol use (Nelson et al., 1999, Hunter-Reel et al., 2010, Hershow et al., 2018). Since alcohol use is pervasive and normative in Vietnam, stigma associated with alcohol abstinence and alcohol refusal can also play an important part in the decision making of PWH (Lancaster et al., 2020, Hershow et al., 2018). The understanding of how alcohol reduction interventions influence these factors will shed light on how to further refine and scale up these interventions in Vietnam and in other settings.

Mixed-methods research, which integrates quantitative and qualitative methods, allows us to explore the research questions from a multi-dimensional lens (Regnault et al., 2017). However, mixed-methods research among PWH with hazardous alcohol use remains rare, and no such evaluation has been conducted among PWH with hazardous alcohol use in Vietnam in particular. This study aims to (1) assess the impacts of two alcohol reduction counseling interventions on readiness to change, alcohol abstinence self-efficacy, social support, and alcohol abstinence stigma among ART clients with hazardous alcohol use in Thai Nguyen, Vietnam at 3 months after randomization and (2) understand the perceived effectiveness of the BI and CoI in changing these factors.

Methods

Study design and setting

Data for this analysis come from a three-arm randomized controlled trial conducted among ART clients with hazardous alcohol use in Thai Nguyen, Vietnam, where the HIV epidemic is primarily driven by injection drug use (Lim et al., 2014). Thai Nguyen is a mountainous, multiethnic province located in the Northeast of Vietnam. Out of 12 government-run outpatient ART clinics in Thai Nguyen, we recruited participants from and conducted the trial in the 7 outpatient ART clinics (6 community clinics and 1 hospital clinic) with the highest numbers of ART clients.

The trial, titled “Reducing hazardous alcohol use & HIV viral load: a randomized controlled trial in ART clinics in Vietnam” (REDART-NCT02720237), compared CoI and BI, both with each other and with a SOC arm. Participants were screened for hazardous alcohol use and were enrolled into the study if they (1) were a current client receiving ART at one of 7 chosen ART clinics in Thai Nguyen; (2) had an Alcohol Use Disorders Identification Test – Consumption (AUDIT-C) score ≥ 4 for men; AUDIT-C score ≥ 3 for women; (3) were 18 years of age or older; and (4) planned on residing in Thai Nguyen for the next 24 months. Participants were excluded if they scored a point of 10 or more on the Clinical Institute Withdrawal Assessment, which evaluated the risks of alcohol withdrawal.

Enrolled participants were individually randomized into one of three study arms at a 1:1:1 ratio. Trained core research staff generated the randomization schedule using permuted-block randomization with a block size of 3 (using SAS 9.4 (SAS Institute, Inc., Cary, North Carolina, USA)). All eligible participants at one clinic were randomized at that clinic before study staff moved on to assess eligibility of participants at the next clinic, so that participants randomly assigned within 1 triplet block all belonged to the same ART clinic. Specifically trained and designated staff applied the randomization schedule to each enrolled participant.

Enrolled participants completed assessments at baseline, 3-, 6-, and 12-months post-randomization. The primary study outcomes of the parent study were percentage of days abstinent from alcohol and viral suppression at 12 months. For the two intervention arms, qualitative interviews were conducted at baseline and 3-months post-randomization for a subset of intervention participants. As we sought to assess immediate intervention effects on readiness to change, alcohol abstinence self-efficacy, social support and alcohol abstinence stigma, only data at baseline and the 3-month follow-up visit, which were collected from March 2016 to August 2017, were used for this analysis. At the end of each quantitative survey or qualitative interview, participants were given 100,000 Vietnamese dong (approximately US $4.30) to compensate for their time, travel, and effort.

Interventions

Two alcohol reduction interventions, CoI and BI, were selected and modified using the situated information, motivation, and behavioral skills model (Hutton et al., 2019). The study group implemented a 3-phase approach to culturally adapt the interventions for PWH in Vietnam: (1) selection of alcohol interventions and in-depth interviews with stakeholders to collect information; (2) integration of data on key characteristics of the interventions’ core elements and (3) in vivo testing with clients and counselors (Hutton et al., 2019). The CoI comprised a total of six individual face-to-face sessions and three optional group sessions, with individual sessions occurring approximately one week apart. This intervention combined MET, which raised awareness of one’s own negative drinking behavior, and CBT, which provided different alcohol abstinence self-efficacy to manage triggers for hazardous drinking. During CoI counseling sessions, participants were also asked to role-play and apply drinking refusal skills to common drinking situations. One of the CoI counseling sessions was dedicated to developing social support by involving a support person to help participants reduce drinking. At the session before this one, the counselor asked participants to find and bring in a supportive person whom they trusted to the following session. During the social support session, the counselor highlighted the role of social support, identified the participants’ sources of social support, helped them practice asking for help and build support for reducing drinking, and engaged the support person in this discussion (if the support person was present).

The BI also combined elements from both CBT and MET but condensed them into a shorter format with a more didactic approach and no time dedicated to role playing. Participants in the BI arm received only two in-person counseling sessions and two counseling phone calls. Contents of the BI sessions included the review of drinking patterns, harmful effects of drinking, and alcohol use behavior change strategies. Moreover, no group sessions were offered for the BI arm.

Even though the interventions did not target alcohol abstinence stigma directly, the counseling sessions included discussions on social triggers to drink, strategies to politely refuse drinking in social contexts as well as the reframing of negative thinking into positive thinking. The group sessions offered in the CoI arm also allowed participants to talk about their drinking with one another and get to know others who were trying to cut down or quit drinking, so they saw that drinking did not have to be a normative activity.

Participants randomized to the SOC arm received standard messages from providers to drink less alcohol and referrals to alcohol treatment and infectious diseases treatment (Hepatitis B, C and other sexually transmitted diseases) at a general hospital in Thai Nguyen. More details of the parent study have been previously described (Go et al., 2020).

Data collection

Quantitative data

Demographic information, alcohol use and data on key intervention targets were collected both at baseline and 3-month follow-up. For all participants, the baseline survey was conducted before any counseling sessions were provided. Questionnaires were administered through face-toface interviews in a private room at the project facility with trained interviewers.

Readiness to change.

A three-item adapted version of the 10-item Readiness to Change questionnaire was administered to participants (Budd and Rollnick, 1996). The original scale was validated among male heavy drinkers, and the measure of readiness to change was shown to be correlated with intention to reduce alcohol intake and reported reduction in intake (Budd and Rollnick, 1996). Participants were asked on a scale of one to ten about the readiness to, importance of and confidence in changing their alcohol use behavior. Participants’ scores were summed; higher scores indicated greater readiness to change alcohol use behavior. The total score ranged from 3 to 30.

Alcohol abstinence self-efficacy.

Participants were administered a 10-item alcohol abstinence self-efficacy scale (McKiernan et al., 2011, DiClemente et al., 1994). This scale was adapted from the 12-item brief alcohol abstinence self-efficacy scale (McKiernan et al., 2011) after piloting the survey among a sample of PWH with hazardous alcohol use, using cognitive interviewing. When being validated among those receiving alcohol treatment services, the brief scale has been proven to be a valid and reliable tool to measure alcohol abstinence self-efficacy (McKiernan et al., 2011). Participants were asked on a scale of one to ten how likely they were to drink (temptation sub-scale) or avoid drinking (confidence sub-scale) under various circumstances, such as when attending social events or experiencing emotional distress or cravings. Scores were summed for each sub-scale, and the score from the temptation sub-scale was subtracted from the confidence sub-scale (McKiernan et al., 2011). The final score ranged from −45 to 45.

Perceived social support.

Participants were administered a 5-item version of the Medical Outcomes Study Social Support Instrument (MOS-SS) (Sherbourne and Stewart, 1991) to measure social support. This modified version was created based on the performance of the scale in previous HIV prevention studies in Vietnam and study staff assessment of cultural relevance (Levintow et al., 2018). Even though the modified version has not been formally tested among PWH, the original MOS-SS scale showed excellent internal consistency (Cronbach’s α=0.97) and convergent validity among methadone maintenance treatment patients in Vietnam (Khuong et al., 2018). Participants were asked on a scale from 1 to 10 how often they received social support (ranging from none of the time to most of the time). Scores were converted to a 0–100 score, with higher scores indicating greater social support (Nguyen et al., 2020).

Alcohol abstinence stigma.

A 7-item alcohol abstinence scale developed by Lancaster et al. was used to capture experienced, internalized and anticipated stigma about alcohol refusal in social settings (Lancaster et al., 2020). The scale showed good internal consistency (Cronbach’s α=0.75) and convergent validity when being validated among a sample of PWH with hazardous alcohol use in Vietnam (Lancaster et al., 2020). Participants were asked on a scale of 1 to 10 to what extent they strongly disagreed or strongly agreed with each item. The answers were then recoded to a five-point Likert scale (ranging from 0 to 4). Participants were categorized as reporting any alcohol abstinence stigma if they reported a score above zero and reporting no stigma with a score of zero.

Alcohol consumption.

Data on alcohol consumption were collected using the full Alcohol Use Disorders Identification Test (AUDIT) questionnaire, which contained 10 questions on the frequency and amount of alcohol consumption scale (Babor, 2001). Alcohol dependence, defined as a strong desire to consume alcohol, difficulties in controlling its use, and persistent use despite harmful consequences (NCCMH UK, 2011), was evaluated with the Mini International Neuropsychiatric Interview (MINI) questionnaire (Sheehan et al., 1998) – a 7-item structured diagnostic psychiatric interview. Answering Yes to 3 or more items indicated alcohol dependence (Francis et al., 2015).

All validated scales described above are presented in the Electronic Supplementary Material.

Qualitative data

For qualitative interviews, we purposively sampled a sub-group of participants from the two intervention arms so that there were clients from all 7 ART clinics and there were both male and female participants from each intervention arm. We selected a sample size of 14, so that we had 7 participants in each arm (BI and CoI) and 2 participants from each clinic (one from each arm). With this sample size, we saw recurring themes across ART clinics by arm, suggesting that the number of participants was sufficient for thematic saturation. The selection of these participants was also spread out over the two-year recruitment period. All in-depth interviews were conducted in person by an interviewer, in a private room at the project office in Thai Nguyen. Two interviewers with many years of experience in conducting qualitative research in public health research went through a one-day training session on the contents of the interview guide and interview skills. During the training, the study team explained to the interviewers the goals of the interviews, the meaning of each question and practiced role-play with them. For each of the first few interviews, the study team reviewed the transcripts and gave the interviewers feedback before they conducted the next interview. At baseline, the interviewers explored participants’ drinking habits, reasons for drinking, barriers to reducing alcohol use, and other perceptions of alcohol use using questions such as “Tell me about the reasons or the people in your life that caused you to drink alcohol” and “If you wanted to reduce your drinking, what might be the biggest challenges?”. At 3-month follow-up, participants were also asked about their experiences with alcohol use since the last interview, changes in drinking and their perceptions of the interventions. Sample questions included “Over the past 3 months, what has helped you to drink less alcohol?” and “Tell me about the counseling sessions that you have had. Which strategy for refusing alcohol did you find the most helpful?”. Detailed questions of the interview guides were included in the Electronic Supplementary Material. All interviews were audio-recorded.

Data Analysis

Quantitative data

We performed descriptive and correlational statistical analyses using SAS 9.4 (SAS Institute, Inc., Cary, North Carolina, USA). Data with enrolled participants at baseline (N=440) and 3 months (N=405) were used. Frequencies and percentages for categorical variables and means with standard deviation (SD) for continuous variables were reported. We used one-way ANOVA for continuous variables and Wald χ² for categorical variables, to assess comparability across study arms at baseline.

For continuous outcomes, we used multiple linear regression to estimate intervention effects on readiness to change, alcohol abstinence self-efficacy, and perceived social support. For categorical outcomes, we used logistic regression to estimate intervention effects on HIV and alcohol abstinence stigma. All multivariable models were adjusted for age, sex, education, marital status, employment status, and factors affecting drinking avoidance at baseline. Thirty-five participants (8.0%) with missing data at follow-up were not included in the multivariable analyses.

Qualitative data

Interviews were conducted in Vietnamese, then transcribed, translated into English, and imported into NVIVO 11 (QSR International Pty Ltd., Chadstone, Victoria, Australia). Interviews were coded and analyzed in English. We applied an iterative approach to coding whereby codes were derived from the interview guides to capture the common reasons for reducing or not reducing alcohol use; we added codes that emerged from the interviews. Three coders (RH, QB, and NB) coded the interviews and conducted a content analysis. Four out of 14 interviews were co-coded by three coders to assess inter-rater reliability. Differences in coding were resolved through discussion. Specifically, for each set of time points (baseline and 3-months), the coding of in-depth interviews was done in three rounds. Three coders first co-coded one interview, discussed the results, and fleshed out the codebook. They then co-coded a second interview, came together to discuss any differences, and made further revisions to the codebook. The coders finally coded the remaining interviews separately.

Mixed-methods analysis

A mixed-methods model was applied to the analysis (Steckler et al., 1992). We used qualitative data to elucidate the mechanisms through which readiness to change and alcohol abstinence self-efficacy were or were not improved as well as the complexity of alcohol abstinence stigma and social support’s roles on drinking habits.

The study was reviewed and approved by the University of Chapel Hill Institutional Review Board (IRB) and the local IRB at the Thai Nguyen Center for Preventive Medicine.

Results

Quantitative Results

Sociodemographic characteristics and intervention uptake

Thirty participants were excluded from the parent study due to the high risks of alcohol withdrawal at enrollment. Another 11 eligible participants were not enrolled because they did not complete the baseline assessment or the laboratory testing. In the end, 440 participants were enrolled into the trial.

The average age of the sample was 40.2 years old (SD=5.7; range=25–60), 72.3% were married/living with a partner and 81.1% reported full- or part-time employment. Participants, overwhelmingly, were male (96.8%). The median AUDIT score in the sample was 12 (Interquartile range: 9–16), and 21.1% of participants reported having alcohol dependence (Table 1).

Table 1.

Participants’ characteristics at baseline by intervention arm.

	Total (N=440)	Standard of care (N=146)	Brief Intervention (N=147)	Combined Intervention (N=147)	p-valuea
Mean age, years (SD)	40.2 (5.8)	40.3 (5.9)	39.8 (5.6)	40.5 (5.8)	0.57
Sex (male) (N(%))	426 (96.8)	141 (96.6)	140 (95.2)	145 (98.6)	0.28
Highest level of education (N(%))					0.80
Technical training/College or university or less	42 (9.6)	16 (11.0)	17 (11.6)	9 (6.1)
High school or less	86 (19.6)	24 (16.4)	30 (20.4)	32 (21.8)
Secondary school or less	246 (55.9)	89 (61.0)	74 (50.3)	83 (56.5)
Primary school or less	66 (15.0)	17 (11.6)	26 (17.7)	23 (15.7)
Marital status (N(%))					0.94
Married/Living with partner	318 (72.3)	105 (71.9)	105 (71.4)	108 (73.5)
Single	66 (15.0)	23 (15.8)	23 (15.7)	20 (13.6)
Widowed/Divorced/Separated	56 (12.73)	18 (12.3)	19 (12.9)	19 (12.9)
Employed full- or part-time (N(%))	357 (81.1)	114 (78.1)	120 (81.6)	123 (83.7)	0.47
Alcohol dependence at baseline (N(%))	93 (21.1)	32 (21.9)	32 (21.8)	29 (19.7)	0.88
Median AUDIT score (IQR)	12 (9–16)	11 (9–15)	12 (9–16)	12 (9–16)
Mean readiness to change (SD)	20.0 (6.6)	20.2 (6.6)	20.2 (6.9)	19.6 (6.5)	0.71
Mean alcohol abstinence self-efficacy (SD)	22.5 (19.5)	22.8 (20.4)	21.9 (18.8)	22.7 (19.4)	0.91
Mean perceived social supportb (SD)	67.0 (26.3)	66.9 (25.7)	66.3 (26.5)	67.7 (26.9)	0.90
Having alcohol abstinence stigma (N(%))	249 (56.6)	81 (55.5)	89 (60.5)	79 (53.7)	0.47
Viral suppression (N(%))	370 (84.1)	117 (80.1)	130 (88.4)	123 (83.7)	0.15

Note. SD=Standard deviation; IQR: interquartile range

^aFor continuous variables, p-values are from the one-way ANOVA test; for categorical variables, p-values are from Wald χ² test.

^bOne participant had missing data on social support and three had missing data on ART adherence at baseline.

In brief, among participants in the CoI arm, 98.0% (144) attended the first counseling session, while 76.2% (112) attended all six required sessions. In the BI arm, 95.2% (140) attended at least one in-person session, while 84.4% (124) attended all two in-person and two phone counseling sessions. More details of the intervention uptake were reported elsewhere (Go et al., 2020).

Intervention effects on readiness to change, alcohol abstinence self-efficacy, perceived social support and alcohol abstinence stigma at the 3-month follow-up visit

At 3 months, compared to the SOC arm, participants in the BI arm had 2.64 more points in the mean readiness to change score (BI: β=2.64, 95% CI: 1.17–4.12), and participants in the CoI arm had 3.50 more points in the mean readiness to change score (CoI: β =3.50, 95% CI 2.02–4.98). Participants in the BI and CoI arm also reported 4.03 and 3.93 more points in the mean alcohol abstinence self-efficacy score at 3 months, as compared to SOC participants, respectively (BI: β =4.03, 95% CI: 0.17–7.89; CoI: β =3.93, 95% CI: 0.05–7.81). There were no significant intervention effects on perceived social support or alcohol abstinence stigma at 3 months (Table 2).

Table 2.

Intervention effects on readiness to change, alcohol abstinence self-efficacy, perceived social support and alcohol abstinence stigma at the 3-month follow-up visit.^a

	Undjustedb		Adjustedc
	β coefficientd (SE)	95%CI	β coefficientd (SE)	95%CI
Intervention effects on readiness to change
BI	2.59 (0.74)	1.13–4.05	2.64 (0.75)	1.17–4.12
CoI	3.34 (0.74)	1.88–4.80	3.50 (0.75)	2.02–4.98
Intervention effects on alcohol abstinence self-efficacy
BI	4.10 (1.95)	0.26–7.94	4.03 (1.97)	0.17–7.89
CoI	3.49 (1.96)	(−0.36)–7.33	3.93 (1.97)	0.05–7.81
Intervention effects on perceived social support
BI	0.92 (3.07)	(−5.13)–6.96	1.04 (3.03)	(−4.91)–7.00
CoI	3.83 (3.07)	(−2.21)–9.87	3.02 (3.03)	(−2.94)–8.99
	aOR	95%CI	aOR	95%CI
Intervention effects on alcohol abstinence stigma
BI	0.82	0.50, 1.33	0.79	0.48, 1.30
CoI	1.37	0.84, 2.23	1.35	0.82, 2.22

Note. Estimates significant at p-values<0.05 are in bold; BI: brief intervention; CoI: combined intervention; SE=Standard error; aOR=Adjusted odds ratio; CI=Confidence interval.

^aStandard of care is the reference group in all models.

^bFor each unadjusted model, factors affecting drinking avoidance at baseline were included as covariates.

^cFor each adjusted model, covariates included factors affecting drinking avoidance at baseline, age, sex, education, marital status, and employment status.

^dβ coefficients presents the mean difference in the continuous scores of readiness to change and alcohol abstinence self-efficacy between arms.

Qualitative Results

A total of 14 in-depth interviews were conducted at baseline (after the baseline quantitative survey had been completed) and 14 interviews at 3-month post-intervention, with 7 participants in each intervention group at each round. For those in the CoI arm, 3 participants completed a baseline interview without the 3-month follow-up interview as they withdrew from the study and did not complete all sessions. As such, 3 additional participants were invited to replace them and only completed the 3-month in-depth interview. The interview duration range was from 47 to 88 minutes, and the mean duration was 62 minutes. The mean age of those interviewed was 39 years old, and one in each intervention group was female.

At the baseline visit

Readiness to change and alcohol abstinence self-efficacy

Regarding readiness to change, at baseline half of all participants thought that they could reduce their alcohol use due to their own motivation. Various health concerns were the most frequently mentioned motivations for reducing drinking. These participants wanted to limit their alcohol use to avoid health problems of drinking, such as liver or stomach problems, tiredness, fatigue, discomfort and HIV-related health issues. Financial reasons were not mentioned by any participants as a reason for reducing drinking. Twelve participants said they were motivated to change their drinking behaviors because of their HIV diagnosis, explaining that they had been informed of negative consequences of drinking on their health status by their HIV providers.

However, to some participants, HIV-related stress and sadness could also make them drink more, as described by one male participant:

“… I often think of my daughter, it is no problem for me to be infected, but thinking about my wife and daughter, I am sad, then I drink more, during that time I drink more. Drinking to forget things, you know, there are things that you can’t re-do. For me, it does not matter, but affecting my wife and child, I am very much regretful” (Male, CoI arm, PTID 601512)

Four participants said that they had difficulties changing their drinking habits, explaining that they could not control their brain or control themselves. Another four participants mentioned that they had used different techniques to try to limit their alcohol consumption prior to receiving the intervention, such as using health issues as an excuse not to drink. However, these participants explained that it was very challenging, and even impossible to refuse alcohol at social events such as weddings, funerals or Tet – the Lunar New Year in Vietnam.

Perceived social support and alcohol abstinence stigma

Eleven participants talked about family as a reason for not drinking. They mentioned their wives, children and mothers as people who encouraged them to reduce alcohol use by giving them verbal reminders, encouraging them to drink less or hiding their alcohol. Interestingly, family was also listed as an important reason for drinking alcohol for five participants. Drinking usually occurred when participants went to social events (e.g., weddings, funerals) with their family, when they had family issues that created stress or when they were worried about the risk of transmitting HIV infection to their wife and children. Contrary to female family members who usually encouraged participants to reduce drinking, male family members often had a negative influence on their drinking. For example, one participant explained how he inherited addictive drinking behaviors from his male family members:

“…since my great grandfather, to my grandfather and my father, they all drank until the day they died. They could all hold their liquor very well. It’s just that they did not die because of drinking. Therefore, it also affects me” (Male, CoI arm, PTID 501089)

A major theme described by participants was that by taking part in social situations with friends and business colleagues, there was increased social pressure to drink. Many participants talked about drinking in social events as a tradition, a culture that was an inseparable part of their everyday life. Meeting with friends, especially at special events such as weddings, Tet, or business meetings with colleagues were mentioned as a frequent reason for drinking. Even though no participants directly labeled their attitudes toward drinking as alcohol abstinence stigma, some participants considered refusing to drink at such events disrespectful, and even “weird” behaviors that they did not want to engage in.

At 3-month follow-up visit

The second round of interviews were conducted shortly after the intervention had been completed. In general, participants had positive feedback on the interventions, stating that the interventions did not only improve their knowledge on negative consequences of heavy drinking on health but also provided them with many tools to help reduce their alcohol use. There were no substantial differences in perceived changes in readiness to change, alcohol abstinence self-efficacy, social support and alcohol abstinence stigma of participants in two intervention arms.

Readiness to change

All participants talked about the positive impacts of the intervention on their readiness to change, describing how they were motivated to reduce alcohol use and how confident they were about their ability to control their drinking. Participants mentioned that after receiving the interventions they understood the harms of drinking and knew what they needed to do to limit this behavior. They were also more aware of the amount of alcohol consumed instead of drinking large amounts without thinking about it, or “drinking as a habit” as they used to do. Some participants emphasized the importance of one’s “bravery”, “determination” and “will”, explaining that no one could force them to drink if they did not want to.

“I have no problems related to my mind or my thought. I just determined that I wouldn’t drink anymore, then I didn’t drink. When I have made a decision on my mind that I wouldn’t drink anymore, then I could do it. Making up your mind is the most important thing” (Male, BI arm, PTID 601528)

Improving one’s health was again the most common motivation for participants to reduce drinking at the 3-month visit. They talked about their wishes to avoid interaction between alcohol and ART, to be healthier and to live longer. Other reasons for reducing drinking included being afraid that drinking would interfere with work and their families’ welfare.

“Because I wish my health is better, so that I can take care of my child, now the child is still small, and I am the head of a family, I want to be healthy to look after my child, and take care of my family’s economy” (Female, BI arm, PTID 400381)

Alcohol abstinence self-efficacy

The majority of participants in both interventions stated that they were able to use various strategies to cut down or quit drinking after receiving the interventions. These strategies included refusing to drink at social events, avoiding social events, finding social support, adopting positive thoughts and having other distractions.

More specifically, twelve participants talked about limiting their alcohol consumption at social events by refusing to drink, including BI participants who were not explicitly taught about refusal skills. Some noted mentioning being sick, having hepatitis or other liver diseases to peers as a reason for refusing alcohol, while those who were comfortable disclosing HIV status mentioned taking HIV medication as excuse for not drinking. Other participants talked about actively avoiding social situations to reduce alcohol consumption or using alternative activities such as fishing, cleaning their house, playing games and self-reflection helped distract them from alcohol cravings.

Some participants also talked about the differences between men and women regarding alcohol reduction. They believed that women, especially HIV-positive women, drank less than men and were less likely to be alcoholic. Moreover, women were perceived by both male and female participants to have fewer challenges reducing their drinking, since they did not gather at social events for a drink as much as men. Even when they did, they were usually not forced to drink as men were. One participant believed that men wanted to show off their drinking abilities and had more difficulties refusing a drink because of their egos, while another mentioned that women might be less likely to participate in the intervention for fear of being labeled “alcoholic”.

After receiving the interventions, six participants (three from each arm) thought that it was not easy to refuse drinks offered by others. They explained that they were able to drink less or have a less strong drink, but were not able to totally refuse drinking at social events due to the strong pressure to drink from friends.

“Yes, sometimes I still had to drink because I couldn’t refuse them, I still drank but just a little. That day I said I would definitely not drink but then I still drank a little, didn’t dare to drink much. As I said, when I first came, I refused to drink but they didn’t let me do so. Then I drank a little to please them” (Male, CoI arm, PTID 601501)

We were able to ask CoI participants but not BI participants about their perceptions of group sessions, since these sessions were only offered in the CoI arm. All participants who joined group sessions of the CoI arm considered the group sessions helpful, except for one participant who thought that applying refusing skills at social events was unrealistic:

“Answer: There is one man in this group who made the entire group laugh. For example, he asked: “In case two men try to invite each other to drink, for any event, then how do you refuse?”. I think maybe in the real life he may be frantic, but in here he just responded: “Please do empathize, I am taking medications now” (laugh). Then others added what he would do if it was insisted that he drank, and there were many offers, he answered: “I already said no drinking, then it means no drinking”. In practice, in real society, it is not all the time that you can show that you are sick and with the disease.

Question: What do you think about that situation, the situation he talked about?

Answer: For me, that situation is not practical. Practically, in real life, he would never refuse like that, he just drinks at once.” (Female, CoI arm, PTID 400507)

Perceived social support

Ten participants mentioned the support of family in helping them reduce drinking. Family members, especially spouses and children of participants, usually reminded them of the need to reduce alcohol use or criticized them if they drank too much. Friends were also mentioned as a source of support by three participants, who believed that friends were understanding of their reasons for not drinking.

Because participants in the CoI arm received a separate counseling session on developing social support, they were asked about the impacts of the designated support person on their behaviors. All seven CoI participants described their experience with a support person, but only two participants said that the support person encouraged them to reduce their drinking. The remaining five participants did not find having a support person very helpful for different reasons. For some participants, the support person either lived too far away, was not available to come to support them, or even invited them to have a drink. One participant said that the support person was able to gain knowledge but did not help him much in reducing his alcohol use because they rarely discussed the topic of limiting alcohol use outside of the counseling session.

Alcohol abstinence stigma

At the 3-month follow-up visit, difficulties staying abstinent from alcohol at social events were still mentioned in many interviews. Feeling forced to drink at these events were common among participants, who explained that their friends and family kept inviting them to drink and that they felt “lost” or “uncomfortable” if they were the only ones not drinking. Two participants shared the belief that drinking was necessary to socialize and maintain relationships with friends and business partners, while one participant said that it was hard to encourage his friends to limit their drinking just like he did.

“Sharing [the motivation to reduce drinking] is difficult, you know, I feel that it is hard. In a table, there are people who like to drink, there is only me who does not like drinking much, and not wanting to drink, then they do not like me, it is hard, not easy” (Male, BI arm, PTID 200726)

Discussion

Our study found positive impacts of two alcohol reduction counseling interventions on readiness to change and alcohol abstinence self-efficacy among ART clients with hazardous alcohol use. Our qualitative data enriched quantitative results by providing more insights into key facilitators and barriers of drinking as well as perceived effects of the interventions on these factors. In this trial, after receiving the interventions many participants described strong determination and feelings in control of their own drinking habits, which might be attributed to improved knowledge of health-related consequences of drinking and tools for drinking avoidance. Indeed, the first counseling session of the CoI was dedicated to building motivation for change for participants (Hutton et al., 2019). Since the BI did not offer a separate session on motivation for change as the CoI, this discrepancy between the two interventions might be a reason for the stronger effects on readiness to change for those who received the CoI.

For both intervention groups, we saw significant improvements in alcohol abstinence self-efficacy. Our qualitative interviews provided additional understandings of the alcohol abstinence self-efficacy in the context of social interactions where alcohol consumption was expected and encouraged regardless of health status. Both the CoI and BI interventions focused on teaching participants problem-solving and alcohol abstinence self-efficacy to reduce drinking. Participants in the CoI arm also had the opportunities to role-play by using their refusal skills in various drinking situations and join group sessions, where they discussed their drinking behaviors and refusal skills. However, the qualitative interviews revealed that a possible unintended consequence of the group sessions was a reinforcement of alcohol abstinence stigma; participants telling each other that refusing to drink at social events was unusual and therefore challenging. To our surprise, participants in the CoI arm had higher alcohol abstinence stigma compared to the SOC arm, even though this increase was not statistically significant. Even though CoI participants were taught how to refuse drinks and given more opportunities to practice refusal skills during their counseling sessions, some might not have been very successful in executing these refusal skills. They might have faced more challenges utilizing these strategies in real-life situations, which could lead to more alcohol abstinence stigma and less alcohol abstinence self-efficacy among this group.

In this study, we placed a greater emphasis on qualitative data because these offered the unique ability to capture challenges to reducing alcohol use associated with social actors, cultural norms, and practices. In-depth interviews with PWH did not only help to explain the effectiveness of the interventions (or lack thereof) shown through quantitative data but also shed light on how drinking behaviors of participants were influenced by alcohol abstinence stigma and their broader social environment, even in the context of the interventions. For example, we did not see any changes in perceived social support among participants in the quantitative analysis. This was unsurprising, since our interventions did not focus extensively on providing social support. Our qualitative findings did highlight the role of family, particularly female family members, in supporting changes in hazardous drinking behaviors. However, in the in-depth interviews, we were able to contextualize this finding by exploring how social pressure of drinking was so strong in Vietnam that characterizing “social support” in this way might be a misnomer. Recent research has attempted to further understand the quality of social support and how it can impact vulnerable groups (Newsom et al., 2003, Mitchell et al., 2016). Individuals who routinely engage in hazardous alcohol use with family and friends may actually perceive negative social support that inhibits healthy behaviors (Owens et al., 2018, Mitchell et al., 2017). Previous alcohol research among people with HIV in Thai Nguyen also demonstrated that men often face extreme social pressure to drink (Hutton et al., 2019, Hershow et al., 2018). In addition, the CoI may have helped participants engage a close family or friend in their alcohol reduction efforts but could not directly address the quality of the support person. The social support component of the CoI can be improved by explicitly training participants to identify positive social support that can help them reduce drinking, and not just any support person. Participants should also be encouraged to have more conversations with the support persons about how to further reduce or quit their drinking outside of the counseling contexts.

Our results also corroborated findings from other studies which showed how individuals with social networks not tied to drinking had more motivation, better alcohol abstinence self-efficacy, and more chances of success in reducing their drinking (Nelson et al., 1999, Litt et al., 2009, Hunter-Reel et al., 2010). Focusing on changing individual behaviors might not be sufficient to effectively reduce drinking and maintain low levels of alcohol consumption among hazardous drinkers. Since stigma exists not only on the individual level but also on interpersonal, community, and structural levels (Hatzenbuehler et al., 2013), tackling alcohol abstinence stigma would require multilevel interventions with more holistic approaches that can successfully address the stigma that has been deeply rooted in the cultural norms regarding drinking in countries such as Vietnam. We recommend that future alcohol interventions also focus on discouraging cultural drinking practices that put a lot of pressure on people to drink and promote alcohol abstinence stigma.

Limitations

Our study has some limitations. First, the scales used to measure alcohol abstinence stigma and social support in our study have not been widely validated. However, the only validation study of the alcohol abstinence stigma scale was conducted within the larger REDART trial, making the scale highly relevant to our analysis. Future studies should also develop and test more complex social support measures as people with HIV might receive positive social support from close family and friends and negative social support from their broader, community-based social network. In addition, an overwhelming proportion of our sample was male. Even though we were able to explore some differences in the perceptions of drinking avoidance between male and female participants, our ability to explore this difference in-depth is limited with only two female participants. Other studies have found that the difference in alcohol consumption between sexes in Vietnam is also considerably larger than in other countries (Lincoln, 2016, Giang et al., 2008). With the majority of hazardous drinkers living with HIV in Vietnam being male, the low proportion of female participants in our sample was somewhat representative of the underlying population of PWH with hazardous alcohol use in Vietnam (Tran et al., 2013, Nguyen et al., 2020). However, we admit that our sample might not be generalizable to other PWH populations with greater proportion of women with hazardous alcohol use. Finally, we only had a small sample size for the qualitative component and did not have a formal power calculation for the quantitative component. Even if our quantitative analysis might lack power, our study still showed promising findings on how the interventions can lead to immediate positive changes in proximal factors affecting alcohol use at 3-month post-intervention. Moreover, the concern for small sample sizes is less relevant in qualitative research, which aims to explore perceptions of participants on specific themes, rather than determining a statistic representative of the underlying population (Guest et al., 2006). Since the pathway through which interventions affect behaviors might be complex and often non-linear, qualitative research allowed us to unpack the intervention effects, thereby providing richness and depth to our findings.

Conclusions

Our study contributed to the understanding of the effects of alcohol reduction interventions on key factors affecting drinking, including alcohol abstinence self-efficacy and readiness to change. Moreover, our findings provided important insights on the complexity of contextual factors and how they influenced alcohol reduction in the social drinking culture in Vietnam. With a mixed-methods approach, we were able to use qualitative data not only to inform quantitative results but also to give a voice to the vulnerable population of PWH in Vietnam with hazardous alcohol use.

HIGHLIGHTS.

• Alcohol interventions based on CBT and MET were effective among people with HIV.

• Readiness to change and alcohol abstinence self-efficacy were improved at 3 months.

• Participants faced barriers to reducing alcohol use due to social norms.

• Interventions addressing social support and alcohol abstinence stigma are needed.