Sexualized alcohol and drug use among men who have sex with men and the PrEP retention in Thailand
Department of Psychiatry, Faculty of Medicine, Chiang Mai University, 110 Inthawaroros Road, Sri Phum, Muang Chiang Mai, Chiang Mai 50200, Thailand
Department of Psychiatry, University of California San Diego, 4168 Front St, San Diego, CA 92103, USA
Research Institute for Health Sciences, Chiang Mai University, 110 Inthawaroros Road, Sri Phum, Muang Chiang Mai, Chiang Mai 50200, Thailand
Global Health Research Center, Faculty of Medicine, Chiang Mai University, 110 Inthawaroros Road, Sri Phum, Muang Chiang Mai, Chiang Mai 50200, Thailand
Integrated Substance Abuse Programs, Semel Institute for Neuroscience and Human Behavior, University of California Los Angeles (UCLA), 10911 Weyburn Ave, Ste. 200, Los Angeles, CA 90024, USA
Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles (UCLA) David Geffen School of Medicine, 760 Westwood Plaza, Los Angeles, CA 90095, USA
Department of Community Medicine, Faculty of Medicine, Chiang Mai University, 110 Inthawaroros Road, Sri Phum, Muang Chiang Mai, Chiang Mai 50200, Thailand
⁎Corresponding author. wichuda.j@cmu.ac.thAbstract
Introduction
Sexualized alcohol use (SAU) and sexualized drug use (SDU) involve the use of alcohol and/or drugs in sexual contexts, which may impact the HIV pre-exposure prophylaxis (PrEP) continuum of care. This study examines associations between SAU, SDU, and PrEP retention among Thai men who have sex with men (MSM), and explores the acceptability of mobile health interventions.
Methods
A quantitative study with embedded qualitative data was employed, with one hundred MSM recruited from a community clinic to complete a survey assessing SAU, SDU, and other health risks. The psychosocial syndemic count was calculated based on symptoms of alcohol and drug use, depression, anxiety, and experiences of physical or sexual trauma. Factors associated with PrEP retention were analyzed using multivariable logistic regression. Additionally, thirty participants who reported SAU and SDU participated in semi-structured qualitative interviews. Rapid thematic analysis was performed to assess the acceptability of mobile health interventions aimed at improving the PrEP continuum.
Results
In the sample, 27 % reported SAU only, 12 % SDU only, and 23 % combining SAU and SDU. SDU included alkyl nitrite (51.6 %), Cannabis (14.5 %), and stimulants (9.7 %). PrEP retention was not associated with syndemic count or SDU, but was associated with SAU and experiences of sexual violence. Qualitative interviews indicated enthusiasm for mobile health interventions, particularly those offering PrEP reminders, incentives for healthy behaviors, and improved PrEP access.
Conclusion
SAU and sexual violence were identified as barriers to PrEP retention among MSM in Thailand, while SDU was not. Mobile health interventions emphasizing pro-health incentives and harm reduction may enhance PrEP adherence in this population.
Highlights
- •Sexualized alcohol use linked to lower PrEP retention in Thai men who have sex with men.
- •Exposure to sexual violence is an independent barrier to PrEP retention.
- •Intoxication during sexualized substance use drives missed PrEP doses.
- •Mobile health tools are highly accepted for PrEP reminders and incentives.
1Introduction
Thailand reports one of the highest prevalences of HIV in the Asia-Pacific region, and rates are highest among MSM (van Griensven et al., 2022). Although HIV incidence among MSM is increasing, efforts to improve access to HIV pre-exposure prophylaxis (PrEP), a medication taken to prevent HIV transmission, have been limited. Expanding PrEP access and optimizing adherence are important strategies to decrease the number of new transmissions and align with the World Health Organization’s (WHO) global target of ending HIV in 2030 (Andrew E et al., 2022, World Health Organization, 2023).
Sexualized alcohol use (SAU) involves the intentional use of alcohol before or during sexual intercourse to enhance or facilitate sexual experiences and is commonly reported among MSM (Carey et al., 2019, Todaro et al., 2024, Wang et al., 2022). SAU often co-occurs with, or leads to, sexualized drug use (SDU) or socially termed “chemsex,” which is the use of drugs to facilitate, enhance, and/or prolong sexual experiences. Although SAU is problematic, SDU has been a subject of increasing concern globally, particularly among young MSM with SDU rates, ranging from 3 % to 29 % (Maxwell et al., 2019, Wang et al., 2023), depending on setting and geographic location. Amphetamine, Mephedrone, and gamma-hydroxybutyrate/gamma-butyrolactone (GHB/GBL) have been frequently reported among MSM in the USA and Western Europe (Bourne et al., 2018, Maxwell et al., 2019) Consistent with this regional context (Lee et al., 2021, Tan et al., 2021b), a sample of Thai MSM at high risk of HIV reported alkyl nitrite (45.9 %), and the injection of crystal methamphetamine (40.2 %) as the most common substances used before or during sex (Cheung et al., 2024).
Despite this evidence, comprehensive Thai-specific prevalence data regarding SAU and SDU among the population remain limited. Available data highlight the substantial integration of alcohol into these patterns: among a cohort of MSM living with HIV in Bangkok, 83.3 % reported alcohol consumption, with risky alcohol use correlating with Hepatitis C and sexually transmitted infections (Muccini et al., 2024). Furthermore, a latent class analysis in a Thai MSM sample identified distinct substance use typologies where, beyond group of "exclusive chemsex use",a separate and substantial group of "sexualized substance use" (13.9 %) was characterized by high probabilities of consuming alcohol (76 %) and Kratom cocktails (79 %) before or during sex (Cheung et al., 2024).
While global studies often highlighted that SDU among MSM is more strongly linked to poor HIV health outcomes than SAU, both sexualized alcohol and drug use can lower inhibitions and increase risky behaviors (Blair et al., 2022, Scott-Sheldon et al., 2016). This reduction in inhibition increases HIV risk behaviors, including condomless anal intercourse, sex with multiple partners, and higher frequency of sexual activity, thereby leading to greater risk of sexually transmitted infections and HIV transmission (Martin et al., 2014, van Griensven et al., 2010). Furthermore, SAU and SDU are also associated with various psychosocial challenges, such as depression, anxiety, suicidal ideation, substance dependence, and psychotic symptoms. (Íncera-Fernández et al., 2021). The overlap of SAU, SDU, mental health issues, and high-risk sexual behavior among MSM strongly aligns with the syndemic theory. This framework posits that the co-occurrence and synergistic interaction of multiple epidemics—like substance use and mental health—are driven by adverse social factors such as stigma, discrimination, and trauma (Bourgeois et al., 2021, Halkitis and Singer, 2018, Hart and Horton, 2017, Pollard et al., 2018). This accumulation of psychosocial factors, often measured as a “syndemic count,” is additively associated with increased HIV transmission and worse adherence to HIV treatment among people living with HIV (Friedman et al., 2015, Singer and Clair, 2003). Crucially, this theoretical framework is highly relevant to understanding and addressing barriers to PrEP access, adherence, and retention within this high-risk population (Bourgeois et al., 2021, Hart and Horton, 2017, Tsai et al., 2017).
In Thailand, the use of social media and hook-up apps to facilitate SDU is common. This practice is primarily organized online, leveraging not only geo-location social networking apps but also public social media such as Twitter and closed groups on instant messaging platforms. The visibility of SDU content on certain platforms contributes to the normalization of these behaviors, potentially providing an entry point to hi-fun, sex parties with SDU, for uninitiated MSM (Witzel et al., 2023). Additionally, social media serves multiple roles within this subculture, being used to source drugs and connect with other people who use drugs, while simultaneously providing information on harm reduction, safe use practices, and access to health sectors. Men who reported sexualized substance use were more likely to report "often" meeting sexual partners online (Cheung et al., 2024, Witzel et al., 2023). Crucially, the online route is a necessary tool for HIV prevention and PrEP access: nearly all participants in one high-risk sample reported meeting sexual partners online (97.2 %) (Cheung et al., 2024). Although many participants were already aware of PrEP, public health interventions targeting certain high-risk groups—such as those who engage in alcohol and poly-substance use but are less aware of their HIV risk—should engage them via online campaigns on geo-location social networking apps (like GRINDR, Jack’D, and Hornet) to raise PrEP awareness and facilitate uptake. Mobile health (mHealth) interventions are a feasible strategy to promote substance use prevention, harm reduction, and self-tracking, as well as to promote PrEP initiation, retention, and adherence. They can be delivered by application (app), text messaging, or the internet (Kazemi et al., 2021, Kazemi et al., 2017, Nelson et al., 2020). Given the magnitude of the potential benefit of PrEP among MSM with SAU and SDU, research is needed on how to create an mHealth intervention that is most acceptable to this population (Hoenigl et al., 2020, Wang et al., 2020, Wu et al., 2021, Yeo and Ng, 2016).
This study aims to describe the relationship between SAU and SDU, psychosocial health factors, and PrEP retention among Thai MSM using a syndemics framework. This study also employed qualitative interviews to explore the potential role of mHealth intervention in promoting PrEP retention, specifically among this population who report SAU and SDU.
2Materials and methods
2.1Participants
This study was conducted between November 2022 and April 2023. A convenience sample was recruited at a community-based HIV prevention clinic in Chiang Mai run by a national organization supporting sexual health and gender and sexual diversity rights in Thailand. We recruited the first 100 MSM with a negative HIV test, aged 18 years and older, presenting to the clinic for PrEP who consented to complete a questionnaire covering general demographic data, mental health conditions, SAU, SDU, and sexual risk behavior, and to share health record information regarding routine 3-month PrEP follow-up. Thirty people endorsing SAU and/or SDU on the study questionnaire were recruited for an in-depth interview.
2.2Quantitative measures
Questionnaires included three main components: Demographic data (e.g., age, education, relationship status), alcohol use, drug use, and history of SAU and SDU, mental health (e.g., depression, anxiety, trauma), and sexual HIV transmission risk behavior. The history of SAU and SDU was defined by asking: (1) ‘Have you participated in sexualized drug use, chemsex, and party and play (PnP) in the past 12 weeks?’, or (2) ‘Have you used any of the following substances to enhance sexual activities in the past 12 weeks?’. The participants were asked to define which group of substances they used for sexual activities by selecting from the list of the Thai version of the Alcohol, Smoking, and Substance Involvement Screening Test (ASSIST). This 12-week recall period was selected to align with a standard recall window used in substance use research, ensuring the assessment captured recent, relevant behaviors.
Because of resource issues relating to management of suicidal ideation at our research site, we assessed depressive symptoms using a modified version of the 9-item Patient Health Questionnaire (PHQ-9), omitting question 9 (suicidal ideation). The resultant PHQ-8 is a reliable measure of depression (Tyree et al., 2019). Current depression was defined as a PHQ-8 score ≥ 10. We used the Generalized Anxiety Disorder-7 (GAD-7) to measure anxiety. The GAD-7 score ≥ 10 was used to identify moderate anxiety. We assessed exposure to physical or sexual trauma using the Trauma History Screening (THS). Any “yes” response to the physical trauma items was noted as positive for exposure to physical trauma. Any “yes” response to the sexual trauma item was noted as positive for exposure to sexual trauma. The severity of alcohol and drug use disorder symptoms was assessed with the Alcohol Use Disorder Identification Test (AUDIT) and Drug Use Disorder Identification (DUDIT). The DUDIT was translated and culturally adapted with review and approval of the assessment tool developer, Dr. Anne H. Berman. AUDIT scores ≥ 8 indicated hazardous use of alcohol, while DUDIT scores ≥ 6 indicated problematic use of drugs other than alcohol. The presence of HIV transmission risk was assessed using seven questions relating to sexual activity according to the risk factors for HIV, consistent with established guidelines for PrEP use (in Appendix 2) (Centers for Disease Control and Prevention, US Public Health Servic, 2021). For the primary outcome, this study defined PrEP retention as completing a routine 3-month PrEP follow-up. All participants were initially assessed by the aforementioned tools as being at risk for HIV and in need of PrEP based on established guidelines.
2.3Statistical analysis
Descriptive statistics were used to describe demographic data, mental health conditions, SAU, SDU, and related sexual health risks. We constructed a psychosocial syndemics count ranging from 0 to 6 based on clinically meaningful substance use (AUDIT ≥ 8, DUDIT ≥ 6), depression (PHQ-8 ≥ 10), anxiety (GAD-7 ≥ 10), physical trauma, and/or sexual trauma. Logistic regression was used to calculate an odds ratio (OR) to assess the association between psychosocial factors, mental health conditions, transmission risk behavior, SAU, SDU, and PrEP retention. Based on the univariable logistic regression analyses, factors associated with PrEP non-retention at a 3-month follow-up visit with a p-value ≤ 0.2 were included in multivariable logistic regression analyses. Backward model selection was utilized in the multivariable regression analysis to identify factors associated with the primary outcome, PrEP retention.
To rigorously test the syndemics framework and the robustness of our findings, we ran a theory-driven multivariable logistic regression predicting PrEP retention, keeping key psychosocial variables in the model regardless of their individual significance. This approach directly evaluates syndemic theory, which holds that co-occurring psychosocial factors interact and jointly worsen health outcomes. The primary exposure of interest was the psychosocial syndemics burden, which was calculated based on six pre-defined conditions. Accordingly, this theory-driven model included all components used to construct the syndemic count: depressive symptoms (PHQ-8), anxiety (GAD-7), symptoms of alcohol use disorder (AUDIT), symptoms of drug use disorder (DUDIT), exposure to physical trauma, and exposure to sexual trauma. To comprehensively model the cumulative risk, the syndemics burden was included in the model both as a continuous count (ranging from 0 to 6) and as a categorical variable (e.g., 0, 1, 2 or more conditions and identified as ‘categorial syndemic burden). Finally, age was included as a standard demographic adjustment covariate. All analyses were done using Jamovi 2.3 (Kerby, 2014, R Core Team, 2021, The Jamovi Project, 2022).
2.4Qualitative interviews
This study utilized a sequential embedded multi-method approach, where the quantitative phase was completed and preceded the qualitative phase. Following the completion of the quantitative part, we conducted in-depth interviews with a purposive subsample of 30 individuals: 15 reported only SAU, and 15 engaged SDU with or without SAU. Open-ended, semi-structured interview questions addressed a broad range of experiences with SAU/SDU, the relationship between SAU/SDU and PrEP retention, and perspectives on potential mobile health interventions. Interviews ranged from 45 to 80 min and were conducted in person in a private room inside the clinic, online via videoconferencing, or over the telephone, depending on participant preference. For this manuscript, we report specifically on those questions related to the role of technology in sexual and drug use behavior and how it may be leveraged to improve health and PrEP retention. All 30 in-depth interviews were conducted by the study's Principal Investigator and lead author (AO). To mitigate the potential for interviewer bias and to encourage candid disclosure on sensitive topics such as substance use and sexual trauma, the PI maintained a non-judgmental stance and emphasized strict confidentiality.
Recorded interviews were transcribed by two professional transcribers (Nevedal et al., 2021). Two authors (AO and SL) reviewed transcripts using a rapid thematic analysis to identify key themes and subthemes that were iteratively refined. They independently developed an initial coding scheme of re-emerging themes to organize the word segments of the transcripts into a coded structure. Microsoft Excel was utilized to manage and display qualitative codes (Ose, 2006). Code lists were then reviewed and compared, and a consensus coding scheme was developed as an Excel matrix (Burla et al., 2008). Quotes that best represented each theme were selected from the database for inclusion in this article by using the consensus from both coders and other investigators. In summary, the analysis process incorporated steps to enhance objectivity, including the utilization of rapid thematic analysis, transcription by two professional transcribers, and the independent review and comparison of the coding scheme by two authors (AO and SL) to establish consensus.
The research was approved by the Research Ethics Committee, Faculty of Medicine, Chiang Mai University. Participants who completed the survey once at the baseline visit and the in-depth interview were compensated 500 THB (~ 14 USD) and 700 THB (~ 20 USD), respectively, for the time and effort.
3Results
3.1Participant characteristics
Of the first 105 MSM who presented to the clinic for a PrEP appointment, five participants declined to participate, citing time constraints. Table 1 describes psychosocial factors between people without substance use (PWoSU), SAU, SDU, and combining SAU and SDU. Twenty-seven percent (27 %) of participants engaged in only SAU, 12 % engaged in only SDU, and 23 % combination of SAU and SDU. The common types of SDU include inhalants, alkyl nitrites (51.6 %), cannabis (14.5 %), and stimulants (9.7 %). Most participants (81 %) used geospatial networking applications to meet sex partners in the previous 12 months.Psychosocial Factors Overall N(%), Mean (SD) (N = 100) PWoSUN(%),Mean(SD) (N = 38) SAU and/or SDU (N = 62) Any SAU N(%), Mean (SD) (N = 50) Compared to PWoSU Any SDU N(%), Mean(SD) (N = 39) Compared to PWoSU Combining SAU with SDU N(%), mean (SD) (N = 27) Compared to PWoSU Demographic Data Age 29.7 (8.3) 29.0 (8.6) 30.4 (7.4) 0.42 29.3 (7.5) 0.88 29.4 (6.77) 0.83 Regular Income 61 (61.0) 23 (60.5) 34 (68.0) 0.47 19 (48.7) 0.30 15 (55.6) 0.69 Education lower than high school 11 (11.0) 4 (10.5) 6 (12.0) 0.83 6 (15.4) 0.53 5 (18.5) 0.35 Living alone 46 (46.0) 15 (39.5) 22 (44.0) 0.67 20 (51.3) 0.30 11 (40.7) 0.92 Relationship status: single 78 (78.0) 32 (84.2) 36 (72.0) 0.17 28 (71.8) 0.19 18 (66.7) 0.10 Coexisting physical illness 10 (10.0) 6 (15.8) 3 (6.0) 0.13 3 (7.7) 0.27 2 (7.4) 0.31 Coexisting mental illness 11 (11.0) 6 (15.8) 4 (8.0) 0.25 4 (10.3) 0.47 3 (11.1) 0.59 Psychosocial Factors Exposure to physical trauma 19 (19.0) 8 (21.1) 8 (16.0) 0.54 8 (20.5) 0.95 5 (18.5) 0.80 Exposure to sexual trauma 10 (10.) 4 (10.5) 6 (12.0) 0.83 6 (15.4) 0.52 6 (22.2) 0.20 GAD-7 equal or greater than 10 5 (5.0) 2 (5.3) 3 (6.0) 0.88 3 (7.7) 0.66 3 (11.1) 0.38 PHQ-8 equal or greater than 10 16 (16.0) 5 (13.2) 10 (20.0) 0.40 10 (25.6) 0.17 9 (33.3) 0.051 AUDIT equal or greater than 8 60 (63.8) 17 (51.5) 37 (75.5) 0.03* 28 (73.7) 0.05 22 (84.6) 0.01* DUDIT equal or greater than 6 12 (12.1) 0 7 (14.3) 0.02* 12 (31.6) < 0.01* 7 (26.9) < 0.01* GAD-7 score 2.78 (3.7) 2.87 (3.4) 2.56 (4.0) 0.70 3.67 (4.3) 0.37 3.78 (4.8) 0.37 PHQ-8 score 4.06 (4.7) 3.95 (4.2) 4.02 (5.1) 0.94 5.56 (5.4) 0.15 6.00 (5.6) 0.10 AUDIT score 10.18 (7.0) 7.61 (5.8) 12.5 (7.6) < 0.01* 13.1 (8.5) < 0.01* 15.6 (8.8) < 0.01* DUDIT score 1.73 (3.1) 0 2.33 (3.8) < 0.01* 4.00 (0.5) < 0.01* 2.0 < 0.01* Sexual Health Risks in 12 months Using sexual hookup applications 81 (83.5) 30 (78.9) 41 (85.4) 0.43 36 (94.7) 0.04* 26 (96.3) 0.05* Transactional Sex 8 (8.1) 3 (7.9) 4 (8.2) 0.96 4 (10.5) 0.69 3 (11.5) 0.62 Condomless sex 57 (57.6) 18 (47.4) 30 (61.2) 0.20 25 (64.1) 0.14 16 (59.3) 0.34 More than 5 sexual partners 51 (51.5) 13 (35.1) 29 (58.0) 0.04* 29 (74.4) < 0.01* 20 (74.1) < 0.01* History of sexually transmitted disease 25 (25.0) 9 (23.7) 13 (26.0) 0.80 10 (25.6) 0.84 7 (25.9) 0.84 Non-consensual sex act 5 (5.1) 1 (2.7) 4 (8.0) 0.30 3 (7.7) 0.33 3 (11.1) 0.17 Encounter sexual violence 11 (11.0) 5 (13.2) 4 (8.0) 0.43 6 (15.4) 0.78 4 (14.8) 0.85 Having sex with people with unknown HIV status 73 (73.0) 24 (63.2) 42 (84.0) 0.03* 31 (79.5) 0.11 24 (88.9) 0.02* Having sex with people with HIV 20 (20.6) 3 (8.3) 13 (26.5) 0.03* 13 (34.2) 0.01* 9 (34.6) 0.01* PrEP Retention in 3 months 77 (77.0) 32 (84.2) 34 (68.0) 0.08 28 (71.8) 0.19 17 (73.9) 0.05 Syndemics Count - Syndrmics Count as continuous data 1.87 (1.5) 1.48 (1.2) 2.11 (1.6) 0.07 2.50 (1.7) 0.01* 2.71 (1.9) 0.01* - Categorial Syndemic Burden 0 10 (11.0) 6 3 0.06 2 < 0.01* 1 0.03* 1 32 (35.2) 14 16 7 5 2 and more 49 (53.8) 13 27 27 18
Participants who reported SAU demonstrated significantly higher rates of problematic substance use compared to PWoSU (those who endorsed neither SAU nor SDU). Specifically, the SAU group was more likely to exceed the AUDIT and DUDIT cutoff for hazardous alcohol and problematic drug use. Furthermore, the SAU group reported significantly higher rates of some risky sexual behaviors than PWoSU. Conversely, the total psychosocial syndemics count was not significantly different between those reported SAU and PWoSU (Table 1).
Participants who reported SAU demonstrated significantly higher rates of problematic substance use compared to PWoSU. Specifically, the SAU group was more likely to exceed the AUDIT cutoff for hazardous alcohol use and the DUDIT cutoff for problematic drug use. Furthermore, the SAU group reported significantly higher rates of some risky sexual behaviors than PWoSU (e.g., more than 5 sexual partners in a year, having sex with people with unknown HIV status.) Conversely, the total psychosocial syndemics count was not significantly different between those who endorsed SAU and PWoSU (Table 1). Participants who reported SDU reported more to exceed the DUDIT cutoff for problematic drug use compared to PWoSU. They were also significantly more likely to report more than 5 sexual partners in a year, having sex with people with HIV or unknown HIV status, than PWoSU. Additionally, the total syndemics count was significantly higher compared to PWoSU. (Table 1)
3.2PrEP retention
Those who denied any SAU or SDU in the previous 12 weeks had the highest rate of PrEP retention (84.2 %), while the lowest retention was found in participants who reported SAU group (68.0 %) (Table 2). In univariable logistic regression analysis, only the presence of SAU, including combined SAU and SDU, was significantly associated with worse PrEP retention (Table 2).Psychosocial Factors Lack of PrEP retention (N = 23)N(% by comlum), Median (IQR) PrEP retention (N = 77)N(% by column), Median (IQR) Univariable logistic regression(p) Odds Ratio (CI) Age 26 (23.5,31.0) 28.0 (23.0,37.0) 0.63 1.01 (0.96–1.08) Regular Income 14 (60.9) 47 (61.0) 0.99 1.01 (0.39–2.62) Education lower than high school 3 (13.0) 8 (10.4) 0.72 1.01 (0.39–2.62) Living alone 10 (43.5) 36 (46.8) 0.78 1.14 (0.45–2.92) Relationship status: single 17 (73.9) 61 (79.2) 0.59 1.35 (0.46–3.97) Any physical health issues 1 (4.3) 9 (11.7) 0.32 2.91 (0.35–24.2) Mental health diagnosis 2 (8.7) 9 (11.7) 0.69 1.39 (0.28–6.94) Exposure to physical trauma 6 (26.1) 13 (16.9) 0.33 0.57 (0.19–1.74) Exposure to sexual trauma 2 (8.7) 8 (10.4) 0.81 1.22 (0.24–6.18) GAD-7 equal or greater than 10 2 (8.7) 3 (3.9) 0.37 0.42 (0.07–2.72) PHQ-8 equal or greater than 10 3 (13.0) 13 (16.9) 0.66 1.35 (0.35–5.23) AUDIT equal or greater than 8 17 (73.9) 43 (60.6) 0.25 0.54 (0.19–1.54) DUDIT equal or greater than 6 3 (13.0) 9 (11.8) 0.88 0.90 (0.22–3.63) GAD-7 score 0 (0,3) 2 (0,5) 0.75 1.02 (0.89–1.17) PHQ-8 score 3 (0.5–7.5) 2 (0−6) 0.63 0.98 (0.89–1.07) AUDIT score** 11 (7.5–14.0) 9 (5.5–11.5) 0.09 0.95 (0.88–1.01) DUDIT score 1 (0−3) 0 (0−2) 0.37 0.94 (0.82–1.08) Sexual Health Risks in 12 months Using sexual hookup applications 18 (85.7) 63 (82.9) 0.76 0.81 (0.21–3.15) Transactional Sex 0 8 (10.4) 0.99 NR Condomless sex 11 (50.0) 46 (59.7) 0.42 1.48 (0.57–3.84) More than 5 sexual partners 11 (47.8) 40 (52.6) 0.69 1.21 (0.47–3.08) History of sexually transmitted disease 4 (17.4) 21 (27.3) 0.34 1.78 (0.54–5.85) Non-consensual sex act 2 (9.1) 3 (3.9) 0.34 0.40 (0.06–2.59) Encounter sexual violence** 5 (21.7) 6 (7.8) 0.07 0.30 (0.08–1.11) Having sex with people with unknown HIV status 18 (78.3) 55 (71.4) 0.52 0.69 (0.23–2.10) Having sex with people with HIV 4 (17.4) 16 (21.6) 0.66 1.31 (0.39–4.4) SAU or SDU** 17 (73.9) 45 (58.4) 0.19 0.50 (0.17–1.4) - SAU** 16 (69.6) 34 (44.2) 0.04* 0.35 (0.13–0.94) - SDU 11 (47.8) 28 (36.4) 0.33 0.62 (0.24–1.60) - Combined SAU and SDU** 10 (43.5) 17 (22.1) 0.047* 0.37 (0.14–0.99) Total syndemics count (continuous) 1.38 (0.59) 1.44 (0.72) 0.67 1.08 (0.76–1.54) Categorial syndemic burden 1.00 (1−2) 2.00 (1−2) 0.71 1.14 (0.56–2.30)
Several variables met the criteria for inclusion in multivariable logistic regression analysis: AUDIT score, transactional sex, exposure to sexual violence, any SAU or SDU, any SAU, and combined SAU and SDU (Table 2). The multivariable logistic regression model demonstrated that exposure to sexual violence (aOR=0.24 [0.06, 0.97], p = 0.045) and SAU (aOR=0.28 [0.10–0.82], p = 0.02) were independently associated with PrEP retention (Table 3.1).Model Fit Measures Overall Model Test Deviance AIC R²McF χ² df p 98.8 107 0.08 9.10 3 0.03 Model Coefficients adjusted by age 95 % Confidence Interval Predictor Estimate SE Z p Odds ratio Lower Upper Intercept 1.71 1.08 1.59 0.11 5.51 0.67 45.30 Exposure of sexual violence 1 – 0 -1.44 0.72 -2.00 0.045 0.24 0.06 0.97 Sexualized alcohol use 1 – 0 -1.27 0.54 -2.32 0.02 0.28 0.09 0.82 age 0.02 0.03 0.43 0.67 1.02 0.94 1.09
4Theory-driven multivariable regression results
The results of the theory-driven multivariable logistic regression model (Table 3.2a) demonstrated that the association between SAU and PrEP retention remained robust in magnitude, even when adjusted for the continuous psychosocial syndemics count. Specifically, SAU was associated with reduced odds of retention (aOR=0.378, 95 % CI: 0.132, 1.08; p = 0.07). Although the association for SAU became marginally non-significant in this comprehensive model, the magnitude of the adjusted odds ratio remained comparable to the estimate found in the initial reduced model (aOR=0.28).Model Fit Measures Overall Model Test Deviance AIC R²McF χ² df p 94.5 102 0.04 3.86 3 0.28 95 % Confidence Interval Predictor Estimate SE Z p Odds ratio Lower Upper Intercept 0.81 1.11 0.73 0.47 2.24 0.25 19.84 Sexualized alcohol use 1 – 0 -0.97 0.54 -1.81 0.07 0.38 0.13 1.08 age 0.02 0.03 0.68 0.49 1.02 0.96 1.09 Syndemics Count 0.14 0.19 0.75 0.45 1.15 0.80 1.67
Furthermore, to test the cumulative effect of psychosocial factors , the psychosocial syndemics count (ranging 0–6) was included as a continuous variable in the model, but it was not significantly associated with PrEP retention (aOR=1.153, 95 % CI: 0.795, 1.67, p = 0.45).
We also modeled the cumulative risk using a categorical syndemics burden variable (Table 3.2b), which yielded similar findings. In this model, SAU retained a strong trend toward reduced retention (aOR=0.397, 95 % CI: 0.14, 1.16, p = 0.093). The categorical syndemics burden also showed no statistically significant association with PrEP retention when comparing participants with one condition (aOR=0.274, p = 0.26) or conditions (aOR=0.695, p = 0.76) against those with a burden score of zero. Consistent with the findings from the reduced model, age was not a significant predictor of PrEP retention in either theory-driven modelModel Fit Measures Overall Model Test Deviance AIC R²McF χ² df p 91.4 101 0.07 6.90 4 0.14 95 % Confidence Interval Predictor Estimate SE Z p Odds ratio Lower Upper Intercept 1.96 1.67 1.17 0.24 7.07 0.27 187.43 Sexualized alcohol use 1 – 0 -0.92 0.55 -1.68 0.09 0.40 0.14 1.16 age 0.02 0.04 0.44 0.66 1.02 0.95 1.09 Cetegorial Syndemics burnden 1 – 0 -1.29 1.16 -1.12 0.26 0.27 0.03 2.65 2 – 0 -0.36 1.17 -0.31 0.76 0.70 0.07 6.87
4.1Qualitative interviews
4.1.1SAU and SDU
The demographic data of the interview subsample are reported in Table 4. Participant were queried, in part, about their perspectives on potential mobile health interventions and the role technology may play in improving the health and PrEP retention of MSM with any SAU or SDU. They suggested that mobile health interventions could be beneficial for promoting PrEP use in their community, and explained that MSM in their community have extensive experience using social networking applications and are, therefore, already comfortable navigating mobile tools. More than a third (11 Ps) reported using a variety of geospatial networking applications to seek out sex, and a few (5 Ps) reported using these apps to acquire drugs for sexual purposes.Demographic data All SAU and/or SDU N (%), Mean (SD) SAU (N = 16) N (%), Mean (SD) Combining SAU with SDU (N = 14) N (%) OR Mean (SD) Age 29.3 (7.55) 26.9 (5.94) 32.1 (8.40) Regular income 20 (66.7) 10 (33.3) 10 (33.3) Educational lower than high school 3 (10) 0 (0) 3 (30.0) Marital status: single 22 (73.3) 12 (40.0) 10 (33.3) Living situation: alone 10 (33.3) 4 (13.3) 6 (20.0) Presence of the history of mental illness 2 (6.7) 1 (3.3) 1 (3.3) Presence of the history of physical illness 3 (10.0) 1 (3.3) 2 (6.7) PrEP retention in 3 months 23 (76.7) 12 (40.0) 11 (36.7)
Some interviewees highlighted the significant risk for HIV transmission for MSM who use alcohol and other drugs in the context of sexual activities, and the value of mobile health interventions that specifically target this high-risk group. Sharing a personal experience of sexual risk in the context of heavy drinking, one interviewee argued that mobile health interventions should first target those at highest risk, ‘It was an afterparty in an event. I blacked out after drinking lots of alcohol. I woke up and felt that my [anus] was hurt and bleeding. I can’t remember what happened and became frustrated. …. I was advised to take PEP for the incidence.’ (29-year-old participant)
Some participants discussed specific details of technology and valuable features that promote PrEP services and retention. Most agreed that mobile technologies were a necessary approach for educating and informing the MSM community and delivering important health interventions. However, very few (2) reported currently using social media to receive education or to specifically connect with PrEP providers, despite believing that technology should be better leveraged to this end. Several interviewees highlighted the challenges of being bombarded with excessive information through a range of media platforms, making it difficult for mobile interventions to compete. Stressing this point, one interviewee stated, ‘The LGBT population usually engages with so much social media such as Twitter, Instagram, and Facebook’ …. ‘people may not be interested in an intervention that provides massive information.’ Too much information, he argued, could lead to a cursory or ‘rough read’ of health information which, in turn, could ‘lead to misunderstandings (20-year-old participant)’.
Some noted the value of interactive and engaging application features that provided a reward structure for adherence to PrEP or other pro-health accomplishments, similar to health information apps developed for dieting, exercise and nutrition. Highlighting this, one interviewee noted, ‘I think of an application [similar] to one for intermittent fasting; it has questions for diet, and exercise. If we reach the goal, the app will give us the tokens that we can use for upgrading to premier users or other rewards. I think it could be applied to the application that you’re designing’ (38-year-old participant).
Interviewees also shared that any mobile intervention ought to be timely and relevant. For example, patients who struggle with medication adherence during periods of intoxication may need reminders about when to take PrEP, when clinic appointments are upcoming, or where they can access medications in cases of an emergency including when HIV post-exposure prophylaxis (PEP) may be needed. One participant stated, ‘Just the reminder of taking PrEP is enough for me. For example, if I forget to take PrEP, the (mobile) app should also provide the timeframe of taking PrEP that is still efficient to prevent infections. It should further provide a good period I should take PrEP in case that I wanna drink at night’ (25-year-old participant).
5Discussion
This study is, to our knowledge, the first in Thailand and Asia to use a quantitative approach with embedded qualitative methods to examine how SAU and SDU influence PrEP retention and to explore the role of mHealth interventions. Quantitative results showed that most participants had recent SAU and/or SDU, behaviors that elevate HIV transmission risk among MSM. SAU—but not SDU—was significantly associated with lower PrEP retention. The qualitative component provided essential context, revealing that intoxication reduced participants’ ability to adhere to PrEP, clarifying the mechanism through which SAU hinders retention. It also identified acceptable mHealth strategies, including reminders and pro-health incentives, to address this barrier. These findings highlight substantial risk in this population and offer context-specific insight into PrEP challenges for MSM in Northern Thailand. Further work is needed to understand retention issues and other stages of the PrEP care cascade, such as adherence.
Qualitative findings further explained the quantitative results: individuals engaging in SAU and SDU experienced reduced capacity to adhere to PrEP while intoxicated, compromising drug effectiveness and increasing vulnerability to HIV. Prior studies similarly show that substance use, stigma, and limited healthcare access impede PrEP continuation (Marcus et al., 2016), and uncertainty about PrEP–substance interactions also deters initiation and sustained use. Participants expressed strong interest in mHealth features such as reminders and incentives, which directly target adherence lapses during intoxication.
The finding that SAU, but not SDU, was significantly associated with lower PrEP retention offers unique, context-specific insights into barriers facing MSM in Northern Thailand. This diverges from much of the global PrEP literature (de Sousa et al., 2023, Viamonte et al., 2022), which has placed considerable focus on SDU and "chemsex," often highlighting the concern for PrEP adherence among those who use stimulants like methamphetamine. One possible explanation for the lack of association with SDU in our study may be linked to the specific substance profile of the sample: 51.6 % of SDU participants reported using alkyl nitrites, while only 9.7 % reported stimulants. Other research from Asia suggests that alkyl nitrite use alone may not warrant the classification of "chemsex," which is typically reserved for more problematic drugs (Tan et al., 2021a, Wang et al., 2023). The inclusion of these less consequential drugs in the SDU definition may have weakened the association with the primary endpoint of retention. Conversely, SAU may represent a more culturally central or accessible risk behavior in this Thai context (Dadras, 2024), leading to more frequent and sustained disruption of the PrEP routine compared to SDU.
The experience of sexual violence was previously reported as an associated factor of PrEP retention in women at risk of HIV. The explanation was related to intimate partner violence (IPV), including sexual violence, which could lead to remembering barriers and the need to get away from their place for their safety after any form of violence occurs. IPV is a powerful syndemic condition that contributes to worse health outcomes among MSM as well, and findings related to remembering barriers and safety might also explain the relationship between sexual violence and PrEP retention in our sample (Reddy et al., 2023, Roberts et al., 2016).
Although psychosocial syndemic counts in the SDU and combined SAU and SDU groups were higher than the group with no SAU or SDU, contrary to our hypothesis, we found no association between psychosocial syndemic counts and PrEP retention. This result requires critical discussion regarding methodological limitations and their impact on testing the syndemics framework. First, this finding may be influenced by reliance on bivariate analysis and the small sample size, which may limit the power to detect complex interactions. To address the issue of risk overestimation, we have conducted an additional Poisson regression model in Appendix 1. Second, we acknowledge the exclusion of structural syndemics (e.g., unstable housing), which, according to syndemic theory, often interact with psychosocial conditions to amplify negative health outcomes. The exclusion may limit the analysis's completeness. Third, retention was only measured at a single 3-month follow-up time point, which may not capture how the accumulation of syndemics worsens retention over a longer duration. Despite the null finding for the overall count, the total syndemics count was significantly higher in the SDU and combined SAU and SDU groups compared to PWoSU, confirming that the cumulative burden of psychosocial factors exists within this high-risk population, even if the count itself did not predict short-term retention. Additional research is needed to examine how the accumulation of psychosocial syndemics might worsen retention over time or affect other important aspects of the PrEP cascade, such as PrEP awareness or adherence (Blashill et al., 2020).
Participants suggested that useful information might relate to PrEP and PEP, but they also suggested that sharing new, relevant information would be useful. As a greater proportion of the SAU and SDU groups had clinically meaningful symptoms of a substance use disorder, they may also benefit from messages related to substance use treatment. Although participants suggested that an intervention be interactive and incentivize pro-health behaviors such as adherence, the nature of rewards was not specified. Considering concern for exposure to sexual violence in the setting of SAU or SDU, rewards could include earning free rides on rideshare apps so that people can get home safely.
This study has several limitations. First, a pilot study of a convenience sample of people engaging in PrEP care at a community clinic may not be representative of all those who are receiving or would benefit from PrEP. Also, rates and characteristics of MSM engaging in SAU and/or SDU may differ in other contexts. However, the clinic serves as a key entry point for PrEP services for MSM in the region, making it a practical and relevant site to study PrEP retention and related behaviors. It also offers a unique opportunity to collect real-world data from individuals actively engaged in PrEP care within this specific community, which remains valuable despite its limited generalizability. Secondly, since backward model selection is used in the multivariable logistic regression analysis. This variable selection process carries the inherent risk of residual confounding, as some relevant covariates that were marginally associated with PrEP non-retention in univariable analysis may have been dropped from the final reduced model. This limitation means the reported independent associations should be assessed cautiously, and we recommend readers consider the effects of covariates that were marginally associated with PrEP non-retention. However, we also present a theory-driven multivariable model that retains key covariates in Table 3.2a,b. Third, a significant limitation is the restricted definition of PrEP retention, defined solely as completing a routine 3-month PrEP follow-up. This short timeframe limits our understanding of long-term PrEP continuation and sustainability. Additionally, this study did not distinguish between people who were new PrEP starts, those who were stable on PrEP, or those who were restarting after a period of non-retention, which may differentially influence short-term retention rates. Furthermore, we were unable to determine the specific reasons why participants did not complete their 3-month follow-up visit. Finally, Finally, the qualitative component included a potential source of bias, as all 30 in-depth interviews were conducted by the study's Principal Investigator and lead author (AO). This position might influence participant disclosure in sensitive topics such as drug use, sexual violence, or adherence challenges. To mitigate the effect of this potential interviewer bias, the analysis process incorporated independent review and consensus coding by two authors (AO and SL).
6Conclusion
Notwithstanding these limitations, the present study is one of the few that utilized a quantitative approach with an embedded qualitative component to describe the relationship between SAU, SDU, and PrEP retention in Asia. Recent SAU and exposure to sexual violence were significantly associated with not completing a 3-month PrEP follow-up visit (no retention). Furthermore, our findings also highlight the potential to capitalize on a high rate of social media connectedness to design a mobile health intervention to enhance PrEP service delivery. Mobile health interventions may be most acceptable if they are informative and possess interactive features that provide a reward structure for achieving health behaviors.
Research materials or sponsorship
Study materials or equipment for facilitating the participants were provided by the Faculty of Medicine, Chiang Mai University,and MPLUS Clinic Chiang Mai. No other professional relationships (e.g., employment, board positions) that could influence the submitted work.
Non-financial interests
No personal or ideological affiliations likely to bias analysis or interpretation.
Other potential conflicts
- •No patent applications or intellectual property rights related to findings.
- •No trip support or travel grants from commercial sources relevant to the study period.
Transparency declaration
– “All authors confirm that all potential conflicts of interest — financial or non-financial — have been fully disclosed. No additional undeclared conflicts exist.”
Funding
This project was supported by the Fogarty International Center of the National Institutes of Health (10.13039/100000002NIH) under Award Number D43TW009343 and the University of California Global Health Institute (UCGHI). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or UCGHI. This project is partially support by Chiang Mai University, regarding resource utilization. Other authors have no conflicts of interests to declare. Funding entities had no role in study design; collection, analysis, or interpretation of data; writing of the manuscript; or the decision to publish.
Declaration of Competing Interest
The authors declare that they have no competing financial interests or personal relationships that could be perceived to have influenced the work reported in this paper.
Appendix 1
| Log likelihood ratio tests | |||
|---|---|---|---|
| X² | df | p | |
| Exposure of sexual violence | 1.00 | 1 | 0.32 |
| Sexualized alcohol use | 1.26 | 1 | 0.26 |
| Age | 0.02 | 1 | 0.89 |
| Log likelihood ratio tests | |||
|---|---|---|---|
| X² | df | p | |
| Age | 0.10 | 1 | 0.76 |
| Sexualized alcohol use | 0.78 | 1 | 0.38 |
| Syndemics Count | 0.13 | 1 | 0.71 |
Appendix 2The HIV transmission risk questions used to define risk behavior
| Have you had sex with a man who has HIV? |
| Have you had sex with a man whose HIV status you did not know? |
| Have you had gonorrhea, chlamydia, or syphilis in the last 12 months? |
| Have you had greater than 5 sex partners in the last 12 months? |
| Do you always use condoms during anal sex? |
| Have you ever traded sex for money or drugs? |
| Have you ever used social media or mobile application for casual sex? |
Appendix 3: Summary of the qualitative data analytic progression
| Analytic Frameworks | Themes | Categories | Codes | Illustrative Quotations |
|---|---|---|---|---|
| Association between SAU, SDU and PrEP retention | Sexualized substance use as a primary motivator for PrEP initiation | Using PrEP as a protective strategy | Sexualized substance use is one of the reasons of using PrEP | “It was an afterparty in an event. I blacked out after drinking lots of alcohol. I woke up and felt that my [anus] was hurt and bleeding. I can’t remember what happened and became frustrated. …. I was advised to take PEP for the incidence.” |
| Loss of control due to alcohol is a primary motivation for PrEP | “Alcohol is a reason I decided to take PrEP because I lost self-control when drinking, and decision-making capacity decreased” | |||
| PrEP increases feelings of comfort/relaxation during sex | “It makes us feel more relaxed and comfortable if we have sex” | |||
| Sexualized substance use as a major barrier to adherence | Disruption of Routine and Timing | SDU has effects on using PrEP (specifically timing/consistency) | “Drinking is a main cause (of inconsistency) due to wake-up time and the amount we drink. If we drink heavily, we might not sleep until noon” | |
| Heavy intoxication leads to forgetting the dose | “If heavily drunk, there is a tendency to not take the pill or forget it completely” | |||
| Disrupted sleep patterns or staying with clients cause doses to be missed/delayed | “Sometimes I leave the medicine in the room and don't carry it, and then I sleep with a client all night” | |||
| Late dosing upon remembering (mitigation strategy) | “Sometimes when I go out for some drinks, I forget to take the medicine, but when I remember, I take it immediately” | |||
| Perceived medical interaction of PrEP and sexualized substance use | Perceived interaction effects | Personal belief that alcohol does interact (due to blood/vitamins) | “PrEP might also have interactions (with alcohol) because it relates to our blood; this is a personal opinion” | |
| Denial the effects of substance use | Denial of the effects of substance use (Alkyl nitrite/inhaled drugs) on PrEP | “I think it is unrelated because they are used separately. Also, popper use involves inhalation, but PrEP is swallowed, so they should not be related” | ||
| PrEP as a risk compensation of sexualized substance use | Risk compensation | PrEP provides an additional safety layer/shield | “PrEP is like a shield that protects and takes care of oneself” | |
| Taking PrEP leads to reduced condom use frequency | “Receiving PrEP makes (me) feel safe, which may lead to not wearing a condom often” | |||
| Acceptability of mobile health interventions aimed at improving the PrEP continuum | Acceptance with concerning points to designs mobile health intervention | Acceptance with suggesting probable features in the app | Positive view about App functionality (Reminder/Tracking/Appointments) | “I think of an application [similar] to one for intermittent fasting; it has questions for diet, and exercise. If we reach the goal, the app will give us the tokens that we can use for upgrading to premier users or other rewards. I think it could be applied to the application that you’re designing” |
| Concerning points | Fear of external individuals obtaining data | “If external individuals can take out the data, it might cause harm to the person who provided the information” | ||
| Being bombarded with excessive information | “The LGBT population usually engages with so much social media such as Twitter, Instagram, and Facebook” …. “people may not be interested in an intervention that provides massive information.” | |||
| Existing phone reminders suffice | “Just the reminder of taking PrEP is enough for me. For example, if I forget to take PrEP, the (mobile) app should also provide the timeframe of taking PrEP that is still efficient to prevent infections. It should further provide a good period I should take PrEP in case that I wanna drink at night” |
Acknowledgments
This project was supported by the Fogarty International Center of the National Institutes of Health (NIH) under Award Number D43TW009343 and the University of California Global Health Institute (UCGHI). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or UCGHI. This project is partially supported by Chiang Mai University, regarding resource utilization.
We would like to acknowledge the MPLUS Clinic Chiang Mai team, which facilitates and collects data from a specific population to enhance our understanding of this important key behavior related to HIV.