Background: In British Columbia (BC), people with HIV are disproportionately affected by the drug toxicity crisis. Ritonavir, a potent cytochrome 3A4 inhibitor, boosts other protease inhibitors and slows metabolism of several synthetic opioids, notably fentanyl, raising its blood levels. We hypothesized that, among people with HIV who use(d) drugs (PWUD), a) exposure to ritonavir increases the odds of fatal opioid overdose (FOO) versus non-exposure; and b) exposure to cobicistat, a structural analogue of ritonavir, has a similar impact.
Methods: We included all deaths between 01-Jan-2012 and 31-Dec-2024 among PWUD with HIV in the BC Centre for Excellence in HIV/AIDS’s Drug Treatment Program (DTP). Individuals were considered PWUD if injection drug use was recorded on DTP clinical forms and/or a urine drug screen was ever performed. FOOs were identified via ICD-10 codes indicating opioid poisoning. Chi-squared, Fisher’s exact, Wilcoxon’s Rank Sum tests, and logistic regression were used to determine associations between a) exposure to ritonavir, and b) exposure to ritonavir and/or cobicistat at time of death and FOO compared to non-FOO deaths.
Results: Of 1,320 deaths among PWUD, 303 (23%) were FOOs. Ritonavir exposure at time of death occurred among 66 (22%) of FOO and 196 (20%) of non-FOO deaths (p=0.44); for cobicistat numbers were 33 (11%) and 86 (9.0%, p=0.24). Ritonavir exposure at time of death was not significantly associated with FOO (adjusted odds ratio [aOR] 1.19, 95% CI 0.85-1.66), adjusted for sex, age, homelessness, residence in Vancouver’s Downtown Eastside, and era of ART initiation. Results were similar when considering exposure to cobicistat and/or ritonavir (aOR 1.21, 95% CI 0.90-1.63).
Conclusion: These preliminary results indicate no association between exposure to ritonavir or cobicistat and likelihood of a fatal opioid overdose death in a large cohort of PWUD with HIV. We hope to corroborate these findings through ongoing time-to-event analyses.

Context: Adherence to antiretrovirals (ARV) is essential for people with HIV, but many face barriers. We developed the 7-item I(interference)-Score, a new electronic patient-reported outcome measure (ePROM) of ART adherence barriers to be used in HIV care. We explored its effectiveness in a Montreal clinic.
Methods: We conducted this 6-month one-arm implementation pilot study at the McGill University Health Centre, Montreal, among adults with HIV on ARV willing to use a smartphone to complete the I-Score. Patients visited their physician at Time 1(T1), 3 months(T2), and 6 months(T3). They completed the I-Score before medical visits. Patients completed a survey on socio-demographics (T1) and self-reported adherence, from 1(‘very poor’) to 7(‘excellent); physicians completed a checklist on actions taken based on I-Score results; and viral loads (VL) were collected from medical records (T1 and T3). We provide descriptive statistics.
Results: 28 patients participated in a total of 78 visits. Average age was 48 years, 12/32(38%) were women, and 19/32(59%) were migrants. Mean self-reported adherence was stable, 5(‘very good’) from T1 to T3, and so was the number of patients with an undetectable VL: 24/28(86%) at T1 and 24/27(89%) at T3. Most patients (26/28;93%) identified at least one adherence barrier, leading physicians to: provide 25/28 patients (89.3%) with education; order new tests for 12/28(43%) patients (e.g., liver function; tied mainly to Habits & Activities barriers, which include substance use); switch ARV for 3/28(11%) patients (Medication barriers); prescribe a new medication for 8/28(29%) patients (e.g., medication for attention deficit disorder; Thoughts & Feelings barriers); and refer 17/28(61%) patients to specialists (e.g., social worker; Social Situation barriers).
Conclusions: A vast range of issues were detected with a single tool, facilitating diverse clinical actions. The I-Score may improve patient-provider communication and guide physician efforts to support optimal adherence.

Heterosexual-identified men who have sex with men (H-MSM) experience sexual identity and behavior discordance yet it is unknown how H-MSM compare to concordant heterosexual men as well as gay, bisexual, and queer (GBQ+) men regarding PrEP knowledge and uptake. We surveyed adult cisgender men in Canada, the United States, and the United Kingdom to gain greater insight on their PrEP knowledge and uptake.

Methods
Purposive sampling was utilized to recruit men via online and offline venues. Data collection consisted of an online questionnaire for adult cisgender men, available in English, French, and Spanish. Data were collected between November 28, 2024 and December 31, 2025.

Results
The survey sample consists of 297 GBQ+ men; 208 concordant heterosexual men; and 92 H-MSM. A one-way ANOVA showed significant differences between the three groups on PrEP knowledge and uptake. H-MSM had significantly less PrEP knowledge (M = 3.48, SD = 3.87) than GBQ+ MSM (M = 6.35, SD = 4.05), p < .001. Among H-MSM, 15.8% of participants reported using PrEP. The logistic regression model predicting PrEP uptake had a good fit to the data. The Hosmer and Lemeshow chi-square goodness-of-fit test showed χ2(8) = 4.017, p = .856. The -2 Log likelihood of the model was 105.631. The model explained 56.9% of the variance in PrEP uptake (Nagelkerke R2). Overall, the model correctly predicted 88.0% of participants who use PrEP. Factors associated with greater PrEP uptake included motivation to use Grindr to find a date, motivation to use Grindr to find people to drink or use drugs with, sexual identity commitment, and PrEP knowledge.

Conclusions
This examination aims to improve understanding of H-MSM as a distinct population. This study provides key insights into avenues for HIV prevention with men whose sexuality and sexual behaviours may be outside of commonly accepted norms.

Background:
Suboptimal antiretroviral therapy (ART) adherence remains a major challenge in HIV care. Mobile health tools such as the MARVIN chatbot can support self-management and generate large volumes of patient messages with untapped potential for adherence monitoring. We developed large language model (LLM)–based classifiers to identify adherence barriers and stratify risk from patient-generated text.

Methods:
Training datasets included MARVIN training corpora, MARVIN-user conversations, qualitative interviews from the I-Score development study (a 7-item PROM of ART adherence barriers), LLM-generated synthetic examples, and external validation data from Portail VIH/SIDA du Quebec and online HIV community forums. Following a participatory design approach, 3 people with HIV, 3 clinicians, 1 resident, and 1 engineer refined annotation guidelines across three workshops. Two sentence-level tasks were defined: I) barrier classification across seven I-Score domains (Thoughts/Feelings, Habits/Activities, Social Situation, Economic Situation, Medication, Care, and Health) plus None; and II) four-level adherence risk classification (High, Medium, Low, None). Two annotators labeled 10% of the data, achieving inter-annotator agreement (κ=0.83 for barriers; 0.71 for risks). The full corpus was then annotated. Small-scale LLMs (e.g., Flan-T5) were fine-tuned for multi-class classification and compared with prompt-tuned large-scale LLMs (GPT-4.1, Gemma-3, DeepSeek-R1, and LLaMA-3.2) on test and external validation datasets using mean Macro-F1.

Results:
The curated dataset included 15,480 annotated sentences. On external validation, Flan-T5-xl achieved the highest performance for barrier detection (Macro-F1 = 0.71, 95% CI 0.68-0.74), outperforming GPT-4.1 (0.62) and other large-scale LLMs (0.45-0.59; P<0.001). For risk stratification, Flan-T5-large performed best (Macro-F1 = 0.57, 95% CI 0.53-0.60), again exceeding GPT-4.1 (0.43) and other LLMs (0.28-0.47; P<0.001). Similar patterns were observed on held-out test data.

Conclusion:
LLM-based triage models can effectively identify ART adherence barriers and risk from patient messages, offering a scalable approach for individualized HIV care. Integration into MARVIN is underway, with real-world validation planned to assess clinical utility.

Long-acting injectables (LAI) are safe, effective forms of PrEP whose discreet administration may help overcome stigma associated with oral PrEP’s visible pill-taking. Among sexually active GBM in the PrEP Implementation Project (PRIMP), we examined relationships between PrEP-related stigma, LAI-PrEP acceptability, and PrEP “reachability”.

PRIMP recruited sexually-active adult GBM in Ontario and British Columbia to a cross-sectional survey in 2022. We restricted our analysis to HIV-negative participants with prior PrEP knowledge. Participants indicated their PrEP-use status (current/former/never) and acceptability of LAI-PrEP (yes/unsure/no). PrEP-related stigma was measured with the 12-item HIV PrEP Stigma Scale (HPSS; Cronbach’s α=0.81). We tested the association between stigma and acceptability stratified by PrEP-use status using multinomial logistic regression (yes vs no, unsure vs no), adjusting for potential confounders. We also estimated the proportions of PrEP-indicated participants potentially “reachable” by PrEP modalities (defined as use/interest in oral or LAI-PrEP) and used multivariable logistic regression with backward selection to examine characteristics associated with reachability by current/emerging PrEP products.

Of 970 included participants, stigma was highest in never users, lowest in current users, and not significantly associated with LAI-PrEP acceptability or uncertainty (Table). Of 752 PrEP-indicated participants, 65.7% were using oral PrEP and 28.4% were non-PrEP users interested in current/emerging PrEP products, leaving 5.9% potentially unreachable. After backward selection, greater stigma was negatively associated with reachability (OR=0.18, 95%CI=0.09-0.36).

PrEP stigma was not associated with willingness to try LAI PrEP across PrEP-use status but was associated with remaining unreachable by existing formulations. Reducing PrEP stigma is critical to increasing PrEP coverage.

Introduction: Sexually transmitted and blood-borne infections (STBBIs) are increasing in Canada. Barriers to STBBI services delay early diagnosis, treatment and access to prevention strategies. The STI Care Now initiative is designed to expand access to STBBI testing, prevention, and treatment services in Nova Scotia. The initiative includes HIV, chlamydia/gonorrhea (CT/NG) self-testing, CT/NG treatment, and HIV pre-exposure prophylaxis (PrEP) services.

Methods: Individuals self-referred through an online webform. HIV self-tests and CT/NG self-testing kits with prepaid return postage were mailed to participants or picked up at community-based organizations and high schools. Pharmacists conducted HIV PrEP assessments and prescribed virtually. CT/NG results were provided by email or phone, and individuals with a CT/NG diagnosis were assessed virtually by a pharmacist and treated or referred when needed. Initiative utilization metrics were collected, and patient experience feedback was obtained through an electronic questionnaire.

Results: Between July 8, 2024 and January 7, 2026, the initiative received 6944 self-referrals from 5304 individuals. 7180 testing kits were issued, including 2128 HIV self-tests. 35% of individuals self-reported their HIV self-test results, with 601 non-reactive, 144 invalid, and 0 reactive results. For CT/NG self-testing, the kit return rate was 55% (3943/7180). Among returned tests, 219 were positive for CT and 40 for NG, yielding a 6.6% positivity rate. All individuals with positive results were successfully linked to care. 298 individuals requested HIV PrEP. 338 PrEP assessments were completed (including 191 eligibility, 64 initial prescription, and 83 refill assessments), and 134 prescriptions were written (99% emtricitabine/tenofovir disoproxil fumarate, 1% emtricitabine/tenofovir alafenamide). The questionnaire had a 55% response rate (2136 responses) and indicated high satisfaction, including ease of use and convenience.

Conclusion: STI Care Now successfully facilitated access to HIV testing, HIV PrEP, chlamydia and gonorrhea testing and care, and has potential for future use with other STBBIs.

ntroduction Increases in bacterial sexually transmitted infections (bac-STIs), such as gonorrhoea, chlamydia and syphilis, have affected gay, bisexual, and other men who have sex with men (gbMSM). A new strategy to prevent bac-STIs involves giving doxycycline as postexposure prophylaxis (doxy-PEP), and research shows that this intervention can prevent chlamydia and syphilis by 70% and gonorrhoea by 33%–50%. Consequently, the US CDC released guidelines in 2024 recommending doxy-PEP for gbMSM with ≥1 bacterial STI diagnosis in the previous 12 months.

Methods We reviewed public health STI surveillance data between 1 January 2021 and 31 December 2024 for gbMSM in Ottawa, Canada, and analysed the number of infections and episodes of infections per person. We estimated the number of individuals who would need doxy-PEP to prevent a single bac-STI episode.

Results During the study period, 1819 unique gbMSM experienced 2834 positive bac-STI testing episodes (PTE), during which 3114 bac-STIs were diagnosed. Consistently, three-quarters of gbMSM did not have a subsequent infection, whether they were diagnosed with 1 infection, 2 infections or ≥3. Considering the average effectiveness of doxy-PEP in this study population, the average number needed to treat (NNT) if doxy-PEP were prescribed to all gbMSM to prevent a first PTE would have been 60. The NNT among those with their first PTE to prevent a second PTE was 8; among those with their second, the NNT was 7.

Conclusions Based on these data, and in alignment with the CDC guidelines, we conclude that doxy-PEP would likely have the most balanced population-level bac-STI prevention effect if given to gbMSM with ≥1 bac-STI diagnosis within the preceding 12 months. Providing doxy-PEP to all gbMSM would likely result in an overuse of antibiotics, and providing doxy-PEP only after a second PTE would result in fewer infections averted for the same proportion treated.

People who use drugs (PWUD) remain disproportionately affected by HIV in Canada. Despite its proven effectiveness, HIV pre-exposure prophylaxis (PrEP) uptake among PWUD is minimal. We quantified barriers to PrEP use among PWUD engaged in care in Vancouver’s Downtown Eastside (DTES).

PWUD age 19 and older from 6 addictions and primary care clinics in DTES completed questionnaires about their willingness to use PrEP (yes, maybe/don’t know, no) and potential barriers to using PrEP. We computed the proportion of participants reporting each barrier, and adjusted risk ratios (aRRs) and ‘population attributable fractions’ of being unwilling to use PrEP (vs. willing/uncertain) for those reporting each barrier using multivariable modified Poisson regression. We repeated this analysis, restricting to participants who reported willingness or uncertainty about taking PrEP to estimate contribution of each barrier to uncertainty.

288 participants most-commonly cited low risk perception (56%), lack of knowledge (45%), and concern about side effects (36%) as barriers to taking PrEP (Table). Low risk perception was significantly associated with unwillingness to take PrEP (aRR 3.51 [2.16-5.84]), accounting for 58% (95% CI 39%-72%) of unwillingness in the sample. Among participants who were willing or unsure about PrEP, low knowledge was the only barrier significantly associated with uncertainty (aRR 1.88 [1.11-3.23]), with 33% (9%-53%) of uncertainty in the population attributable to this barrier.

Even among participants with indications of HIV risk, low risk perception represented a significant barrier to PrEP uptake among PWUD in DTES. Educational campaigns may prove most effective for engaging those uncertain about PrEP.