Factors associated with non-use of HIV pre-exposure prophylaxis among people who inject drugs aware of this prevention method in Mozambique, 2023-2024

¹Field Epidemiology Training Program (FETP), National Institute of Health (INS), Maputo, Mozambique, ²National Institute of Health (INS), Maputo, Mozambique, ³Field Epidemiology Training Program (FETP), Mozambique

^&Corresponding author
Nazario Carlitos, Field Epidemiology Training Program (FETP), National Institute of Health (INS), Maputo, Mozambique

Introduction    

People who inject drugs (PWID) experience a disproportionate burden of HIV globally. An estimated 14.8 million individuals aged 15-64 injected drugs in 2022, representing a major public health concern [1,2]. Since 2008, sub-Saharan Africa has observed a notable increase in injecting drug use, contributing to a growing regional epidemiological challenge [1,2]. In this region, PWID face overlapping risks, including HIV transmission through non-sterile syringe sharing and unprotected sexual practices, as well as structural vulnerabilities such as stigma, discrimination, violence and limited access to health services [1,3-5].

Despite global progress in HIV prevention, PWID remains disproportionately affected. Between 2010 and 2022, new HIV infections among PWID decreased by 24%, however, in 2022 their relative risk of infection was still 14 times higher than that of the general adult population, exceeding the risk among sex workers and second only to men who have sex with men (23 times higher) [6]. Global estimates for 2024 indicate a median HIV prevalence of 7.1% among PWID ten times higher than the global adult prevalence of 0.7% [7].

In Mozambique, HIV prevalence in the general population was 12.5% in 2021 [8]. Data from the second round of the biobehavioral survey among PWID (BBS-PWID-II) in 2023 confirmed consistently higher prevalence in this group, ranging from 13.3% in Nampula to 29.9% in Maputo, despite a notable decline since 2014 [5,8-10]. Notably, fewer than half of HIV-positive PWID are linked to care and treatment, underscoring persistent structural barriers [5].

Pre-exposure prophylaxis (PrEP) is a highly effective, proven biomedical intervention that reduces new HIV transmissions [11-13]. It involves the daily use of antiretroviral drugs by HIV-negative individuals who are at a substantial risk of infection. Scientific evidence shows that when taken consistently, PrEP can reduce the risk of HIV transmission by 70% to 90%, especially when combined with other preventive measures like condom use and harm reduction [13,14].

In Mozambique, PrEP was formally integrated into HIV prevention services in 2020, with support from the Ministry of Health (MISAU) and the U.S. President's Emergency Plan for AIDS Relief (PEPFAR) [15]. However, PrEP uptake among PWID remains limited, largely due to scarcity of harm-reduction programs, low awareness, and high levels of stigma [5].

Evidence concerning determinants of PrEP uptake among PWID in Mozambique is scarce. Understanding these determinants is essential to design targeted and equitable HIV prevention strategies. This study aimed to identify factors associated with PrEP never used among PWID already aware of this strategy in five major Mozambican cities, hypothesising that structural and vulnerability-related factors play a more prominent role than individual sociodemographic characteristics.

Methods     

Study design and settings: we conducted a secondary analysis of data from the BBS-PWID-II, implemented from 2023 to 2024 in Mozambique. The main survey study used an observational, quantitative, analytic cross-sectional design. This secondary analysis specifically examined factors associated with never having used PrEP among participants who reported prior awareness of PrEP as an HIV prevention strategy.

Study population and sampling: the survey was conducted in five major urban centres: Maputo, Beira, Tete, Quelimane, and Nampula. Eligibility criteria included: age ≥16 years, self-reported drug injection within the previous 12 months, and provision of written informed consent. Participants were excluded if visibly unable to consent due to intoxication, provided inconsistent screening responses, or refused participation. Respondent-driven sampling (RDS) was used to recruit participants due to the hidden and highly stigmatized nature of PWID. Initial “seed” participants recruited members of their peer networks using coded referral coupons, generating recruitment chains that varied in length across cities. Additional RDS implementation details, including number of seeds per city, coupon distribution procedures, incentive structure and recruitment chain diagnostics, are fully documented [10,16-18]. For this secondary analysis, the analytic sample was restricted to participants who reported awareness of PrEP at the time of the interview, in order to specifically examine factors associated with non-use among individuals with prior knowledge of this prevention method. Awareness was defined as a positive response ("yes") to the survey question: "have you ever heard of PrEP?”

Data collection procedures: trained interviewers administered a standardized, structured questionnaire adapted from previous biobehavioral surveys conducted among key populations in Mozambique, including the BBS-PWID 2014, and based on WHO recommendations for second-generation HIV surveillance [9,19]. The instrument collected information on sociodemographic characteristics, sexual behaviours, drug-use patterns, PrEP knowledge and uptake, access to health services, violence, stigma and discrimination. Participants were additionally offered rapid biological testing for HIV, hepatitis B virus (HBsAg, Abbott Diagnostics, Japan) and hepatitis C virus (anti-HCV, SD Bioline HCV, Standard Diagnostics, Republic of Korea), conducted on-site by trained health personnel. HIV testing followed the Mozambican national serial algorithm using Determine™ HIV-1/2 as the screening test and Uni-Gold™ HIV as the confirmatory assay. All pre- and post-test counselling was conducted confidentially, and results were recorded anonymously using only each participant's alphanumeric study identifier. Participants with reactive results for HIV and/or hepatitis were linked to care and treatment services by trained peer navigators and community focal points. Individuals could decline any testing component without affecting their participation.

Measures and variables: the primary outcome was “PrEP never used”, defined as having never used PrEP at any point in life among participants with prior awareness of this strategy. Independent variables were organized into domains: sociodemographic characteristics (age group, sex, marital status, education level, and city of residence); sexual and prevention behaviours (number of sexual partners in the previous 6 months, condom use at last sexual intercourse, attendance in community at HIV educational activities in the previous 12 months, and time since last HIV test); drug-use characteristics (substance most frequently injected); vulnerability and social risk indicators (self-reported overdose in the past 12 months, drug dependence treatment, experiences of aggression, sexual violence and discrimination in the past 12 months); and perceived HIV risk (self-reported, used as a proxy for risk awareness). Sparse categories with fewer than five observations were collapsed to ensure analytical stability.

Bias: potential sources of bias include recall bias and social desirability bias due to self-reported behavioural data. Selection bias may also be present due to the use of respondent-driven sampling. Efforts to minimize bias included the use of standardized data collection tools, trained interviewers, and confidential interview procedures.

Study size: the study size was determined by the number of participants recruited in the BBS-PWID-II survey. For this secondary analysis, all participants who reported awareness of PrEP were included (n=427), ensuring adequate statistical power to detect associations between explanatory variables and PrEP non-use.

Quantitative variables: continuous variables such as age were categorized into meaningful groups (16-24, 25-34, and ≥35 years) based on distribution and analytical relevance. Time since last HIV test was categorized as ≤3 months and >3 months to reflect programmatic thresholds. Other categorical variables were grouped to ensure sufficient sample size within categories.

Statistical analysis: data cleaning procedures included verification of coding accuracy, internal consistency, and identification of outliers. Variables with more than three categories and low frequencies (<5 observations) were recategorized to improve statistical robustness. Descriptive analyses summarized sociodemographic, behavioral, and health service-related characteristics. Missing values were retained as a distinct "no information" category in descriptive tables (Table 1, Table 2) to accurately report the completeness of the sample and allow for an assessment of its integrity. Statistical analysis was performed using R version 4.4.1. Univariable and multivariable logistic regression analyses were conducted to identify factors associated with PrEP never-use. Variables were initially selected for inclusion in the multivariable model based on a p-value <0.20 in univariable analysis. In addition, variables considered conceptually relevant based on prior literature and epidemiological plausibility (e.g., age and sex) were retained in the model regardless of statistical significance. Adjusted odds ratios (aOR) with 95% confidence intervals (CI) and p-values were reported, and statistical significance was set at P<0.05. For the multivariable regression, a complete case analysis approach was used, excluding observations with missing data on the selected covariates. Although RDS generates weights based on network size, we did not apply weighting. Previous methodological assessments show that when the objective is to assess associations within the sample rather than population prevalence, unweighted regression yields valid estimates [20-22].

Ethical considerations: the survey protocol was reviewed and approved by the INS Institutional Bioethics Committee (CIBS-INS) and the National Bioethics Committee for Health (CNBS) under number 877/CNBS/22, as an activity involving human subjects. PWID participated in the survey after providing written informed consent. The protection of participants' identities was ensured by the INS, which was responsible for data collection and management, through the use of anonymous identification codes in accordance with the ethical principles of confidentiality, privacy, and respect for individual autonomy. The ethical principles of integrity and data confidentiality were respected. We used only the data necessary to achieve the defined objectives.

Results     

Sociodemographic characteristics among participants aware of PrEP: a total of 427 participants who reported awareness of PrEP were included in this analysis, of whom 74% (314/427) had never used it. Most participants were aged 16-24 years (62%, 264/427), male (69%, 296/427), single (81%, 348/427), and had secondary or higher education (85%, 368/427) (Table 1).

Pre-exposure prophylaxis (PrEP) never-use varied significantly by city of residence. The lowest proportion of non-use was observed in Nampula (60%, 105/175), while higher proportions were documented in Quelimane (80%, 117/147), Maputo (85%, 39/46), Tete (89%, 31/35) and Beira (92%, 22/24) (Table 1).

Sexual behaviours and access to HIV prevention services: participants who had attended HIV educational sessions in the previous 12 months reported a lower prevalence of PrEP non-use (69%, 132/192) compared with those who had not (77%, 182/235; P=0.04). PrEP never-use was also lower among participants who were tested for HIV within the previous three months (66%, 139/212) than among those tested earlier (81%, 175/215; P<0.001). No statistically significant associations were observed for number of sexual partners or condom use at last sexual intercourse (P=0.44 and P=0.16, respectively).

Participants reporting overdose in the past 12 months (55%, 28/51; P=0.001) and those reporting sexual violence (46%, 11/24; P=0.002) also showed significantly different proportions of PrEP never-use compared with peers without these experiences. No statistically significant differences were observed for drug dependence treatment, aggression or discrimination (P=0.20, P=0.80, and P=0.29, respectively) (Table 2).

Factors associated with PrEP non-use: in the multivariable logistic regression model, city of residence and vulnerability indicators remained independently associated with PrEP non-use. Compared with Nampula, residence in Quelimane (aOR 2.34; 95% CI: 1.36-4.11; P=0.002), Maputo (aOR 3.48; 95% CI: 1.47-9.34; P=0.007), Tete (aOR 4.59; 95% CI: 1.56-17.2; P=0.011), and Beira (aOR 6.04; 95% CI: 1.62-39.5; P=0.02) was associated with higher odds of never having used PrEP. Participants whose last HIV test was more than three months ago also had higher odds of never having used PrEP (aOR 1.71; 95% CI: 1.02-2.89; P=0.04). Conversely, reporting overdose in the past 12 months (aOR 0.38; 95% CI: 0.19-0.77; P=0.007) and sexual violence (aOR 0.31; 95% CI: 0.12-0.79; P=0.02) was associated with lower odds of PrEP never-use (Table 3).

Discussion     

This study aimed to assess factors associated with PrEP non-use among people who inject drugs aware of this prevention method in Mozambique. A high proportion of participants reported never using PrEP, indicating substantial gaps in uptake among this population. Geographic disparities were observed, with higher odds of non-use in several cities compared to Nampula. In addition, longer time since last HIV testing was associated with higher odds of non-use, while recent overdose and experiences of sexual violence were associated with lower odds of non-use.

Sociodemographic factors such as age, sex, and education were not independently associated with non-use of PrEP, suggesting that structural and health system determinants better explain PrEP uptake, consistent with evidence from other settings [1,23,24].

Geographic disparities were evident. Multivariate analysis indicated that PWID in Quelimane, Maputo, Tete, and Beira were more likely to report never-use. These inequities may reflect variability in service availability, outreach coverage, stigma environments, and network-driven demand creation [1,2]. Global assessments similarly report uneven PrEP roll-out for PWID, often concentrated in urban areas supported by external funding [25].

Behavioural indicators displayed limited association with PrEP uptake. More than half of PWID reported having two or more sexual partners in the 12 months prior to the survey, but this group showed only a modestly lower uptake of PrEP. Similarly, although most participants did not use condoms at their last sexual encounter, PrEP uptake did not differ significantly between condom users and non-users. These findings suggest that isolated sexual behaviors are not the primary determinants of PrEP engagement. Rather, they align with studies emphasizing that structural, social, and contextual factors are more decisive for adherence than individual sexual practices [11,23,25,26].

Although the proportion of PrEP never-use was above 50% in both groups, there was an eight-percentage-point difference in never-use among PWID who participated in HIV lectures versus those who did not, suggesting that this intervention may contribute to improved engagement with prophylaxis. This variable showed a significant association in the crude model but lost significance in the adjusted model. Studies show that educational activities are essential components of effective programs, as they can increase knowledge about the intervention, reduce stigma, and improve adherence and use [11,26,27].

Pre-exposure prophylaxis (PrEP) uptake was markedly higher among PWID tested for HIV in the past three months compared with those tested earlier. In adjusted model, testing more than three months ago was associated with higher odds of non-use, reinforcing the role of HIV testing services as gateways to prevention. This observation is consistent with literature that highlights HIV testing services as opportune platforms for offering PrEP [11,23,25,28].

The vast majority of PWID reported using heroin as their primary injectable drug, while only a small proportion frequently injected other drugs, such as cocaine and methamphetamine. In both groups, the proportion of PrEP never-users exceeded 70%, suggesting that uptake of prophylaxis is limited regardless of the drug injected. These findings highlight the need for strategies that integrate PrEP with harm reduction services, such as needle exchange programs and methadone treatment, favoring community-based distribution models, as evidenced by previous studies [1,24].

Episodes of overdose and experiences of sexual violence, traditionally seen as indicators of extreme vulnerability, were associated with lower odds of never using PrEP. These findings should be interpreted with caution, as the cross-sectional design does not allow causal inference and the underlying mechanisms remain unclear. The observed associations may reflect increased contact with health services or heightened risk perception among affected individuals, potentially creating opportunities for engagement with prevention services. Interventions leveraging peer navigation, crisis-care touchpoints, and trauma-informed approaches may therefore be critical for this subgroup. Studies emphasize that recognizing one's own vulnerability is a relevant factor for PrEP acceptance [1,4,23,26].

This analysis has several limitations. First, given the cross-sectional design, the findings should be interpreted as associations rather than causal relationships. Second, restricting the analysis to participants aware of PrEP may introduce selection bias, as this subgroup may differ systematically from those unaware of the intervention, thereby limiting generalizability to all people who inject drugs. Third, the variable assessing reasons for not using PrEP had substantial missingness data and was excluded, reducing insight into motivational barriers. Fourth, behavioural and violence-related variables were self-reported and may be subject to recall and social desirability bias. Finally, although RDS facilitates recruitment of hidden populations, its network-based structure may introduce selection bias. In addition, while unweighted regression may limit how well the results apply to the wider population, it is appropriate for examining associations within the study sample, in line with the objectives of this study. This study has strengths, including the use of recent data from biobehavioral surveys in several cities in Mozambique focusing on PWID, a key population often underrepresented in HIV prevention research. These results contribute to the evidence base needed to inform more targeted HIV prevention strategies in this context.

Conclusion     

Non-use of PrEP among PWID with prior awareness in Mozambique remains high. This study provides evidence on factors associated with never-use of PrEP in this population. City of residence and timing of HIV testing were associated with higher odds of never-use, while a history of overdose and experiences of sexual violence were associated with lower odds. Participation in HIV lectures showed an association in crude analysis but not after adjustment. These findings highlight geographic disparities and the role of contact with HIV services in PrEP uptake, suggesting the need for targeted strategies to improve access and engagement with PrEP among PWID.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Conception and study design, data analysis and interpretation: Nazario Carlitos, Auria Ribeiro Banze, and Cynthia Semá Baltazar; data extraction, manuscript drafting, and guarantor of the study: Nazario Carlitos; manuscript revision: Auria Ribeiro Banze, Erika Valeska Rossetto, and Cynthia Semá Baltazar. All the authors read and approved the final version of this manuscript.

Acknowledgments     

We express our sincere gratitude to the National Institute of Health of Mozambique, the Faculty of Medicine of Eduardo Mondlane University, and the coordination of the Field Epidemiology Training Program (FETP), Mozambique. We also extend our deep gratitude to Roberta Horth (CDC Mozambique), Rachid Muleia (National Institute of Health), and Fabião Maússe (FETP Mozambique) for their support.

Tables     

Table 1: sociodemographic characteristics of people who inject drugs aware of pre-exposure prophylaxis and proportion who never used it, Mozambique, 2023-2024

Table 2: risk behaviors and access to health services among people who inject drugs aware of pre-exposure prophylaxis and proportion who never used it, Mozambique, 2023-2024

Table 3: multivariate logistic regression analysis of factors associated with pre-exposure prophylaxis never used among people who inject drugs in Mozambique, 2023-2024

References