Prevalence and risk factors associated with HIV-1 infection among people who inject drugs in Dar es Salaam, Tanzania: a sign of successful intervention?

Background Prevalence of HIV infection among people who inject drugs (PWID) has been reported to be higher than that of the general population. The present study aimed to estimate the prevalence of HIV infection and associated risk factors among PWID in Dar es Salaam, Tanzania, following the introduction of a comprehensive HIV intervention package (CHIP) for PWID in the country in 2014. Methods We conducted an integrated bio-behavioral survey (IBBS) among PWID using respondent-driven sampling (RDS) in Dar es Salaam, Tanzania, between October and December 2017. Data on socio-demographic characteristics and risky behaviors were collected through face-to-face interviews. Blood samples were collected and tested for HIV infection. We accounted for weighting in the analyses, and logistic regression was performed to assess risk factors for HIV infection. Results A total of 611 PWID (94.4% males and 5.6% females) with a median age of 34 years (IQR 29–38) were recruited. The overall prevalence of HIV infection was 8.7% (95% CI 6.5–10.9). The prevalence of HIV infections for males and females were 6.8% (95% CI 4.7–8.9%) and 41.2% (95% CI 23.7–58.6%) respectively. Adjusted weighted logistic regression analysis (WLRA) showed that being a female (aOR 19.1; 95% CI 5.9–61.8), injecting drugs for more than 10 years (aOR = 7.32; 95% CI 2.1–25.5) compared to 1 year or less and being 45 years or older (aOR = 34.22; 95% CI 2.4–489.5) compared to being 25 years or younger were associated with increased odds of HIV infection. Use of a sterile needle at last injection decreased odds of HIV infection (aOR = 0.3; 95% CI 0.1–0.8). Conclusions The present study observed a decline in prevalence of HIV infections among PWID in Dar es Salaam (8.7%) compared to a previous estimate of 15.5% from an IBBS conducted in 2013. Despite the decrease, HIV prevalence remains high among PWID compared to the general population, and women are disproportionally affected. The decline may be possibly attributed to the on-going implementation of CHIP for PWID, highlighting the need for strengthening the existing harm reduction interventions by incorporating access to sterile needle/syringe and addressing the layered risks for women.


Background
Globally, it is estimated that there are 15.6 million people who inject drugs (PWID), and 30% of these are women [1]. In Tanzania, a national consensus size estimate from 2014 indicated the total number of people who inject drugs (PWID) to be 30,000 individuals [2]. Injection drug use poses health problems in both lowmiddle and high-income countries [3]. Among notable complications of injection drug use is the transmission of blood-borne infections, including HIV and viral hepatitis [4]. PWID are at risk of drug overdoses, including suicide related to drug use [5,6]. In addition, they can be affected by other mental health problems [7]. Furthermore, injection drug use is also associated with stigma, rejection, and incarceration [8,9]. PWID are disproportionally affected by HIV infection as compared to the general population. The global average prevalence of HIV infection among people who inject drugs is 17.8% [1]. Several studies conducted in Tanzania between 2010 and 2018 have reported that the prevalence of HIV infection among PWID ranges between 11.0 and 51.1% [2,[10][11][12][13]. These estimates are 3 to 11 times higher than the prevalence of HIV infection in the general population which is currently 4.5% [14]. In response to the high prevalence of HIV infection observed among PWID, the Government of Tanzania has rolled out a national comprehensive HIV intervention package (CHIP) for PWID in 2014. CHIP contains HIV intervention components such as biomedical interventions, social-behavioral interventions, HIV diagnosis, treatment and care, and management of co-infection and co-morbidities [15]. A total of six medication-assisted treatment (MAT) clinics have been established in mainland Tanzania along with several other harm reduction interventions including a pilot needle exchange program in one of the districts in Dar es Salaam. Since the rollout of the CHIP among key populations in Tanzania, no survey has been done to investigate the prevalence of HIV infection and the risk factors among PWID. The present study provides data on the prevalence of HIV infection and associated risk factors after the introduction of the CHIP in Tanzania.

Study design and setting
A cross-sectional survey was carried out between October and December 2017 in Dar es Salaam, the largest metropolitan city in Tanzania, with an estimated population of 6.7 million in 2020 [16]. This study was part of a broader integrated bio-behavioral survey that was conducted among key populations in Dar es Salaam. The city was chosen because of the high number of PWID [2] and a high prevalence of HIV infection [11,12].

Sample size calculation
Sample size estimates were based on the measurement of precision of point estimates for the prevalence of HIV infection among PWID. The sample size and power calculation for this study was based on an estimated prevalence of 16% from the IBBS which was conducted in Dar es Salaam in 2014 [12]. The 95% confidence interval and a 5% standard error were used to calculate an appropriate sample size of 610 study participants.

Study population
The study participants were PWID aged 18 and above residing in Dar es Salaam. Residency was defined as having lived in the city for at least 12 months before the survey. PWID were included in the study if they had injected drugs in the past 3 months, presented with a valid study coupon, and consented to participate in the study.

Sampling procedures and recruitment
The study participants were recruited through respondent-driven sampling (RDS). RDS is a recruitment and sampling method that combines modified snowball sampling with a mathematical model that weights the sample to compensate for the fact that it was collected in a non-random way [17]. Research assistants in consultation with the study investigators purposefully selected and recruited five seeds (i.e., the initial survey participants). The selection was done so that the seeds came from multiple locations to improve access to potentially hidden networks of PWID. Subsequently, each seed person recruited a maximum of three other participants by giving them recruitment coupons, and this process continued so that the desired sample size and the statistical power were reached. Participants were screened for eligibility by a person who had injected drugs in the past and recovered after therapy. The screener used a standardized screening tool to ensure objectivity. Interviews were done at a private, safe location that was easily accessible to the participants. The screener verified eligibility for all study participants by looking for injection track marks and asking about knowledge of local terms, injection practices, and drug prices. The participants were only included in the study if they possessed a valid recruitment coupon. Each coupon was uniquely barcoded to link recruiters with recruitees, and the barcode was also used to connect participants with their consent forms, questionnaires as well as HIV test results. A total of 8000 Tanzanian shillings (3.50 USD) and 4000 Tanzanian shillings (1.75 USD) were paid to each participant for transport reimbursement and incentive for secondary recruitment, respectively. The amounts used for primary and secondary incentives were approved by the ethical review committee as the average travel costs to the study site.

Data collection training
A team of five research assistants with prior experience in research was recruited and provided with a two-day training on the study protocol, study instruments, research ethics, HIV testing, as well as the procedures for transportation and storage of biological samples. The team was oriented on the use of an open data kit (ODK) for data collection. ODK is a free and open-source data collection software that was customized for our research purpose. Additionally, the field team was trained on interviewing skills and sensitivity in working with key and vulnerable populations.

Data collection tool
A structured questionnaire was adapted from a previous IBBS which was done in 2013 among PWID in Dar es Salaam, Tanzania. The questionnaire was translated from the English to Kiswahili and then back translated to English to ensure face validity (a simple form of validity based on a subjective assessment). The version of the questionnaire used during the interviews was in Swahili, the language spoken by the majority of Tanzanians. The questionnaire was pre-tested and also evaluated with a group of PWID peer educators to ensure appropriate use of language and alignment with cultural and peer norms. Using handheld tablets, trained research assistants administered the questionnaire and collected data using an ODK, which were uploaded directly into the Muhimbili University of Health and Allied Sciences (MUHAS) server daily. The questionnaire gathered information on socio-demographic characteristics, injection practices, sexual risk behaviors, experience of violence, and alcohol use from the PWID participants. Also, Blood samples were also taken and tested for HIV infection.

Socio-demographic characteristics
Socio-demographic characteristics were gathered using a questionnaire to describe the sample population. It included age (continuous), gender (male or female), marital status (never married, married, separated/divorced/ widowed), average level of income per month (continuous), employment status (none, self-employment, employed, informal employment), and levels of education (none, primary education, secondary education).

Injection practices and other risky behaviors for HIV infection
Injection practices and other HIV risk behaviors were measured as binary variables and included the following: frequency of alcohol consumption in the past one month, non-injection drug use other than alcohol in the last 3 months, ever shared needles, having shared needles in the past month, use of sterile needle at last injection, ever used male condoms, experience of physical violence (beaten) in the past 12 months, and forced to have sex in the past 12 months.

Biological data
Pre-and post-test counseling and HIV testing were performed by a nurse and a laboratory technician at the study site, and the national HIV testing algorithms were followed. All specimens were screened using SD Bioline HIV-1/2 3.0 test manufactured by Standard Diagnostics, Inc., Korea [18,19], and positive samples were confirmed using Determine HIV1/2 test manufactured by Alere group of companies [20]. For discordant samples, the tiebreaker test was Unigold HIV test manufactured by Trinity Biotech, Ireland, and was used as a determining test [19].

Data management and statistical analysis
Data analysis was conducted using the RDS Analysis Tool (RDSAT) and STATA version 15.1. To be able to generate sample weight using RDSAT software package, participants were asked for the number of PWID they knew that they could recruit into the survey. They were also asked for the number of PWID they have seen during the last month, a number which was used in the analysis of data when weighing the participant's network size. Per RDSAT data analysis procedures, data were weighted by calculating weight as an inverse of the participant's network size and controlled for clustering by multiplying the weight by the sample size and then dividing by the sum of the weight. Categorical variables were summarized by calculating frequencies and percentages. The median and interquartile range were used to summarize the age variable. The chi-square test was used to compare differences in HIV infection by various categorical variables. Weighted logistic regression analysis was used to estimate bivariate of the associations between various risk factors and HIV infection. Only covariates with a p value threshold of < 0.25 in the bivariate analyses were included in the logistic regression model. Crude and adjusted odds ratios with 95% confidence intervals were reported. All the analyses were two-tailed, and a p value of less than 0.05 was considered statistically significant.

Characteristic of study participants
The distribution of the socio-demographic characteristics of the study participants is shown in Table 1. A total of 611 PWID with a median age of 34 years (IQR 29-38) were recruited through eight recruitment waves. The majority of the study participants were males (94.4%). About half of those who took part were unmarried (52.9%), whereas 28.5% were divorced/separated or widowed. Most of the participants were within the age group 25-44 years (84.5%). About a third (30.4%) of the participants reported a monthly income above 200,000 Tanzanian shillings (about 86 USD), whereas 20.3% reported a monthly income of less than 50,000 shillings (less than 22 USD). The largest proportion of the study participants had completed primary school (67.6%). Almost three quarters (74.7%) of the study participants reported being self-employed. Table 2 shows the distribution of HIV risk factors. Almost one-third (29.0 %) of the study participants said they had used needles previously used by others in the

Prevalence of HIV infection
The overall weighted prevalence of HIV infection was 8.7% (95% CI 6.4-11.8). Sex-disaggregated analysis  Table 3 shows the results of bivariate and multivariate regression analyses. We found that female gender, older age, longer duration of injecting drug use, and using a sterile needle during the last injection were independently associated with HIV infection among PWID. Adjusted weighted analyses of risk factors for HIV infection indicated that being female (aOR = 19.1; 95% CI 5.9-61.8) was associated with increased odds of HIV infection compared to males. We also found that being aged 45 years or older (aOR = 34.2; 95% CI 2.4-489.5) was associated with an increased odds of HIV infection compared with those who were younger than 25 years. Persons who had injected drugs for more than 10 years had about seven times the odds of HIV infection as compared to persons who had injected drugs for less than a year (aOR = 7.3; 95% CI 2.1-25.5). However, using a sterile needle during the last injection was independently associated a 70% decreased odds of HIV infection (aOR = 0.3; 95% CI 0.1-0.8).

Discussion
With data from 611 PWID, we assessed the prevalence of HIV and associated risk factors among PWID in Dar es Salaam, Tanzania. We found that 8.7% of PWID residing in Dar es Salaam, Tanzania, were infected with HIV. This estimate is almost half of the prevalence reported by a similar IBBS conducted in 2013 in the same geographical area and by the same research team (15.5%). In 2011, the results of a survey among PWID attending outreach services in Dar e Salaam found HIV prevalence to be 51.1%, almost six times higher than the prevalence observed in the present study [11]. The decrease in HIV prevalence among PWID suggested by the findings of this study is in line with what has been observed in the general Tanzanian population over the past decade, during which the prevalence of HIV has decreased from 5.7% in 2008 to 4.7% in 2016 [14]. The underlying reasons for the observed decline could be multifactorial, e.g., mortality among HIV-infected PWID or practice of safer injecting and sexual practices among members of this population. It is also possible that the observed decline has occurred as a consequence of safer behaviors resulting from the rollout of the harm reduction services and comprehensive HIV interventions that were implemented in the country in 2012 and 2014 respectively. The latter included MAT services and the community-based and peer-led outreach services provided by several civil society organizations (CSOs) in the vicinity. Globally, the trends of the HIV epidemic among PWID have not been well established. There are different reports indicating both an increase and a decrease in HIV infections in different parts of the world. Studies have shown a reduction in HIV infection among people who use drugs in some areas in the USA and Greece [21,22], while other studies have shown increases, such as in Scotland [23]. Notable factors attributed to the observed reduction of HIV infection among PWID in Greece and Australia included the access to a combination of several prevention strategies such as medication-assisted treatment, antiretroviral treatment for HIV infection, and extensive needle/syringe exchange programs [22,24].
We found that there is a gender disparity in HIV prevalence among PWID. Women had almost nineteen times higher the odds of HIV infection compared to men. As in the previous studies, we observed a higher prevalence of HIV infection among WWID than among their men counterparts [11,12]. This gender disparity in HIV infection may exist because women who inject drugs have multiple layers of HIV infection risk. Women who inject drugs may have additional risks related to unsafe sexual practices, such as multiple concurrent sexual partnerships, having an injection partner who is also a sexual partner, and engaging in commercial sex to be able to purchase drugs [25].
The duration of injecting drug use and the age of PWID were found to be associated with HIV infection in the present study. However, there is a potential collinearity between age of the PWID and the year spent injecting drugs. This may mean that older PWID are more likely to have spent many years injecting drugs and make it possible that both variables measure the same construct (i.e., duration of exposure to drug injection). Having injected drugs for more than 10 years and being aged 45 years or older significantly increased the odds of HIV infection. This finding is similar to other studies among PWID [26]. The cumulative risk resulting from older age and a long duration of exposure to HIV risk behaviors may explain this association. Additionally, the availability of HIV prevention and treatment services for PWID which were established less than 10 years ago could also partly explain these findings. This finding may mean that before the rollout of harm reduction services, PWID were more likely to engage in unsafe injection practices and also other behaviors associated with HIV risk that could explain the increased odds of HIV infection among the older PWID and those who  [2]. By the end of 2015, only 2500 out of nearly 30,000 PWID were on MAT in Tanzania [2]. It is imperative that the efforts to scale up harm reduction interventions are increased to prevent continued HIV risk exposure among PWID. This study found that the use of clean syringes and needles was independently associated with decreased odds of HIV infection among PWID. These findings concur with other studies that have found needled and syringe programs (NSPs) to be effective in reducing HIV infection [24,27]. It is important to carefully evaluate the current National Guideline for a Comprehensive HIV Package for Key Population and strengthen the NSPs as part of harm reduction for PWID in Tanzania.

Limitations
There are some limitations to this study. First, this is a cross-sectional study which provides a snapshot of the magnitude (prevalence) of HIV infection among PWID. However, it does not allow for evaluation of the temporal relationships between HIV infection and the predictors of interest. Secondly, there is a possibility of desirability bias due to the collection of sensitive information related to injecting and sexual behaviors. Thirdly, there is a potential for selection bias inherent in the respondent-driven sampling which is a non-probabilistic sampling approach. To mitigate this, seeds were selected in such a way that they could represent the diversity of PWID, including gender (male and female), various age groups, and analysis, and took into account network size and used special software (RDSAT) designed for this kind of recruitment. Lastly, out of 611 participants, 35 were WWID constituting only 5.6% of the study sample. Despite the fact that previous studies in the same city have also reported fewer women who inject drugs and higher HIV infection rates, the lower proportion of women in this study may have biased gender comparison [12]. Difficulties in recruiting WWID in Tanzania context can be explained by the idea that, WWID are subjected to increased risk of physical violence (such as assaults and robbery) as well as stigma and discrimination which in turn generate mistrust against healthrelated activities and eventually push them in hiding and Only covariates with a p value threshold of < 0.25 tin the bivariate analysis were included in the multivariate analysis AOR adjusted odds ratio, cOR crude odds ratio, CI confidence interval *A p value of 0.05 and below was considered statistically significant **Unweighted Chi-square analysis was used to summarize of the percentage of HIV infection by sex being less accessible through the study's RDS [28]. In addition, some WWID are engaging in commercial sex work which happens mostly during the night time and hence make it difficult to be accessed during the day time when RDS was taking place.

Conclusion
The results from this study suggest that the prevalence of HIV among PWID in Dar es Salaam has decreased considerably in the past 4 years. Despite the declining trends of HIV infection for both the general population and among PWID, the risk of being HIV positive for PWID is almost twice that of people in the general population. The decline in the prevalence of HIV could be attributed to increased access to comprehensive HIV interventions for key populations in Dar es Salaam, which includes harm reduction intervention and access to HIV testing, care, and treatment. The results may be a sign of successful interventions that are helping to realize the global targets of 90-90-90 and three zeros [29]. Comprehensive HIV interventions should be rolled out across the country, especially in the areas where there is a substantial number of PWID.