Article Text
PDF
Abstract
Introduction Pre-exposure prophylaxis (PrEP) is an effective HIV prevention tool when taken as prescribed. However, suboptimal use may challenge its real-life impact. To support female sex workers in their efforts to prevent themselves from HIV, it is essential to identify factors that contribute to early disengagement from PrEP care. In this study, we aimed to estimate the risk of early disengagement from PrEP services among female sex workers in Tanzania and associated factors using a socioecological model as a guiding framework.
Methods The study was conducted as part of a pragmatic mHealth trial for PrEP roll-out in Dar es Salaam in 2021. We estimated the risk of early disengagement, defined as not presenting for the first follow-up visit (within 56 days of enrolment), and its associations with individual, social, behavioural and structural factors (age, self-perceived HIV risk, mental distress, harmful alcohol use, condom use, number of sex work clients, female sex worker stigma and mobility) using multivariable logistic regression models, with marginal standardisation to obtain adjusted relative risks (aRR).
Results Of the 470 female sex workers enrolled in the study, 340 (74.6%) did not attend the first follow-up visit (disengaged). Mental distress (aRR=1.14; 95% CI 1.01 to 1.27) was associated with increased risk of disengagement. Participants who reported a higher number of clients per month (10–29 partners: aRR=0.87; 95% CI 0.76 to 0.98 and ≥30 partners: aRR=0.80; 95% CI 0.68 to 0.91) and older participants (≥35 years) (RR=0.75; 95% CI 0.56 to 0.95) had a lower risk of disengagement.
Conclusions and recommendations Early disengagement with the PrEP programme was high. Mental distress, younger age and having fewer clients were risk factors for disengagement. We argue that PrEP programmes could benefit from including mental health screening and treatment, as well as directing attention to younger sex workers and those reporting fewer clients.
- HIV
- prevention strategies
- epidemiology
- public health
- environmental health
Data availability statement
Data are available upon reasonable request. The data can be obtained upon reasonable request by the principal investigator, EJM; email: elia.mmbaga@medisin.uio.no.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Oral pre-exposure prophylaxis (PrEP) in an effective HIV preventive drug for key populations at increased risk of HIV, but studies have shown suboptimal use and engagement with PrEP services, challenging its ‘real life’ impact.
There is limited evidence from epidemiological studies on factors influencing engagement with services among female sex workers in sub-Saharan Africa
WHAT THIS STUDY ADDS
Using a socioecological model to guide investigation, this study found that younger age, mental distress and having few sex work clients were associated with early disengagement
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The study highlights the need for integrated programmes for key populations that addresses more than one disease or social circumstance at a time, such as integrated mental health and HIV services and tailored interventions for younger female sex workers.
Introduction
Female sex workers are one of the so-called key populations in the HIV epidemic and are 30 times more likely than other adult women to contract HIV.1 Their vulnerability to HIV is shaped by a complex interplay of individual, social, behavioural and structural factors, including, but not limited to, unprotected sex, multiple sexual partnerships, stigma, gender-based violence (GBV) and legislation that unfavourably impacts sex work.2 Approaches that have been effective in reducing HIV risk among female sex workers include behaviour change communication, harm reduction services, peer-led programmes, as well as initiatives and strategies focused on solidarity and empowerment.3 Over the last decade, oral pre-exposure prophylaxis (PrEP), antiretroviral (ARV) medication taken as a daily pill, was found to be efficient in preventing HIV acquisition.4 Following this, in 2015 WHO recommended the use of PrEP as a complementary prevention approach for populations at high risk of HIV infection. Many countries, including Tanzania, have later incorporated oral PrEP into their HIV prevention programmes.
Even though trials have demonstrated PrEP’s efficiency, several studies have identified implementation challenges along the PrEP care continuum5–12 that may challenge its anticipated impact. The ‘care continuum’ or ‘prevention cascade’ typically includes aspects such as awareness and/or knowledge, motivation, uptake, adherence, engagement and continuation in programmes and even measurements of restarting after stopping use.13 14 A recent systematic review including several population groups globally found high discontinuation rates, defined as loss to follow-up or self-reported PrEP stoppage, with an overall pooled prevalence of 41% (48% in sub-Saharan Africa) within the first 6 months of PrEP initiation. Factors that were most commonly reported to be associated with discontinuation were young age, being female and being transgender. Only three of the 56 included studies were conducted among female sex workers, among whom the pooled prevalence was higher at 50% (25.7–75.4%).15 Hendrickson et al proposed that one of the success criteria of a PrEP programme is that the client either remains engaged in care (presents to follow-up) or has discontinued PrEP in consultation with a healthcare provider.16 This enables the provider to conduct risk assessments of individual users, discuss other prevention options with them if they wish to discontinue PrEP and to detect any cases of seroconversion during PrEP use. We argue that attending the first follow-up visit is particularly important as the need for more information relating to side effects and correct use is expected to be highest early on. Additionally, the indication for PrEP use is less likely to have changed within such a short time frame.14
Tanzania is implementing daily oral PrEP programming in several regions, including in the country’s largest city Dar es Salaam, where approximately 15% of female sex workers are estimated to be living with HIV.17 Studies from Tanzania have found that willingness to use PrEP is high among female sex workers and female bar workers, where members of the latter occupational group are known to commonly exchange sex for money.18 19 However, the first demonstration trial in the country conducted among female sex workers and other key population groups revealed that less than half of those starting PrEP attended the first follow-up visit,20 highlighting the need for investigation into factors that might influence disengagement from PrEP care.
Socioecological frameworks can aid in this understanding. These frameworks21 exist in slightly different forms, but have in common that they aim to encompass the numerous factors that can have impact on risk of disease or that can influence health behaviour from the micro (individual) level to the macro (or structural) level. Drawing on such frameworks offers the advantage of preventing overly simplistic assumptions. For instance, it ensures that disengagement from PrEP services is not merely viewed as an individual behaviour, but a practice influenced by a complex interplay of individual, behavioural, social and structural factors. Specific frameworks to aid in the understanding of HIV risk and HIV-related behaviour using a socioecological approach have previously been proposed.22–24 In this study, we study the association between early disengagement from a PrEP programme, defined as not attending the first scheduled follow-up visit, and eight factors that have been previously shown to be associated with HIV or the PrEP continuum in various population groups. That said, it is noteworthy that knowledge from quantitative studies on factors concerning PrEP use among female sex workers is scarce. The factors explored in this study are situated in different layers of the socioecological model, although several factors can be seen as belonging to more than one layer. At the individual level, studies have shown that older age at PrEP initiation can predict continuation among female sex workers,25 26 while high self-perceived HIV risk has been linked to PrEP uptake and continuation.27 Often referred to as psychosocial factors, poor mental health and alcohol use have been linked to adherence to or continuation with PrEP.25 26 On the interpersonal/network level, we explore if sexual and/or prevention practices such as condom use and number of sex work clients are factors that influence disengagement with PrEP services. On the community level, the hypothesis is that stigma related to sex work influences disengagement from PrEP, as stigma can negatively affect health service use.28 Finally, we investigate if mobility for sex work, sometimes viewed as a sociostructural factor, increases the risk of disengagement, as this link has been suggested in a recent study.29
This study seeks to narrow the evidence gap on factors associated with disengagement in PrEP programmes among female sex workers in one of the first PrEP programmes in Tanzania, with the ultimate aim of informing PrEP programming locally and regionally.
Methods
Study design and setting
This paper presents an analysis of a cohort of female sex workers from an intervention site of a pragmatic mobile health (mHealth) PrEP study conducted in the city of Dar es Salaam, Tanzania.30 31 Dar es Salaam is the financial capital of Tanzania, with a prevalence of HIV among female sex workers among the highest in the country.32 The study was conducted in collaboration with a local sex worker-friendly satellite clinic, set up within a gated compound in a central part of the city. During participants’ first visit they were prescribed PrEP and received the study intervention, an mHealth app called ‘Jichunge’ (‘protect yourself’) to promote adherence to medication and follow-up visits. This included onboarding and instructions about app features, such as an alarm to remind the participants of pill taking, gamification aspects and the opportunity to contact staff in case of any questions. The clinic was open Mondays to Fridays, and some Saturdays, from morning to afternoon and was highly flexible when it comes to accommodating participants’ schedules including catering for drop-in. For the subanalysis presented in this paper, a single proportion formula with 50% used as proportion of disengagement and a 5% margin of error would require a minimum sample size of 385. The total sample size of the mother study, 470, was therefore sufficient to estimate the outcome.30
Study participants
Female sex workers were defined as women who reported to have sold sex for money or goods in the last 3 months before study recruitment. Eligibility criteria were being 18 years or older, having lived in the city of Dar es Salaam for at least 6 months preceding the survey, owning a smartphone, ability to give informed consent and being eligible for and interested in starting PrEP. Participants were recruited between March and July 2021 using respondent-driven sampling (RDS). RDS is a method commonly used when a sampling frame is not available.33 The method can be considered a modified ‘snowballing’ sampling strategy combined with a mathematical model that weights the sample to mitigate the biases associated with oversampling or undersampling certain groups. We identified three initial female sex workers (‘seeds’) to start the recruitment process and added more seeds as the recruitment progressed. They were given three coupons each and asked to recruit others from the study population using these coupons as study invitation. In the selection of the ‘seeds’, we sought a variation of sociodemographic and sex work characteristics, such as diversity in age, socioeconomic status, residential neighbourhood and type of sex work (ie, where they got in touch with clients). Most of the seeds were good recruiters, while one of the seeds did not recruit, and we were informed later that another seed might not have distributed the coupons herself. Although aiming for a diversity in type of sex work, most of the recruited participants reported to work from bars. Implementation challenges are common in RDS studies.33 We sought to mitigate these by including more seeds with varying characteristic, recommended if recruitment chains die out or recruitment is too slow.34 The final number of seeds was 9. We also learnt through qualitative accounts that some of the women who got in touch with clients through internet/phone, street or brothels also sometimes worked from bars (bars could, for instance, be located adjacent to brothels), which indicated that the ‘type of sex work’ category was more fluid in our context. The recruitment procedure was repeated with new recruits yielding new waves of participants and continued until we reached the desired sample size. In line with RDS,34 the women received a reimbursement for participation (8000 Tanzanian shillings≈US$3.5) and for recruiting new participants (4000 Tanzanian shillings≈US$1.75). Participants were screened for eligibility by study staff, including peer educators. The participants were assessed for PrEP eligibility by the local clinic using standard national criteria: being HIV negative and at substantial risk of HIV infection with no suspicion of acute HIV infection, serum creatinine clearance >60 mL/min and being willing to consent to and use PrEP as prescribed.35 Eligible participants were prescribed PrEP at enrolment, and follow-up was scheduled as monthly visits. The clinic provided PrEP to all those who were eligible regardless of mobile phone ownership, which was only a criterion for study recruitment/enrolment.
Participant and public involvement
The research group has a long-standing collaboration with peer educators from key populations in Tanzania. These were integral in the research assistant team and assisted in the development of the mother study intervention (the mHealth app), the recruitment (identified ‘seeds’) and data collection.
Data collection procedures
Trained research assistants administered a questionnaire through face-to-face interviews at enrolment. Data on motivation for PrEP use (eg, self-perceived HIV risk), sex work history and practices, as well as sociodemographic characteristics and socioenvironmental factors (eg, mobility, stigma, substance use) and questions related to mHealth, were collected. The interview lasted for about an hour and was conducted in the national language, Swahili, by Swahili-speaking staff. Participants were registered when they came for their first follow-up appointment. Participants’ responses during interviews were directly plotted into a web-based questionnaire solution (‘Nettskjema’) and submitted into a highly secure platform: Services for Sensitive Data developed and operated by the University of Oslo.36
Outcome variable
Disengagement from care was the outcome variable, defined as no evidence of having attended the first scheduled follow-up appointment in the PrEP programme. The follow-up appointment was typically scheduled 28 days after the enrolment visit. In the operationalisation of the outcome variable, we included an additional 28 days to allow for delay, in line with other studies in the field before declared as a missed appointment.9 25
Exposure variables
Exposure variables were age, self-perceived HIV risk, mental distress, harmful alcohol use, condom use, number of sex work clients, female sex worker stigma and mobility for sex work. Mental distress was defined as a positive screen for depression using the Patient Health Questionnaire 2 (PHQ-2)37 and/or a positive screen for anxiety disorder using the 2-item questionnaire Generalized Anxiety Disorder (GAD-2).38 PHQ-2 asks about the frequency of depressed mood and lack of joy, while GAD-2 inquires about frequency of anxiety symptoms, both during the last 2 weeks. The four answer options range from not at all (0 point) to nearly every day (3 points). We used the standard cut-off of 3 for both screening tools. Harmful alcohol use was assessed with the WHO Alcohol Use Disorder Identification Test.39 We used a Swahili version previously validated among patients with traumatic brain injury in Tanzania40 with minor modifications in the wording. The 10-item instrument inquires about aspects of problematic alcohol use, including frequency and quantity of alcohol intake, harmful consequences of alcohol use and dependency. The maximum score is 40, and a cut-off of 16 is typically used to indicate a harmful pattern of use (or likely dependency). Sexual practices and previous prevention use were assessed by asking ‘Did you use a condom the last time you had vaginal sex with a client?’ and ‘In the past one month: how many clients did you have vaginal sexual intercourse with?’. Stigma was assessed using a female sex worker stigma scale, previously used in Tanzania and the Dominican Republic41 with a reliability coefficient of 0.88. We dichotomised the scale at the median. Mobility was assessed by the question: ‘Within the last 6 months have you travelled to another city or county to perform sex work?’. Self-perceived HIV risk was categorised into ‘high’, ‘medium/low/no’ and ‘don’t know’.
Confounders
In addition to the exposure outlined above, several factors were considered potential confounders for each exposure-outcome pair (online supplemental file 1). Education level was categorised as primary (or lower) or secondary (or higher); income from sex work was divided into four categories based on participants’ estimations of income. Social support was measured using the Duke-UNC Functional Social Support Questionnaire42 and had a reliability coefficient of 0.88. GBV was measured by two questions: ‘Have you experienced physical violence (like being beaten) during the last 12 months?’ and ‘Have you been forced to have sex during the last six months?’. Answering yes to either of these questions was considered a positive screen for GBV. As a recent study found that new entrants to sex work were less likely to take PrEP every day,43 the number of years since the participant first sold sex (<5 or ≥5 years) was also included as a potential confounder.
Supplemental material
Statistical analysis
As participants were recruited using RDS, those with larger networks have potential to be over-represented in the sample. We therefore weighted the data using individualised weights, proportional to the inverse of a respondent’s network size.33 Categorical variables were summarised by estimating proportions and presented by PrEP disengagement status. To further estimate the associations between exposure variables and the outcome, we first conducted unadjusted logistic regression. We then proceeded building multivariable logistic regression models, one model for each exposure-outcome pair, to obtain adjusted relative risk (aRR). To aid in selection of variables to adjust for in each model we drew directed acyclic graphs (DAG) using DAGitty.net44 guided by prior research, theoretical and plausible assumptions about relationships between variables (online supplemental file 1). Age was decided a priori as a forced variable to be included together with the main exposure in all models. We further used a modelling strategy described by Greenland et al,45 aiming to reduce the mean square error of the effect measure coefficient. To assess the total effect of age, no adjustment was necessary, thus only unadjusted estimates are presented. For all analyses for assessing associations, we used unweighted data based on the evidence that unweighted regression analysis is less prone to bias than weighted analysis of RDS data.46 47 As the outcome was common, following the logistic regression, we used marginal standardisation to obtain relative risk (RR) instead of the OR.48 All analyses were two tailed with a significance level of 5% and conducted in STATA/SE V.16.0 (StataCorp 2019).
Ethics
The Tanzania Commission for Science and Technology granted individual research permits to foreign investigators. Permission to work with the clinic was granted by the clinic authorities. All participants were informed about the study’s aims and proceedings and provided written informed consent. The study addressed national research priorities and strengthened local research capacity with a team made up by majority Tanzanian nationals. More details concerning the research collaboration can be found in the ‘Author Reflexivity Statement’ (online supplemental file 2). The study followed the principles of the Declaration of Helsinki.
Supplemental material
Results
Of the 470 female sex workers enrolled into the study, 340 (74.6% weighted) disengaged from care at month 1 follow-up (figure 1). Table 1 shows the distribution of sociodemographic, behavioural and structural characteristics of the study participants by disengagement status.
Early disengagement with PrEP services (month 1) among female sex workers in Dar es Salaam, Tanzania.
Distributions of baseline sociodemographic, behavioural and structural characteristics of female sex workers by early PrEP engagement status (N=470)
Three-quarters (n=153) of female sex workers below 25 years did not attend month 1 follow-up versus 57.8% (n=26) of women who were 35 years or older. Among women who had sex with less than 10 clients per month, 82.3% (n=102) were disengaged from care at month 1; in contrast, among women with 30 or more clients, 63.0% (n=110) were disengaged. Of the women who scored positive for mental distress, 79.7% (n=114) did not attend month 1 follow-up, in contrast to 69.1% (n=224) of the women with a negative score (table 1).
In regression analysis (table 2), female sex workers with a positive screen for mental distress had a 14% increased risk of disengagement at month 1 follow-up compared with women who had a negative screen (aRR=1.14; 95% CI 1.01 to 1.27). Female sex workers reporting between 10 and 29 clients per month were 13% less likely to be disengaged compared with women with fewer than 10 clients/month (aRR=0.87; 95% CI 0.76 to 0.98), while women with 30 clients or more had 20% reduced risk of disengagement (aRR 0.80; 95% CI 0.68 to 0.91). The fewer the clients the women had, the higher the risk of disengagement: for example, in sensitivity analysis with even finer categories we found an RR=1.36 (95% CI 1.15 to 1.57) for those with zero to four clients compared with those ≥30, while those with five to seven clients had a comparable RR of 1.26 (95% CI 1.05 to 1.47). Age was also associated with early disengagement, with older age being significantly protective (RR=0.75; 95% CI 0.76 to 0.98). We found no statistically significant relationship between our outcome and the four other exposures in adjusted analysis: condom use (aRR=0.92; 95% CI 0.81 to 1.02), harmful alcohol use (aRR=1.02; 95% CI 0.90 to 1.13), female sex worker stigma (aRR=1.04; 95% CI 0.92 to 1.16) or mobility (aRR=0.91; 95% CI 0.80 to 1.01).
Unadjusted and adjusted logistic regression modelling with marginal standardisation, assessing associations between individual, social, behavioural and structural exposures and early PrEP disengagement among female sex workers (n=470)
Discussion
Early disengagement from care was high in this cohort with three-quarters (74.6%) of the female sex workers not attending the month 1 follow-up for PrEP refill and medical check-up. Disengagement was associated with younger age, mental distress and having fewer clients per month.
Like in our study, several PrEP programmes and studies among female sex workers in other sub-Saharan countries have also experienced high rates of disengagement from PrEP services. In the Treatment and Prevention for Female Sex Workers (TAPS) Demonstration Project conducted at two urban clinics in South Africa, 53% of the participants came for month 1 follow-up, while 22% were seen after 12 months.7 Another demonstration study in Benin found that the overall retention at the end of the study was 47.3%,29 while a study conducted at facilities for sex workers and men who have sex with men in South Africa showed that about 50% discontinued PrEP during the first 6 months.27 In Kenya, 40.3%, 26.3% and 14% of the female sex workers were seen after 1, 3 and 6 months, respectively.49 In contrast, a study from public health centres in Senegal saw high retention after 1 month (90%), 6 months (79.9%) and 12 months (73.4%), illustrating that achievement of high attendance to PrEP services among female sex workers is possible.25 Our study found higher early disengagement in Tanzania than in any of the mentioned studies. One possible explanation for this could be that screening and enrolment were conducted at the same visit in our study, while in the TAPS and Benin studies, enrolment was done at a second visit, which could have meant that these participants were more motivated for PrEP care. We, however, note that self-perceived HIV risk was high in our cohort, thus an important aspect of PrEP motivation was indeed present.50
Reasons for disengagement from PrEP health programmes are usually multifactorial. In the studies cited in the previous paragraph, side effects and self-perceived HIV risk,27 mobility among participants29 and not using PrEP consistently7 were some of the factors mentioned. Although a thorough review of possible reasons is outside the scope of this paper, some of the issues brought up by participants in our study were side effects, stigma related to PrEP use and other obligations, such as travels and work. It should also be noted that daily dosing for some was mentioned as an obstacle. Unfortunately, there is insufficient evidence for the effect of event-driven PrEP (PrEP taken before and after sexual risk) for women, and thus this is not a recommended dosing strategy for this group.51 In the Senegalese study, older age was the only factor that was found to be associated with retention; the authors did not elaborate on what made this programme particularly successful. It is, however, noted that sex work is legal in Senegal and registered sex workers are obliged to attend mandatory health services. This programme included both registered and non-registered sex workers which might have had a positive impact on attendance.
Scholars have argued that public health narratives of HIV tend to leave out the importance of the social world in shaping healthy practices, and instead emphasise individual and biomedical interventions as the main solutions.52 Our point of departure was to study disengagement from PrEP services, not mainly as an individual behaviour, but as a practice influenced by social and structural as well as individual and behavioural factors within a socioecological framework. One of these factors was mental distress. Previous studies have pointed to mental health as a central concern for an effective PrEP care continuum.53 54 Mental health outcomes are typically situated at the individual level, but heavily influenced by social circumstances such as GBV, stigma and poverty, and often considered a sociopsychological factor. In this study, mental distress was defined as having symptoms of depression and/or anxiety. Of the participants, about a third screened positive for mental distress (n=143, 32.2%), and these had an 18% higher risk of PrEP discontinuation. Mental distress could have affected health-seeking behaviour through pathways such as avoiding social situations, experiencing less interest and engagement in things and feelings of anxiety.54 Combining mental health services with HIV prevention and treatment could produce synergistic effects. In this way, PrEP care could serve as an entry point for promoting mental health, and vice versa.53 In the Tanzanian PrEP implementation framework, mental health screening is briefly mentioned,55 yet efforts could be made to expand this integration more fully by promoting holistic health services for sex workers that actively seek to address the complexity of their health needs by learning from earlier successful efforts in targeting this key population.3 It is, however, important that screening and subsequent treatment of mental health conditions should not be made a prerequisite for access to PrEP, as PrEP ought to be easily available.53
The second factor associated with early disengagement was having a lower number of sex work partners, a social factor as it involves (at least) two people engaging in sexual contact. The finding could be related to lower HIV risk perception among sex workers with fewer sex clients which is in line with studies that have found that a higher degree of sexual risk can predict higher adherence to PrEP.56 However, when we examined self-perceived HIV risk and disengagement, we did not find a significant association, pointing to other differences between those with higher and lower number of clients also being important. Another explanation could be fear of losing clients and thus income due to PrEP use among those with fewer paying partners, as they might be more dependent on the earning from each single client. Indeed, in a study from Tanzania before PrEP was available, female sex workers expressed concerns about stigma from clients as PrEP could be mistaken for ARVs and thus indicate HIV positive status.57 Work striving to limit public stigma related to both HIV and PrEP use continues to be particularly important. Additionally, services should be keen to discuss PrEP use with clients even in the context of relatively few sex work clients.
We also found that female sex workers aged between 18 and 25 years had an increased risk of early disengagement from PrEP services compared with those aged above 35. This finding corroborates earlier evidence that higher age predicts consistent PrEP use25 26 and is likely related to factors ranging from the individual to the structural level (neurocognitive development, peer influence, socioeconomic and structural vulnerabilities, etc).58 Health services need to be particularly considerate when encountering young women who sell sex, ensuring that they experience friendly services and providers that are suitably trained on their needs. There would also seem to be a need for testing new innovative interventions tailored to this group. Additionally, as young women’s vulnerability to HIV is related to social and structural inequities, actions on the social determinants of health, such as social and economic empowerment, and efforts to alter sexual norms and gender-related norms cannot be underestimated.
There are a few important limitations to this study. First, there is no consensus on how to best perform analysis of associations with RDS data, thus the findings should be interpreted with some caution. We have, however, followed recommendations using unweighted data based on studies in the field. Second, one of the eligibility criteria for study enrolment was smartphone ownership. This could have affected representativeness, as women who own a smartphone may differ from those who may not afford a smartphone. Third, as with many survey datasets, it is not straightforward to assert the direction of effect between exposures and confounder variables measured at baseline, as the temporality aspect is lacking. As this in turn guides adjustment selection, some caution is warranted when interpreting the estimates. However, directed by our DAG, we adjusted for potential confounders in each model, several of which the direction of effect is known, and typically identified as important in this population. Fourth, self-report measures are subject to the risks of desirability bias. We sought to mitigate this by training the research assistants in sensitive communication in relation to key populations, as well as including members of these groups in our team. Considering the relatively high proportions reported of what could be considered less socially acceptable behaviour (harmful alcohol use, specific sexual practices, etc), this might have been less of a problem in our study. Fifth, the study recruited participants from the intervention site who have been receiving reminder messages to attend clinics. There is potential that the intervention might have improved engagement with the PrEP programme, hence underestimating our estimated disengagement rate. However, given that the estimated disengagement is high, the intervention would not affect the conclusions presented in this paper. Finally, we chose to focus on a few specific individual, social, behavioural and structural factors in this study. There are likely other relevant factors that also affected the risk of disengagement, but these were outside the scope of this study.
Conclusion
Our study conducted among female sex workers in Dar es Salaam found high early disengagement with PrEP services, and this was associated with mental distress, younger age and having fewer clients. By situating and discussing these findings within a socioecological framework, we hope to have contributed to disengagement being recognised as a complex phenomenon that is shaped by elements within multiple nested layers and thus to an increased understanding of so-called non-compliance issues in HIV prevention. The study can further provide policy makers, health practitioners and PrEP users with valuable insights into how and where to best focus their efforts. Integrating mental health services with HIV prevention, increasing research focus on young sex workers, while ensuring that information regarding indication and need for PrEP are provided to all sex workers, regardless of client numbers, can be one way forward. Finally, we also call for qualitative research to further explore the sociocultural contexts and individual perspectives that may contribute to early disengagement.
Data availability statement
Data are available upon reasonable request. The data can be obtained upon reasonable request by the principal investigator, EJM; email: elia.mmbaga@medisin.uio.no.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by the Ethical Review Committee of the Muhimbili University of Health and Allied Sciences (MUHAS), the Tanzania National Health Research Ethics Committee (NatREC) and the Regional Committee for Medical and Health Research Ethics (REK) in Norway. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We acknowledge all the participants taking part in the research and the field assistants who collected the data. We additionally extend our gratitude to Melkizedeck Thomas Leshabari and Inga Haaland for their role in planning and execution of the study.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Stephanie M Topp
Contributors All authors conceptualised and designed the study. HOL analysed and interpreted the data, with inputs from all the authors. HOL was responsible for the write-up with the assistance of EJM. KM, CHM and MRK reviewed the work critically for its content, and all authors have approved the final version including being accountable for all aspects of the work. EJM is responsible for the overall content as the guarantor.
Funding This research was funded by the Research Council of Norway through the Global Health and Vaccination Programme (GLOBVAC, project number: 285361). The project is also part of the European & Developing Countries Clinical Trials Partnership (EDCTP2) programme supported by the European Union.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.