Violence across the Life Course and Implications for Intervention Design: Findings from the Maisha Fiti Study with Female Sex Workers in Nairobi, Kenya

We examined violence experiences among Female Sex Workers (FSWs) in Nairobi, Kenya, and how these relate to HIV risk using a life course perspective. Baseline behavioural–biological surveys were conducted with 1003 FSWs June-December 2019. Multivariable logistic regression models were used to estimate the adjusted odds ratio (AOR) and 95% confidence intervals (CI) for associations of life course factors with reported experience of physical or sexual violence in the past 6 months. We found substantial overlap between violence in childhood, and recent intimate and non-intimate partner violence in adulthood, with 86.9% reporting one or more types of violence and 18.7% reporting all three. Recent physical or sexual violence (64.9%) was independently associated with life course factors, including a high WHO Adverse Childhood Experiences (ACE) score (AOR = 7.92; 95% CI:4.93–12.74) and forced sexual debut (AOR = 1.97; 95% CI:1.18–3.29), as well as having an intimate partner (AOR = 1.67; 95% CI:1.25–2.23), not having an additional income to sex work (AOR = 1.54; 95% CI:1.15–2.05), having four or more dependents (AOR = 1.52; 95% CI:0.98–2.34), recent hunger (AOR = 1.39; 95% CI:1.01–1.92), police arrest in the past 6 months (AOR = 2.40; 95% CI:1.71–3.39), condomless last sex (AOR = 1.46; 95% CI:1.02–2.09), and harmful alcohol use (AOR = 3.34; 95% CI:1.74–6.42). Interventions that focus on violence prevention during childhood and adolescence should help prevent future adverse trajectories, including violence experience and HIV acquisition.


Introduction
Violence against women and girls (VAWG) is common around the world, with almost one in three women (31%) experiencing physical or sexual violence (PSV) against them by an intimate partner or sexual violence from someone other than their partner in their lifetime [1]. Intimate partner violence (IPV) prevalence varies substantially between countries and in different geographical locations within countries [2]. In Kenya, the 2014 Demographic and Health Survey estimated that 47% of women aged 15-49 years have ever experienced PSV from a spouse or someone else, with 20% reporting physical violence and 8% reporting sexual violence in the previous 12 months [3]. Although IPV is the most common form of violence experienced, VAWG is multi-faceted, and women can experience violence from different perpetrators (e.g., family members, strangers, etc.) and in different forms (e.g., childhood sexual abuse, forced sexual debut, physical, emotional and/or sexual IPV, elder violence) across their life course [1]. For both women and men, experiences in childhood, such as experiencing violence or witnessing violence against their mother is associated with increased risk of experiencing or perpetrating violence in adulthood [4][5][6][7]. In addition, men who are violent towards women often have a clustering of risk behaviours that can increase the chances that they and/or their sexual partners will acquire HIV and other STIs [4,[8][9][10]. At least three prospective cohort studies from sub-Saharan Africa (SSA) confirm that physical IPV and any IPV are associated with increased HIV incidence [11].
Female sex workers (FSWs) are at increased risk of both violence and HIV compared to all women [12,13]. A systematic review found that the prevalence of workplace violence only (i.e., not including IPV) was 45-75% ever and 32-55% in the past year [14]. In addition, the relative risk of acquiring HIV is 30 times greater among FSWs compared with women of reproductive age from the general population [15]. The criminalisation of sex work in most settings globally means that FSWs work on the margins of society and legality, with little recourse for violence against them, meaning perpetrators can continue with impunity [16]. Police officers and other law enforcement agents frequently perpetrate instead of protect FSWs from violence, including through police arrest and imprisonment, as well as demanding sex in exchange for non-arrest [14,17,18]. FSWs have multiple identities, including as daughters, wives/intimate partners, mothers, and key providers for their families [19]. Thus, in addition to 'workplace' violence, they are also at risk of childhood, family, partner, and other non-sex-work related violence [20,21]. However, little is known about the co-occurrence of violence across the life course among FSWs. In addition, to our knowledge, there have been no studies to date that have examined violence exposure among the children of FSWs. Together these have important implications for the 'transfer' of violence between generations and the need for interventions.
Kenya is located in East Africa and has one of the largest numbers of people living with HIV globally. The adult HIV prevalence is estimated to be 4.0% and is higher in women (5.4%) compared with men (2.6%) [22]. Nairobi is the capital and largest city in Kenya, with a population of approximately 4.4 million people [23]. Nairobi county has an estimated 2032 'hot-spots' where approximately 39,600 women sell sex [24]. Types of hot-spots include bars with lodging (where sex work can take place), bars without lodging, guest houses, streets, sex dens, and uninhabited buildings. Around 73% (29,000) of FSWs in Nairobi are served by seven Sex Worker Outreach Programme (SWOP) clinics which provide peer education and outreach, comprehensive clinical services, including HIV testing and treatment, and condom distribution. Additional programmes provide services for other FSWs. The Maisha Fiti study is a mixed-methods longitudinal study which aims to examine the biological impact of violence and harmful drinking on inflammation in the blood and the genital tract. Using baseline data collected from June-December 2019, the aim of this paper is to examine (i) violence experiences across the life course and (ii) associations of risk factors across the life course with recent violence experiences among FSWs in Nairobi.

Study Design and Sampling
The Maisha Fiti study was designed in consultation with the FSW community in Nairobi, as well as with peer educators and staff working at the seven SWOP clinics. The study was powered to detect genital inflammation among women who had experienced recent PSV. Assuming 2:1 exposure to recent violence, enrolling 750 HIV-negative women would detect a 10% absolute difference in the proportion of women who have genital inflammation (25% vs. 15%) at 90% power. The HIV prevalence among FSWs in Nairobi is approximately 25%, and thus, the target sample size was 1000 FSWs for the study.
All women attending SWOP clinics have a unique enrolment number supported by biometrics (fingerprints). Enrolment numbers were selected from all clinic attendees who had accessed SWOP services in the past 12 months, who were aged 18-45 years, and who did not have an underlying chronic illness (other than HIV) that was likely to alter host immunology. Of 29,000 FSWs enrolled at one of the seven SWOP clinics across Nairobi, 10,292 met these inclusion criteria and were included in the sampling frame. Additional exclusion criteria (assessed during study enrolment) were current pregnancy or breastfeeding. Of the 10,292 FSWs, 1200 were randomly selected for study participation with numbers weighted by the total population of FSWs enrolled in each SWOP clinic. Women aged <25 years were oversampled to enable sufficient power for analyses stratified by age. Thus, although <25 year olds represented 11.69% of women meeting the study inclusion criteria, we randomly selected 21.14% to participate in the study (sampling fraction: <25 year olds 17.6%; 25+ years 8.7%).

Ethics and Informed Consent
The Maisha Fiti study was ethically approved by the Kenyatta National Hospital-University of Nairobi Ethics Review Committee (KNH ERC P778/11/2018), the Research Ethics Committees at the London School of Hygiene and Tropical Medicine (Approval number: 16229) and the University of Toronto (Approval number: 37046). Selected women were telephoned, informed about the study, and invited to the study clinic, where the study team gave them detailed study information both verbally and through a written participant information leaflet. Information about the study was also relayed to the sex work community by seven peer educators (The Maisha Fiti Study Champions). Consenting women undertook a pregnancy test, and those who were not pregnant or breastfeeding were enrolled in the study and completed a behavioural-biological survey. Those found to have experienced recent violence or to have mental health problems or suicidal behaviours were referred to a trained counsellor employed as part of the study team. All women who tested positive for HIV were counselled and referred for HIV care at their SWOP clinic. All women who tested positive for bacterial STIs were offered treatment free of charge.

Behavioural-Biological Survey
Women completed a baseline questionnaire focused on socio-demographics, sexual history and practices, reproductive health, Adverse Childhood Experiences (ACEs), sex work characteristics, intra-vaginal washing practices, mental health problems, and alcohol and substance use. The WHO Violence Against Women 13-item questionnaire, which measures the frequency and severity of Intimate Partner Violence (IPV), was adapted to include violence by non-IPs (e.g., clients, police, strangers etc.) [2]. We asked about lifetime violence experiences and those in the past 6 months. Following consultation with the sex work community, we also asked about (i) drugging and imprisonment (locked up somewhere against their will) by IPs and non-IPs, (ii) experiences of gang rape (ever and in the past 6 months), (iii) rape in the past 7 days, (iv) police arrest and imprisonment ever and in the past 6 months, and (v) if anyone had protected them from violence in the past 6 months.
Our main outcome variable was defined as physical and/or sexual violence in the past 6 months, as PSV is the category that had been the basis of most violence research [6,25]. We examined recent physical/sexual violence by anyone and by type of perpetrator (IP, non-IP). Exposure variables are shown in Figure 1, with further details provided in Table 1.

Laboratory Methods
Urine samples were collected to test for pregnancy, Chlamydia trachomatis (CT), and Neisseria Gonorrhoea (NG) infection. Blood was taken to test for Treponema pallidum (syphilis). HIV status was screened by rapid HIV tests, with positive tests confirmed using HIV DNA Genexpert. Self-collected vaginal swabs were used to test for Bacterial Vaginosis (BV; Gram's stain and Nugent scoring) and Trichomonas vaginalis (TV; OSOM Trichomonas Rapid Test; SEKISUI Diagnostics, Massachusetts, USA).
Our main outcome variable was defined as physical and/or sexual violence in the past 6 months, as PSV is the category that had been the basis of most violence research [6,25]. We examined recent physical/sexual violence by anyone and by type of perpetrator (IP, non-IP). Exposure variables are shown in Figure 1, with further details provided in Table 1.

Conceptual Framework
We developed a conceptual framework (Figure 1) based on a life course perspective [32] to explore the associations of recent physical or sexual violence with distal and proximate exposure variables, drawing on current theories about risk factors and drivers of violence [4,33]. Level 1 variables included ACEs and distal socio-demographic factors. Level 2 variables included distal sexual and reproductive health factors. Level 3 variables included (a) proximate socio-demographic and economic factors; (b) proximate sexual health and sex work characteristics; and (c) proximate mental health and social support factors.

Statistical Analyses
Data were double-entered using CSPro Software (United States Census Bureau, https://www.census.gov/data/software/cspro.html, (accessed on 1 June 2023)) and statistical analyses were conducted in STATA 16.1 (Stata Inc., College Station, TX, USA). We used a hierarchical modelling approach to build multivariable models for each outcome (recent physical/sexual violence by (i) any perpetrator; (ii) an intimate partner; and (iii) a non-intimate partner). Associations were estimated using odds ratios (OR), with p-values obtained using a joint hypothesis via the adjusted Wald test (to allow for sampling weights). Tests for trends were conducted for ordered categorical variables included in the final models. All models were adjusted for age and clinic as a priori defined variables. Level 1 variables associated with recent physical or sexual violence (p-value < 0.1) in univariate analyses were included in an initial multivariable logistic regression model. Variables were retained in a core Level 1 model (Model 1) if independently associated with any of the three violence outcomes (p-value < 0.1). Next, Level 2 variables were examined and adjusted for the core Level 1 variables and were retained if independently associated (p-value < 0.1) with any of the three violent outcomes (Model 2). Similarly, Models 3a-3c were fitted with Level 3a-3c variables, respectively, adjusting for core Level 1 and Level 2 variables. Missing data was reported if >5% of observations were missing.

Sample Demographics and Sex Work Characteristics
Of 1200 sampled women, 1039 met the eligibility criteria, and 1003 (96.5%) consented to participate in the study. Participant characteristics are shown in Table 2. 64.9% of study participants had experienced recent (past 6 months) physical or sexual violence from any perpetrator (intimate or non-intimate partner). The median age of participants was 32 years (range 18-45 years). Most were born in Kenya (98.7%), were Catholic or Protestant (91.3%), and had primary education or less (70.1%). The median number of Adverse Childhood Experiences (ACEs) reported was 6 (range 0-12). The median age at sexual debut was 16 years (range 0-26 years), with one-third of women (31.2%) reporting their sexual debut was not consensual. Just over half of participants (59.8%) reported a current intimate partner (IP) (defined as a lover or boyfriend who does not pay for sex), although only 6.8% were living with a male partner. The most common places for soliciting clients were bars, clubs, or lodges (61.5%) or on the streets (30.0%), and the median number of clients in the previous week was 3 (range 0-70). HIV prevalence was 28.0%, but bacterial STI prevalence was relatively low (10.2%) (CT 5.7%; NG 2.6%; Syphilis 2.1%) ( Table 2).  [31] defined as "high" if participant's highest score was "high" for any substance, "moderate" if participant's highest score was "moderate" for any substance, and "low" if they scored "low" for all substances. Substances asked about included Cannabis, Cocaine, Amphetamine type stimulants, Inhalants, Sedatives or Sleeping Pills, Hallucinogens, and Opioids.

Violence Experience across the Life-Course
Reported ACEs were prevalent among study participants: 41.4% were orphaned, 12.0% had lived on the streets, 77.0% had experienced physical or sexual violence, and 89.9% had experienced war or community violence when they were a child (<18 years old). A small minority (6.3%) reported Female Genital Mutilation (FGM), and 11.6% reported being raped during their sexual debut (Table 3). There was substantial overlap between experiencing PSV, orphanhood, and street homelessness in childhood, with 84.84% reporting at least one of these events and 5.85% reporting all three ( Figure 2).

Violence experience across the life-course
Reported ACEs were prevalent among study participants: 41.4% were orphaned, 12.0% had lived on the streets, 77.0% had experienced physical or sexual violence, and 89.9% had experienced war or community violence when they were a child (<18 years old). A small minority (6.3%) reported Female Genital Mutilation (FGM), and 11.6% reported being raped during their sexual debut (Table 3). There was substantial overlap between experiencing PSV, orphanhood, and street homelessness in childhood, with 84.84% reporting at least one of these events and 5.85% reporting all three ( Figure 2).   Reports of recent violence experience were also common, with 64.9% experiencing PSV by any perpetrator, 31.2% of women experiencing PSV by an IP, and 55.7% experiencing PSV by someone other than an IP in the past 6 months (Table 3). In addition, participants reported being drugged or imprisoned by partners (6.9%) and non-partners (18.5%) and arrested by the police because they were sex workers (30.7%) in the past 6 months. A substantial minority (n = 61; 6.4%) reported being raped in the 7 days prior to the survey (Table 3). Of note, just over one-third of participants (35.9%) said that someone had protected them from violence in the past 6 months, including the police (15.3%), city askaris (3.2%), clients (14.1%) and other FSWs (5.9%) (data not shown).
When we explored experiences of violence polyvictimization during the past 6 months, we found substantial overlap between recent physical or sexual violence by intimate partners, non-intimate partners, and police arrest, with 71.52% of women reporting any of these events and 9.03% reporting all three (Figure 3). Importantly, we also found substantial overlap between recent intimate and non-IP PSV and PSV experiences in childhood, with 86.85% reporting one or more types of violence and 18.67% reporting all three (Figure 4).

Witnessing violence by children of FSWs
We next investigated the violence experiences of children of FSWs, as reported by

Witnessing Violence by Children of FSWs
We next investigated the violence experiences of children of FSWs, as reported by FSWs. Of the 911 participants who had children, 33.5% said their children had witnessed violence against them, with 18.1% witnessing violence against them in the past 6 months (Table 4). In addition, of the 55 participants who had children and who had been imprisoned in the past 6 months, 14.5% said that they had no one to look after their children during their imprisonment (data not shown).

Witnessing violence by children of FSWs
We next investigated the violence experiences of children of FSWs, as reported by FSWs. Of the 911 participants who had children, 33.5% said their children had witnessed violence against them, with 18.1% witnessing violence against them in the past 6 months (Table 4). In addition, of the 55 participants who had children and who had been

Associations with Recent Physical or Sexual Violence Experience by Any Perpetrator
We were interested to understand factors across the life course which were associated with recent PSV experience by any perpetrator, guided by our conceptual framework ( Figure 1). In Model 1 (Table 5), women with recent PSV experience by any perpetrator had a higher prevalence of ACEs (5-8 vs. 0-4 ACEs: AOR 3.01, 95% CI 2.23-4.07; 9-12 vs. 0-4 ACEs: AOR 7.92, 95% CI 4.93-12.74) and were more likely to be aged 25-34 years (rather than younger and older women) ( Table 5, Model 1). In Model 2, after adjusting for level 1 factors, there was strong evidence that recent PSV was associated with a forced sexual debut (AOR 1.97, 95% CI 1. 18-3.29). In Model 3a, after adjusting for level 1 and level 2 factors, participants with recent PSV experience by any perpetrator were more likely to currently have an IP (AOR 1.67, 95% CI 1.25-2.23), less likely to have an additional income to sex work (AOR 0.65, 95% CI 0.48-0.87) and were more likely to report recent hunger in the past 7 days (AOR 1.39, 95% CI 1.01-1.92), compared to women with no recent PSV experience. There was some evidence that having four or more dependents was also associated with an increased risk of recent violence (AOR 1.52, 95% CI 0.98-2.34) ( Table 5, Model 3a). When we examined associations with proximate sexual health and sex work characteristics (Model 3b), after adjusting for level 1 and level 2 factors, recent police arrest (AOR 2.40, 95% CI 1.71-3.39) and not using a condom at last sex (AOR 0.68, 95% CI 0.48-0.98) were associated with recent PSV by any perpetrator (Table 5, Model 3b). When we examined associations with proximate mental health and social support factors (Model 3c), after adjusting for level 1 and level 2 factors, we found that high alcohol risk score (AOR 3.34, 95% CI 1.74-6.42) and CBO membership (AOR 1.91, 95% CI 1.14-3.21) were associated with recent PSV by any perpetrator (Table 5, Model 3c).

Associations with Recent Physical or Sexual Violence Experience by an Intimate Partner
We next examined associations with recent PSV experience by an IP, with some similar findings to recent PSV by any perpetrator (Supplementary Materials: Table S1). Thus, recent PSV experience by an IP was associated with a high number of ACEs (Model 1), having an intimate partner (Model 3a), not using a condom at last sex and recent police arrest (Model 3b), and a moderate or high alcohol use score (Model 3c). In addition, there was evidence that having an additional income from sex work (AOR 1.32, 95% CI 0.97-1.78) (Model 3a), non-use of PrEP, PEP or ARV (AOR 0.67, 95% CI 0.50-0.89) (Model 3b), and mild (AOR 2.22, 95% CI 1.58-3.12) or moderate/severe (AOR 1.40, 95% CI 0.95-2.04) depression or anxiety (Model 3c) were also associated with recent PSV experience by an IP-findings not seen when examining associations with recent PSV by any perpetrator. Conversely, there was no evidence that recent PSV experience by an IP was associated with the economic factors seen in the overall analyses (number of people dependent on her income, recent hunger, not having an additional income for sex work).

Associations with Recent Physical or Sexual Violence Experience by a Non-Intimate Partner
Associations with recent PSV by a non-intimate partner were broadly similar to recent PSV by any perpetrator (Supplementary Materials: Table S2), with street-based sex work additionally associated with increased risk of recent PSV by a non-intimate partner (AOR 1.48, 95% CI 1.08-2.01) (model 3b).

Discussion
We found a high prevalence of recent physical or sexual violence experiences among FSWs in Nairobi, with 64.9% of women reporting PSV by any perpetrator in the previous six months. Key risk factors during childhood and adolescence include a higher prevalence of ACEs and forced sexual debut. More proximate risk factors include having four or more dependents on her income, having an intimate partner, not having an additional income other than sex work, recent hunger, police arrest in the past 6 months, non-condom use at last sex, moderate or high alcohol use risk score, and belonging to an FSW CBO. We also found a high prevalence of PSV experience in childhood (77.0%) and a substantial overlap with recent intimate and non-IP violence in adulthood. This suggests an explicit need to understand violence against FSWs from a life course and intersectional perspective [34], as many FSWs are adversely situated economically, socially, and politically throughout their life course. Taken together, these findings provide strong evidence of polyvictimization, including sequential and concurrent violence experiences, findings which are supported by our qualitative research [35]. A key limitation of our study was the cross-sectional design, meaning the direction of causality cannot be ascertained; our findings and implications should be interpreted with this in mind.
The high prevalence of violence against FSWs that we found here has been described in multiple other settings. In a systematic review from 2014, the lifetime prevalence of workplace violence was 45-75%, and in the past year, workplace violence was 32-55% [14]. Previous studies with FSWs in Mombasa have reported similarly high violence levels, with 63% reporting lifetime physical violence and 44% reporting lifetime sexual violence [36]. Recent studies with FSWs in Russia, Rwanda, South Africa, and the USA have also found evidence of lifetime polyvictimization [21,[37][38][39]. However, we found only one other study with FSWs from SSA, which examined recent violence (rape in the past year) from a lifecourse perspective [40]. The authors found that childhood trauma, food insecurity, and harmful drinking were similarly associated with recent sexual violence and street-based sex work was similarly associated with recent rape by a non-IP.
Among FSWs, the three pillars of a comprehensive GBV response include (i) prevention, (ii) survivor support, and (iii) accountability/justice [41]. A recent systematic review identified 21 FSW violence interventions in 10 countries [41]. Evidence from India suggests that a combination of HIV and workplace violence interventions can successfully impact both HIV and violence outcomes, including police arrest [42][43][44]. Mathematical modelling suggests that reducing violence against FSWs in Kenya (and Ukraine) would also significantly reduce HIV infections among FSWs and the general population [45]. SWOP introduced violence interventions and response mechanisms and reporting as part of its programming in Nairobi in 2014. In line with the national government violence prevention and response strategy for key populations, this included both violence prevention and violence response interventions [46]. Interestingly, proximal protective factors for reduced violence risk among FSWs in our study include financial (reduced number of people she supports on her income, having an additional income to sex work), workplace (not selling sex on the street), and individual (not having an intimate partner); findings supported by our qualitative research and also by FSW studies elsewhere [35]. However, there remains a paucity of evidence regarding effective combination interventions for IPV (i.e., spousal/partner violence) and HIV among key populations, despite the fact that key populations of women bear a disproportionate burden of both IPV and HIV.
It is well recognised that witnessing or experiencing violence in childhood is strongly associated with IPV in adulthood among all women [4][5][6]; our study adds to the literature on FSWs, which finds that ACEs are also associated with increased risk of recent PSV among FSWs, both by intimate partners and by others. Importantly, one-third of FSWs said their children had witnessed violence against them, providing evidence of the inter-generational transfer of violence experience. The WHO INSPIRE technical package has been developed to assist countries in developing programmes to support children exposed to trauma and to prevent violence against children [47]. Economic strengthening for families could also help reduce the risk of ACEs [48,49], as well as reduce the risk of subsequent entry to sex work for economic survival.
The strengths of our study include a large, random sample of FSWs from across Nairobi, the use of validated tools to measure key variables, and the use of biological samples to measure HIV and STIs. In addition, our qualitative interviews and public engagement workshops with study participants and peer educators helped support and interpret our study findings. As noted above, a key limitation was the cross-sectional design; longitudinal data will become available for this study shortly, although long-term cohort studies would be needed to causally associate events in childhood and adolescence with violence experienced during adulthood. Our sampling methodology missed an estimated 10,600 FSWs who are not registered at a SWOP clinic (although they may be registered at other non-SWOP services). These women may be more vulnerable and at increased risk of the study outcome and key exposures (such as harmful drinking and younger age), leading to an underestimation of violence prevalence. However, creating a sampling frame and randomly selecting from all FSWs working in Nairobi was prohibitively expensive for our study.

Conclusions
Levels of violence experienced by FSWs in Nairobi, Kenya, are extremely high and occur throughout their life course. This warrants interventions to both prevent violence-especially at pivotal periods in the life course (childhood, adolescence, first relationships)-as well as programming to support FSWs (and their children) who experience violence [50]. Preventing violence against FSWs would also help protect their children from violence exposure and thus help break the cycle of violence across the life course. Decriminalisation of sex work in Kenya and other countries and addressing violence perpetrated by men would be key steps in supporting this transition [51][52][53].
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/ijerph20116046/s1, Table S1: Multivariable logistic regression-associations with recent physical or sexual violence by an Intimate Partner; Table S2: Multivariable logistic regression-associations with recent physical or sexual violence by a non-Intimate Partner.  Informed Consent Statement: Written informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data that support the findings of this study will be available on request from the corresponding author from June 2023 (two years after the study data collection is completed). The data are not publicly available due to privacy or ethical restrictions.