Brief Report: Adolescent Girls Who Sell Sex in Zimbabwe: HIV Risk, Behaviours, and Service Engagement

Abstract

Background: 

To reduce HIV incidence among adolescent girls who sell sex (AGSS) in Zimbabwe, we need to better understand how vulnerabilities intersect with HIV infection and how those living with HIV engage in care.

Methods: 

In 2017, we conducted social mapping in 4 locations in Zimbabwe and recruited girls aged 16–19 years who sell sex, using respondent-driven sampling or census sampling methods. Participants completed a questionnaire and provided finger prick blood samples for HIV antibody testing.

Results: 

Of 605 AGSS recruited, 74.4% considered themselves sex workers, 24.4% reported experiencing violence in the past year, 91.7% were not in school, and 83.8% had less than a complete secondary education. Prevalence of HIV increased steeply from 2.1% among those aged 16 years to 26.9% among those aged 19 years; overall, 20.2% of AGSS were HIV-positive. In the multivariate analysis, age, education, marital status, and violence from a client were associated with HIV. Among the 605 AGSS, 86.3% had ever tested for HIV, with 64.1% having tested in the past 6 months. Among AGSS living with HIV, half (50.8%) were aware of their status, among whom 83.9% reported taking antiretroviral therapy.

Conclusion: 

The steep rise in HIV prevalence among those aged between 16 and 19 years suggests the window to engage with AGSS before HIV acquisition is short. To accelerate reductions in incidence among AGSS, intensified combination prevention strategies that address structural factors and tailor services to the needs of AGSS are required, particularly ensuring girls enroll and remain in school.

INTRODUCTION

In Zimbabwe, in 2020, HIV prevalence among adolescent girls (3.8%) was almost twice that among adolescent boys (2.1%).¹ This difference is driven by a stark disparity in incidence. In 2020, among persons aged 15–24 years, estimated incidence of HIV among female individuals was 0.76% compared to 0.08% among male individuals.² This difference is likely driven by a combination of biological, behavioral, and structural factors.^3–5 Adolescent girls who sell sex (AGSS) are at particularly high risk because of high partner numbers, often with older men who are more likely to be HIV-positive, poorly developed condom negotiation skills, and extreme economic vulnerability.⁶ In addition to being highly stigmatized, selling sex as an adolescent is criminalized (both for selling sex and selling sex under the legal age of consent), making AGSS wary of accessing health services and fearing referral to police and child protection services.⁷ We explore how vulnerabilities among younger AGSS intersect with HIV infection and how these adolescent girls, when diagnosed with HIV, engage in HIV care.

METHODS

Sampling and Recruitment

Four sites, including Zimbabwe's 2 largest cities, a rural farming/mining community, and a border town, were chosen to represent different types of communities where sex is sold. Adolescent girls were eligible if they exchanged sex for money in the past 30 days, were aged 16–19 years, and were living or working in 1 of the 4 study areas. In August 2017, we deployed rapid social mapping to identify sex work locations and typologies in each site. We listed all sex work venues and assessed AGSS social networks as to whether they were willing to invite their peers.

In the 2 cities, we used respondent-driven sampling (RDS) by purposively selecting seeds based on sex work typologies emerging from the mapping. Each seed completed a questionnaire, had a finger prick blood sample collected for HIV antibody testing, and was given 2 uniquely identified coupons. The seeds each recruited 2 AGSS meeting the inclusion criteria of the study; those who consented to participate were in turn given 2 coupons to recruit 2 peers until we reached 6 recruitment waves. Each recruiter was provided with US$5 to cover costs of participation and an additional $2 for each eligible AGSS recruited. In the border town and rural community, the census sampling method was used to recruit all AGSS identified in venues listed during mapping. The study team visited each site several times, particularly during peak hours such as evenings and weekends, to maximize recruitment.

Data Collection

An interviewer-administered questionnaire was delivered anonymously to all participants collecting sociodemographic data, information on selling sex risk of common mental disorders (CMDs), HIV status and risk perceptions, HIV-related service use, and structural factors known to heighten risk. The questionnaire was administered in participants' language (Shona or Ndebele). Data were collected directly in a computer-assisted survey instrument (Questionnaire Development System, Nova Research Company, Silver Spring, MD). Information on CMD was collected using a 14-item Shona Symptom Questionnaire; a combined score of ≥9 indicated risk of CMD, with a score of ≥11 indicating risk of severe CMD.⁸

Laboratory Procedures

Each participant had a finger prick blood sample taken for rapid HIV testing. Pre-HIV and post-HIV counselling were offered. Samples were tested according to the Zimbabwe National HIV testing algorithm.

Statistical Analysis

For the 2 RDS sites, we examined recruitment trees and assessed convergence of the HIV prevalence estimate and homophily patterns in recruitment. Survey design was accounted for by weighting observations using the RDS-II approach.^9,10 Purposely selected seed participants were dropped, and remaining participants' responses were weighted using the inverse of participant-reported network size with confidence intervals calculated using Taylor linearization.¹¹ The variable used for network size was the final response to 3 questions: (1) how many young women you know personally have sex with men in exchange for material support; (2) how many of these young women whom you know personally would you consider recruiting to this study; and (3) how many of them are aged 16–19 years. For young women with missing responses to the network size question, the value was replaced with the mean inverse network size by site. The 2 census sampling method sites were assigned a weight equal to 1, representing the probability of being selected in each census site.

Analyses were conducted in Stata version 15 (Stata Corp, College Station, TX) using the survey commands with seeds dropped and inverse degree probability weights to obtain the weighted estimates. In our risk factor analysis, variables found to be associated with HIV prevalence in the univariate analysis (P ≤ 0.05) were included in a multivariate logistic regression model, where a fixed term for the site was applied. The mapping protocol and data collection were approved by the Medical Research Council of Zimbabwe (MRCZ/A/2222).

RESULTS

Between September and November 2017, a total of 615 AGSS were recruited: 294 in city 1, 201 in city 2, 79 in town 1, and 41 in town 2. In city 1, each of the 4 seeds led to recruitment chains of 5 or 6 waves. In city 2, half of the 6 seeds led to recruitment chains of between 2 and 5 waves, with the remaining half recruiting 6 waves each. The cumulative RDS-II weighted estimate of the proportion of AGSS who were HIV-positive converged well by the end of recruitment in city 1 and reasonably well in city 2. There was little evidence that AGSS recruited each other preferentially according to HIV status. The 10 seeds were removed from further analyses.

Characteristics of AGSS and of Sex Work

Most of the AGSS (n = 605) were aged 18 or 19 years (61.3%), were not in school (91.7%), had less than a complete secondary education (83.8%), and had never been married (66%) (Table 1). Compared with the other 3 sites, participants in city 2, where 4 of 6 seeds were in school, were younger, had a greater level of current school attendance, and were more likely to have never married or had children (Table 1). In the past 12 months, 1 in 4 AGSS (24.4%) experienced sexual violence, with lower prevalence in the cities than the towns (Table 1). Among AGSS, 29.7% reported potential signs, symptoms, and complications of sexually transmitted infections, and 48% reported symptoms suggestive of being at risk of a CMD, with 25.5% reporting symptoms suggestive of severe CMD.

TABLE 1. - Characteristics of Adolescent Girls Who Sell sex

	Town 1		Town 2		City 1*		City 2*		Total*
	n	%	n	%	n	%RDS	n	%RDS	n	%RDS
Total	79	—	41	—	290	—	195	—	605	—
Age (yr)
16	5	6.3	10	24.4	48	17.6	32	21.8	95	16.3
17	17	21.5	6	14.6	58	20.4	61	31.7	142	22.4
18	28	35.4	14	34.2	76	27.4	57	26.1	175	30.5
19	29	36.7	11	26.8	108	34.5	45	20.3	193	30.8
In school
No	79	100.0	41	100.0	289	99.8	130	63.7	539	91.7
Yes	0	0.0	0	0.0	1	0.2	65	36.3	66	8.3
Education†
None/primary incomplete	27	34.2	16	39.0	35	35.5	27	17.3	105	31.5
Secondary incomplete	42	53.2	21	51.2	51	50.9	111	54.1	225	52.3
Complete secondary/higher	10	12.7	4	9.8	14	13.7	57	28.6	85	16.2
Marital status‡
Single/never married	37	46.8	22	53.7	193	69.3	183	94.6	435	66.0
Married/living together	0	0.0	2	4.9	2	0.4	1	0.6	5	1.0
Divorced/separated	42	53.2	17	41.5	93	29.6	11	4.8	163	32.8
Widowed	0	0.0	0	0.0	2	0.7	0	0.0	2	0.2
How many children do you have†
0	22	27.9	19	46.3	111	38.5	108	58.4	260	41.1
1–2	28	35.4	15	36.6	103	39.9	33	15.4	179	32.5
≥3	29	36.7	7	17.1	76	21.6	54	26.2	166	26.4
Where do you live?§
Relatives' home	12	15.2	6	14.6	80	28.7	161	85.9	259	35.5
Own home	44	55.7	25	61.0	139	45.9	12	4.9	220	41.8
Friends' home	5	6.3	8	19.5	48	16.3	16	6.8	77	11.7
Other	18	22.8	2	4.9	22	8.8	6	2.3	48	10.9
Do not wish to answer	0	0.0	0	0.0	1	0.4	0	0.0	1	0.1
Is having sex with men in exchange for material support the main way you obtain money or support yourself?
Yes	62	78.5	36	87.8	251	89.1	165	82.9	514	84.4
No	17	21.5	5	12.2	39	10.9	30	17.1	91	15.6
Do you consider yourself to be a sex worker?
Yes	51	64.6	28	68.3	244	87.3	127	62.4	450	72.0
No	28	35.4	13	31.7	46	12.7	66	36.8	153	27.9
I do not know	0	0.0	0	0.0	0	0.0	2	0.8	2	0.1
Sex work debut║
Late sex work debut	60	76.0	28	68.3	185	61.9	98	49.6	371	64.3
Early sex work debut	19	24.0	13	31.7	105	38.1	97	50.4	234	35.8
Have you experienced sexual violence in the past 12 months?
Yes	24	30.4	12	29.3	58	21.9	30	17.1	124	24.4
No	55	69.6	29	70.7	232	78.1	165	82.9	481	75.6
During the past 12 months, have you experienced pelvic pain, genital sores or itching, genital warts, or unusual vaginal discharge?
Yes	25	31.7	13	31.7	98	35.3	34	17.9	170	29.7
No	54	68.3	28	68.3	192	64.7	161	82.1	435	70.3
CMD
Not at risk of CMD	31	39.2	20	48.8	167	57.5	124	63.0	342	52.0
At risk of CMD	19	24.1	8	19.5	66	22.0	45	23.1	138	22.5
At risk of severe CMD	29	36.7	13	31.7	57	20.5	26	13.9	125	25.5
In the past 12 months, how often did you drink alcohol?
Never	27	34.2	8	19.5	97	37.2	35	17.2	167	29.0
Once a month	7	8.9	2	4.9	9	3.2	16	12.2	34	7.2
2–4 times a month	6	7.6	5	12.2	38	15.6	37	19.3	86	13.5
2–3 times a week	20	25.3	10	24.4	73	23.5	74	39.4	177	27.8
≥4 times a week	19	24.1	15	36.6	73	20.6	33	12.0	140	22.1
I do not wish to answer	0	0	1	2.4	0	n	0	0	1	0.4

^*Percentages are RDS-II weighted.

^†Percentages are as among those who responded to the question.

^‡In Zimbabwe, marriage may only indicate living with a partner.

^§Other includes dormitory room, no fixed abode, or other.

^║AGSS who had their first sex with a man in exchange for material support at the age of 15 years or younger assigned as early sex work debut.

Overall, 72% of AGSS considered themselves to be a sex worker (Table 1). Among AGSS aged 16–17 years, 70% considered themselves sex workers, 83% stated that having sex with men for material support was their main way to obtain money, and 63% reported that they were younger than 16 years when they started selling sex. Among AGSS aged 18–19 years, a greater proportion considered themselves sex workers (76%), and reported sex work was their main source of income (88%), whereas the proportion who reported starting sex work at an age younger than 16 years was lower (23%).

Risk Factors of HIV Among AGSS

One-fifth of AGSS (n = 122; 20.2%) were HIV-positive, with prevalence increasing steeply from 2.1% among those aged 16 years to 26.9% among those aged 19 years (Table 2). In the multivariable analysis, older age, lower level of educational attainment, being divorced or separated, and having experienced physical violence/abuse from a client were associated with prevalent HIV (Table 2).

TABLE 2. - HIV Prevalence and Risk Factors Associated With HIV

	Total	HIV-Positive		Univariate				Multivariate			Overall P
		N	%	OR	95% CI	P	P	OR	95% CI	P
Total	605	122		—	—	—	—	—	—	—	—
Age (yr)
16	95	2	2.1	0.07	0.01 to 0.36	0.001	<0.001	0.10	0.02 to 0.51	0.006	0.003
17	142	24	16.9	0.56	0.30 to 1.15	0.121	—	0.80	0.37 to 1.76	0.584	—
18	175	44	25.1	1.06	0.60 to 1.87	0.846	—	1.36	0.73 to 2.53	0.339	—
19*	193	52	26.9	1	—	—	—	1	—	—	—
Site
City 1*	290	67	23.1	1	—	—	<0.001	1	—	—	0.260
Town 1	79	30	38.0	1.98	1.13 to 3.45	0.018	—	1.34	0.72 to 2.53	0.358	—
Town 2	41	11	26.8	1.18	0.55 to 2.54	0.671	—	1.16	0.49 to 2.76	0.734	—
City 2	195	14	7.2	0.23	0.12 to 0.46	<0.001	—	0.60	0.26 to 1.42	0.248	—
In school
No*	539	119	22.1	1	—	—	—	1	—	—	0.095
Yes	66	3	4.5	0.12	0.03 to 0.45	0.002	—	0.60	0.14 to 2.69	0.507	—
Education
None/primary (incomplete and complete)	164	51	31.1	1.56	0.92 to 2.64	0.101	0.063	1.62	0.86 to 3.05	0.136	0.013
Secondary incomplete*	329	58	17.6	1	—	—	—	1			—
Secondary complete (O and A levels)	112	13	11.6	0.64	0.29 to 1.38	0.253	—	0.59	0.26 to 1.34	0.205	—
How many children do you have
0*	260	42	16.2	1	—	—	0.203	—	—	—	—
1–2	179	48	26.8	1.66	0.94 to 2.94	0.082	—	—	—	—	—
≥ 3	166	32	19.3	1.45	0.78 to 2.68	0.240	—	—	—	—	—
Marital status
Single/never married*	435	65	14.9	1	—	—	<0.001	1	—	—	0.012
Married/living together as married	5	0	0.0	1	—	—	—	1	—	—	—
Divorced/separated	163	56	34.4	2.99	1.80 to 4.97	<0.001	—	1.76	0.98 to 3.15	0.058	—
Widowed	2	1	50.0	3.5	0.21 to 57.47	0.38	—	3.87	0.31 to 48.79	0.295	—
Where do you live?†
Relatives' home*	259	27	10.4	1	—	—	<0.001	1	—	—	0.208
Own home	220	70	31.8	3.85	2.10 to 7.04	<0.001	—	1.77	0.83 to 3.78	0.137	—
Friends' home	77	12	15.6	1.61	0.64 to 4.08	0.312	—	1.11	0.41 to 3.03	0.840	—
Other	48	13	27.1	5.21	2.21 to 12.29	0.001	—	1.90	0.67 to 5.40	0.225	—
Do not wish to answer	1	0	0.0	—	—	—	—	—	—	—	—
Sex work debut
Late sex work debut*	371	84	22.6	1	—	—	—	1	—	—	0.633
Early sex work debut	234	38	16.2	0.55	0.33 to 0.93	0.027	—	1.03	0.53 to 1.99	0.933	—
Experience of physical violence/abuse from steady partner
No*	454	86	18.9	1	—	—	—	—	—	—	—
Yes	151	36	23.8	1.51	0.89 to 2.58	0.128	—	—	—	—	—
Experience of physical violence/abuse from client
No*	472	84	17.8	1	—	—	—	1	—	—	0.015
Yes	133	38	28.6	2.53	1.48 to 4.32	<0.001	—	1.97	1.05 to 3.67	0.034	—
Experience of sexual violence/abuse
No*	481	95	19.8	1	—	—	—	—	—	—	—
Yes	124	27	21.8	1.4	0.80 to 2.45	0.236	—	—	—	—	—
Experience of violence from police in the past 12 months
No*	555	112	20.2	1	—	—	—	—	—	—	—
Yes	50	10	20.0	1.23	0.52 to 2.92	0.637	—	—	—	—	—
Food insecure
No*	230	44	19.1	1	—	—	—	—	—	—	—
Yes	375	78	20.8	1.12	0.68 to 1.85	0.654	—	—	—	—	—
CMD
Not at risk of CMD*	342	66	19.3	1	—	—	0.318	—	—	—	—
At risk of CMD	138	25	18.1	1.13	0.61 to 2.10	0.685	—	—	—	—	—
At risk of severe CMD	125	31	24.8	1.57	0.88 to 2.81	0.130	—	—	—	—	—
In the past 12 months, how often did you drink alcohol?
Never	167	33	19.8	0.65	0.35 to 1.22	0.178	0.232	—	—	—	—
Once a month	34	7	20.6	0.66	0.23 to 1.90	0.44	—	—	—	—	—
2–4 times a month	86	12	14.0	0.41	0.17 to 1.01	0.054	—	—	—	—	—
2–3 times a week*	177	38	21.5	1	—	—	—	—	—	—	—
≥4 times a week	140	32	22.9	1.02	0.53 to 1.95	0.958	—	—	—	—	—
I do not wish to answer	1	0	0.0	1	—	—	—	—	—	—	—

Variables identified in univariate analysis as being associated with HIV prevalence included in multivariate model; statistically significant results (at 95% level) in italics.

^*Baseline groups are those with the highest number of responses.

^†Other includes dormitory room, no fixed abode, and other.

AGSS Engagement With HIV-Related Services

Among the 605 AGSS, 86.3% had ever tested for HIV and 64.1% in the past 6 months. Among the 483 AGSS who tested HIV-negative, these figures were 84.9% and 66.0%, respectively. A public clinic (53.7%) and a national Sisters with a Voice programme clinic for FSW (17.6%) were the 2 most common venues where HIV-negative AGSS received their last test. Of the 122 AGSS living with HIV, 62 (50.8%) were aware of their status, among whom 52 (83.9%) were taking antiretroviral therapy.

Risk Perception Among AGSS

When asked, “What do you think are the chances that you will become infected with HIV in the near future?,” more than half (58.2%) of the 483 HIV-negative AGSS reported no or small chance. When asked, “do you think you are able to protect yourself from getting HIV?,” most of them (90.9%) replied affirmatively. We found no association between perceived chance of becoming infected soon (P = 0.44) or ability to protect oneself from HIV in day-to-day life (P = 0.86) and having tested in the past 6 months.

DISCUSSION

Although there is a body of literature focusing on adult sex workers in sub-Saharan Africa,¹² or adolescents involved in transactional sex,¹³ there is little data on adolescent girls aged 16–19 years who rely on selling sex for their livelihoods. In our study, most of the AGSS considered themselves to be sex workers and reported selling sex as their main way to obtain money, most of them were not in school, almost half reported symptoms, suggesting they were at risk of CMD, and 1 in 4 reported experiencing sexual violence in the past year. In Zimbabwe's Multiple Indicator Survey in 2019, one-third of adolescent girls reported physical violence by their current or last husband or partner.¹⁴ Young South African women experiencing violence are at a greater risk of acquiring HIV.¹⁵

We reported a steep rise in HIV prevalence among those aged between 16 and 19 years, suggesting high HIV incidence, consistent with a recent study that reported incidence to be up to 7.1/100 person-years among 18- to 24-year-olds who sell sex in Zimbabwe.¹⁶ In addition to age, we found education (lower level of attainment), marital status (being divorced or separated), and having experienced physical violence/abuse from a client to be associated with being HIV-positive. Young women who sell sex in Zimbabwe have previously been reported as being less able to negotiate safe sex and more likely to have higher risk partners compared with older sex workers.¹⁷

Perhaps reflecting awareness of high HIV incidence, most of the AGSS reported having tested in the past 6 months. The high proportion of AGSS reporting having tested is difficult to interpret, considering our finding that almost half of AGSS living with HIV reported being unaware of their status. However, among those reported being HIV-positive, most of them reported being on treatment. This may reflect that AGSS are either not testing or misreporting their HIV-negative status due to a misunderstanding of terminology¹⁸ and/or due to social desirability bias.¹⁹ Alternatively, if both self-reported testing uptake and HIV-negative results are accurate, they suggest a high incidence of HIV in the preceding 6 months.

Despite reporting HIV risk, most of the HIV-negative AGSS believed they could protect themselves and that they were at little risk of infection in the near future, possibly because some AGSS interpret recent negative test results as confirming they are at little risk or that they do not want to admit their level of risk. Of note, our study was conducted before the widescale rollout of pre-exposure prophylaxis in Zimbabwe.

Our study has some limitations. Our sample in city 2 was different from those in the other sites in that 4 of our 6 seeds were in school, and we observed recruitment homophily by school status, with girls more likely recruiting others with a similar school status. Although it is likely this homophily reflects differences in sampling rather than different pathways into sex work, it may explain some of our findings. To inform the future sampling of AGSS, we need to learn more about how school attendance acts as a network determinant.

The steep rise in HIV prevalence among those aged between 16 and 19 years suggests the window to engage with AGSS before HIV acquisition is short. There is an urgent need to reach young entrants into sex work to reduce their myriad vulnerabilities.

Easy access to sexual health services to speed their engagement with prevention and care needs to be prioritized. We also need to scale up evidence-based safety net interventions to tackle the structural factors that place vulnerable young girls at risk of selling sex, including social protection schemes to keep girls in school and gender-transformative interventions to shift harmful gender norms and mitigate the risk of violence.