Pettifor, Audrey E. PhD*; Turner, Abigail Norris MSPH*; Van Damme, Kathleen MD†‡; Hatzell-Hoke, Theresa PhD§; Rasamindrakotroka, Andry MD‡; Nasution, Marlina D. PhD§; Behets, Frieda PhD*†
From the Departments of *Epidemiology and †Medicine, University of North Carolina, Chapel Hill, Chapel Hill, North Carolina; ‡UNC-MAD, Antananarivo, Madagascar; and §Family Health International, Durham, North Carolina
Partial funding for this study was provided by the U.S. Agency for International Development.
Correspondence: Audrey E. Pettifor, PhD, Department of Epidemiology, School of Public Health, McGavran-Greenberg CB#7435, University of North Carolina, Chapel Hill, Chapel Hill, NC 27599-7435. E-mail: email@example.com.
Received for publication May 14, 2006, and accepted October 2, 2006.
Goal: To examine and compare young sex workers’ risk of acquisition of sexually transmitted infections (STIs) with the risk among their older counterparts.
Objective: To evaluate the effect of young age—16–19 years vs. 20 years and older—on risk of incident infection with Neisseria gonorrhoeae (GC) or Chlamydia trachomatis (CT).
Study Design: A randomized controlled trial of 1,000 sex workers in Madagascar was conducted. STI testing was conducted at baseline, 6, 12, and 18 months.
Results: About 13% of the cohort (n = 134) was composed of young women aged 16–19 years. STI incidence rates in this group over the 18-month study period were high: 51.9/100 woman-years (WY) for GC and 47.4/100 WY for CT compared to 27.4/100 WY and 19.1/100 WY for sex workers over age 20, respectively. In multivariable models, young sex workers were at significantly higher STI risk compared with their older peers: The adjusted risk ratio (aRR) for GC comparing younger to older women was 1.50 (95% confidence interval (CI): 1.20, 1.88); for CT, the aRR was 1.72 (95% CI: 1.35, 2.19) and for GC or CT combined, the aRR was 1.42 (95% CI: 1.22, 1.66).
Conclusions: This exploratory analysis suggests that additional research is warranted to identify effective and acceptable prevention strategies that benefit young women, and interventions already proven effective among adolescents should be given high priority for scale-up.
GLOBALLY, SEXUALLY ACTIVE ADOLESCENTS HAVE the highest rates of sexually transmitted infections (STIs) of any age group,1 and young women face increased risk of cervical STIs in particular.2,3 However, researchers of STI prevention trials may be disinclined to include younger women because of human subjects restrictions.4 Consequently, information on STI incidence and the effectiveness of STI prevention interventions among adolescents are largely lacking, particularly in developing country settings.
In many parts of sub-Saharan Africa, sex workers are at high risk for STIs,5–9 and our earlier work suggests that young sex workers in particular are at even greater risk of infection.10,11 Using data generated by a randomized, controlled trial (RCT) among sex workers in Madagascar, we conducted this analysis to better understand the association between young age and risk of infection with Neisseria gonorrhoeae (GC) or Chlamydia trachomatis (CT).
The methods and main results of the RCT have been reported elsewhere.11 Briefly, the trial explored the effect of enhanced peer education-plus-clinician counseling compared with peer education only to promote male and female condom use and to reduce both sexual risk behaviors and biologically confirmed STI acquisition. From 2001 through 2003, 1,000 self-identified sex workers in Antananarivo and Tamatave, Madagascar, were followed bimonthly for 18 months. Eligible women were self-identified sex workers, not pregnant, and at least 16 years old. They were contacted in their community by peer counselors and invited to 2 public dispensaries (one in each city) for STI care and possible study participation. Peer counselors also assisted in tracking enrolled participants and accompanying them to bimonthly visits. Women received free STI care and compensation for time and transportation costs (approximately $3.25 U.S. per visit).
The RCT was approved by ethics committees at the Laboratoire National de Référence (LNR) VIH/SIDA in Madagascar and Family Health International in North Carolina. Because sex workers under age 18 rarely live with parents and may freely seek care at STI clinics, participating adolescents provided informed consent, but additional parental consent was not required.
GC and CT testing was conducted on urine specimens at baseline, 6, 12, and 18 months using ligase chain reaction assays (Abbott LCx Probe System; Abbott Laboratories, Abbott Park, IL). Because HIV rates have remained low in Madagascar,12 this RCT did not assess infection with HIV. Participants received presumptive treatment for GC and CT at enrollment and at 6-, 12-, and 18-month visits.
In the main RCT, the age distribution was similar across intervention groups and among sites, and we detected no modification of the intervention effect by age (data not shown). For this analysis, we examined age continuously and by various categorizations. Because we hypothesized that STI risk would be highest for women under age 20 resulting from physiological and behavioral factors, we focused our analyses on the youngest age group. In addition, we limited the outcomes in our evaluation to the cervical infections GC and CT.
Differences in proportions were assessed using 2-sided P values by Pearson’s χ2 tests for categorical data and t tests for continuous data.13 For multivariate analyses, we used binomial regression models with generalized estimating equations (GEE) to compute the 6-month risk of incident STI by age group.14,15 Three separate models were run to evaluate risk of GC, CT, and GC/CT combined. Effect modification by site and randomization arm was examined using Mantel-Haenszel tests for homogeneity. Backward elimination was used to build each model. We began with fully adjusted models containing dichotomized age at enrollment (main exposure: 16–19 years vs. 20–64 years), site, proportion of sex acts protected by male or female condoms with clients in the past month, total clients in the past week, randomization group, and previous GC or CT infection. Site and randomization group were retained in all models because of a priori hypotheses about their importance; other covariates were retained only if removal from any model resulted in ≥10% change in the association between age and incident STI.
All analyses were performed using SAS (version 9.1; SAS Institute, Cary, NC).
Of 1,000 enrolled participants, 901 attended the 6-month visit, 863 returned at 12 months, and 818 at 18 months (81.8% retention over 18 months). At baseline, participants’ median age was 27 (range, 16–64 years), and adolescents (16–19 years) comprised 13.4% (n = 134) of the total sample (Table 1). We examined STI prevalence and incidence by 4-year age categorizations (Table 2). Because the youngest women had substantially higher STI rates than any other age group at almost every time point, and because we were particularly interested in characterizing differences in STI risk between adolescent and adult sex workers, we subsequently collapsed all women over age 20 into a single age group, including participants aged 20 to 64 years.
Adolescent women (aged 16–19) were recruited in nearly equal proportions at the 2 sites (12.4% of those enrolled in Tamatave and 14.4% of those enrolled in Antananarivo). The 2 randomization groups also had similar proportions of young women: adolescents comprised 12.8% of the peer education-plus-clinician counseling intervention group and 14.0% of the peer education only control group.
There were no significant differences at baseline between the 2 age groups in use of contraception (16.9% of young women vs. 16.4% of older women reported using a method to prevent pregnancy, P = 0.89), use of male condoms during last sex act with a client (53.6% of young women vs. 48.5% of older women, P = 0.27), having a condom break or tear during sex in the past month (19.8% of younger women vs. 20.2% of older women who reported having sex in the past month, P = 0.93), or in mean number of clients in the past week (7.4 clients for younger women compared with 7.0 for older women, P = 0.53). Although 16.0% of young women reported never using condoms in the past month compared with 8.9% of older women (P = 0.03), the mean number of unprotected acts with clients in the past month in the 2 age groups was more similar: 12.1 acts (standard deviation [SD]: 15.9) for young women and 10.3 acts (SD: 17.9) for older women (P = 0.24). Younger women were significantly more likely to report being raped or beaten in the past month than older women (8.2% vs. 3.0%, P <0.01).
At baseline, GC prevalence was 41.0% among 16 to 19 year olds and 19.5% among 20 to 64 year olds (P <0.01); CT prevalence was 29.9% among young women and 12.3% for older women (P <0.01). The incidence rate for GC over the full 18-month period was 51.9 per 100 woman-years (WY) among young women, substantially higher than GC incidence in older women (27.4 per 100 WY). CT incidence was 47.4 per 100 WY in young women compared with 19.1 per 100 WY in older women. Figure 1 outlines the prevalence and incidence of GC or CT combined by age group; young women had higher STI incidence at all time points.
The final GEE binomial regression models estimated the effect of young age on 6-month STI risk (Table 3). No heterogeneity of the age–STI association by study site or randomization group was observed. The final model adjusted for male or female condom use with clients in the past month and GC/CT status at the previous visit. Young women were at significantly increased STI risk compared with older women: the risk ratio (RR) for GC comparing younger women to older women was 1.50 (95% confidence interval [CI] = 1.20–1.88); for CT, the RR was 1.72 (95% CI = 1.35–2.19) and for GC or CT combined, the RR was 1.42 (95% CI = 1.22–1.66). The factor most strongly associated with incident GC or CT infection was testing positive for either GC or CT at the previous visit where testing was performed, 6 months earlier. Women who reported inconsistent consistent condom use with clients in the past month (condoms used in less than 100% but greater than 0% of reported sex acts) were also significantly more likely to experience GC or CT than women who did not report any unprotected sex acts.
After adjusting for differences in sexual behavior, younger women in this sex worker cohort were consistently and significantly at increased risk of GC and CT compared with older women. Biologic factors such as greater cervical ectopy or weaker immune response are likely to place young women at increased risk of infection compared with older women. Young women face a dangerous “double exposure”: sex work places them at increased risk of STI exposure, and once exposed, their immature reproductive tracts are physically more vulnerable to STI acquisition.1–3 Young women may also be more vulnerable to gender power imbalances than older women, potentially affecting their ability to negotiate condoms or refuse unwanted sex.16 The more frequent reports of rape and beatings and inconsistent condom use by younger women in this study support this gender-power hypothesis.
In addition to young age, a second strong predictor of GC/CT was testing positive for a cervical STI at the previous visit where testing was performed. Although all participants received therapy for GC and CT at enrollment and at 6, 12, and 18 months, reinfections may have occurred when women were exposed again to the same sexual networks or the partner who was the contact for their first infection. Although routine STI case management in Madagascar includes referral for partner treatment, GC and CT infections were diagnosed in a U.S. laboratory; thus, results were not available during the study, which precluded effective partner treatment. Future interventions should target sex workers’ partners to promote consistent condom use and STI treatment.
This analysis has several limitations. The RCT was not specifically designed to explore young women’s STI risk and consequently did not include assessments of gender-power factors, partner characteristics, or biologic mediators of STI risk (such as cervical ectopy) that could add depth and context to our observed findings. In addition, sexual behavior data were self-reported and therefore subject to bias, including social desirability. These shortcomings are countered by strengths of the RCT design: the longitudinal design allowed evaluation of incident infections. In addition, adolescent women were included in a sex worker intervention. Data on GC/CT incidence in young people from countries in sub-Saharan Africa are lacking,3 and this analysis contributes to that sparse literature.
Our findings highlight the tremendous STI risk among adolescents who practice sex work in Madagascar. Primary prevention should target parents, teens, and community leaders to promote increased education and employment opportunities for young women. Whenever possible, STI control interventions, including screening and treatment programs, should include these high-risk adolescent women in the spirit of harm reduction.17 This exploratory research suggests that new interventions for young women are necessary both to address the biologic vulnerabilities particular to adolescent women (for example, cervix-protecting methods such as diaphragms and microbicides) as well as their increased “contextual” risk (such as promoting increased gender power, motivating male partners to increase condom use, and avoiding coercive sex in young women). Additional research is warranted to identify effective and acceptable prevention strategies that benefit adolescent women, and interventions already proven effective among adolescents should be given high priority for scale-up.
1. Berman S, Hein K. Adolescents and STDs. In: Holmes K, Sparling P, Mardh P, et al, eds. Sexually Transmitted Diseases, 3rd ed. New York: McGraw-Hill, 1999:129–142.
2. Moench TR, Chipato T, Padian NS. Preventing disease by protecting the cervix: The unexplored promise of internal vaginal barrier devices. AIDS 2001; 15:1595–1602.
3. Sexually Transmitted Infections: Issues in Adolescent Health and Development. Geneva: World Health Organization, 2004.
4. Singh JA, Abdool Karim S, Abdool Karim Q, et al. Enrolling adolescents in research on HIV and other sensitive issues: Lessons from South Africa. PLoS Medicine 2006; 3:0984–0988.
5. Ghys PD, Diallo MO, Ettiegne-Traore V, et al. Effect of interventions to control sexually transmitted disease on the incidence of HIV infection in female sex workers. AIDS 2001; 15:1421–1431.
6. Plummer FA, Nagelkerke NJ, Moses S, et al. The importance of core groups in the epidemiology and control of HIV-1 infection. AIDS 1991; 5(suppl 1):S169–S176.
7. Connolly CA, Ramjee G, Sturm AW, et al. Incidence of sexually transmitted infections among HIV-positive sex workers in KwaZulu-Natal, South Africa. Sex Transm Dis 2002; 29:721–724.
8. Ramjee G, Karim SS, Sturm AW. Sexually transmitted infections among sex workers in KwaZulu-Natal, South Africa. Sex Transm Dis 1998; 25:346–349.
9. Ryan KA, Zekeng L, Roddy RE, et al. Prevalence and prediction of sexually transmitted diseases among sex workers in Cameroon. Int J STD AIDS 1998; 9:403–407.
10. Behets FM, Van Damme K, Rasamindrakotroka A, et al. Socio-demographic and behavioural factors associated with high incidence of sexually transmitted infections in female sex workers in Madagascar following presumptive therapy. Sex Health 2005; 2:77–84.
11. Feldblum P, Hatzell T, Van Damme K, et al. Results of a randomized trial of male condom promotion among Madagascar sex workers. Sexually Transmitted Infect. 2005; 81:166–172.
12. Xueref S, Holianjavony J, Daniel R, et al. The absence of HIV seropositivity contrasts with a high prevalence of markers of sexually transmitted infections among registered female sex workers in Toliary, Madagascar. Trop Med Int Health 2003; 8:60–66.
13. Selvin S. Statistical Analysis of Epidemological Data, 3rd ed. New York: Oxford University Press, 2004.
14. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73:13–22.
15. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: A generalized estimating equation approach. Biometrika 1988; 44:1049–1060.
16. Pettifor A, Measham D, Rees H, et al. Sexual power and HIV risk, South Africa. Emerg Infect Dis 2004; 10:1996–2004.
17. Rekart ML. Sex-work harm reduction. Lancet 2005; 366:2123–2134.