BMC Infectious Diseases BioMed Central

Background: Cambodia's 100% Condom-Use Programme (CUP), implemented nationally in 2001, requires brothel-based female sex workers (FSWs) to use condoms with all clients. In 2005, we conducted a sexually transmitted infection (STI) survey among FSWs. This paper presents the STI prevalence and related risk factors, and discusses prevalence trends in the context of the 100% CUP in Cambodia.


Background
Sexually transmitted infections (STIs) remain major causes of reproductive morbidity and mortality in developing countries; and their high prevalence facilitate HIV transmission [1]. STIs are usually concentrated in core groups such as female sex workers (FSWs) characterized by a high number of partners and poor healthcare seeking behavior [2]. The prevalence of gonorrhoea and/or chlamydia among FSWs was about 20% in 2003 in some Vietnamese provinces that share a border with Cambodia (with geographical variation ranging from 11.3% to 32.7%) [3], and 18% in 2004 in Laos (with provincial variation ranging from 12.3% to 21.9%) [4]. In Chiang Mai, Thailand, the 2004 prevalence of gonorrhoea and chlamydia among FSWs was 14% and 17% respectively [5], whereas the prevalence of gonorrhoea or chlamydia in China's Yunnan province was 24.6%.
In Cambodia, the first STI prevalence study conducted in 1996 found a very high prevalence of STIs among FSWs; 38.7% of these women had gonorrhoea and/or chlamydia and 13.8% had active syphilis [6]. Consequently, programmes focusing on FSWs have been widely implemented. One such intervention is the 100% condom use programme (CUP), which was based on the model successfully implemented in Thailand in 1989. The 100% CUP is a multi-sectoral approach involving local authorities, health staff, police, brothel owners, sex workers, outreach and peer educators, with the aim of enforcing condom use in all brothels and sex-related establishments [7]. This programme encourages FSWs to use condoms consistently with clients and regularly attend STI check-ups at assigned clinics. The strategy is based on promotion of systematic use of condoms in all brothels and entertainment establishments, and referral to free-of-charge STI clinics by stakeholders including peer-educators, establishment managers, health staff, and the local police. A Condom Use Working Group, consisting of police, local authorities, and health staff is set up to monitor, register and refer FSWs to assigned STI clinics and enforce programme implementation [8]. A brothel would be temporarily closed for one month if women in that specific brothel were found STIs 3 consecutive times during the regular STI check-up at the assigned clinic, but women would not receive any penalties. This programme was piloted in 1998 in Sihanouk Ville, a province characterized by a high prevalence of HIV infection (52% in 1996) among brothel-based FSWs. A programme evaluation, conducted eighteen months after the implementation, found a substantial reduction in STI incidence at the clinic designated for use by FSWs. From 1998 to 2000, syphilis incidence declined from 9.0% to 1.8% and trichomoniasis incidence declined from 2.0% to 0.9%. Almost all (95%) FSWs attended the regular monthly check-up at the STI clinics, and more than 90% reported always using condoms with clients. Based on this success, the Prime Minister supported the 100% condom use policy for national use and the programme was implemented nationwide in 2001 [9].
Compared to HIV that is a life long infection, curable bacterial STIs are biological markers that are more likely to reflect recent risk behaviour. While high STI prevalence indicates frequent risky sexual practice and a poor provision or uptake of services, low STI prevalence reflects the improvement in provision of care services or change in risk behaviours [10]. Compared with the 1996 prevalence estimates, the 2001 Cambodia STI Surveillance Survey (SSS) results indicated a significant decline of major STIs among FSWs; gonorrhoea decreased from 24.0% to 14.2%; chlamydia from 23.3% to 12.1%; and syphilis from 14.0% to 5.7% [11]. The high turnover of sex workers observed in the 2003 Behavioural Surveillance Survey (BSS) (i.e., 50% of FSWs reported that they had been selling sex for less than one year) [12] is a major concern because it renews the pool of FSWs who have not yet been exposed to sexual health promotion messages or monthly STI exams. In this paper, we examine the relationship between duration of sex work and STI prevalence among FSWs. We then discuss the trends in prevalence of STIs from 1996 to 2005 with regards to the effectiveness of the 100% CUP in Cambodia.

Data collection
This report is based on data from Cambodia's national SSS 2005 conducted among brothel-based FSWs from March to August 2005 in the 8 capital cities of the following provinces: Phnom Penh, Kampong Cham, Prey Veng, Battambang, Banteay Meanchey, Siem Reap, Koh Kong and Sihanouk Ville. Brothel-based FSWs were defined as women mainly working at the brothels in the red light areas who have no other employment other than selling sex to clients. Prior to data collection, the National Center for HIV/AIDS, STI and Dermatology (NCHADS) and Provincial AIDS Offices together with local NGOs conducted mappings of all known brothels located in all survey provinces, which served as sampling frames. All women working in selected establishments and not menstruating at time of interview were eligible. Participants were selected through two-stage cluster sampling. At the first stage, brothels were selected with equal probability. At the second stage all women working in selected brothels were invited to visit the nearby local STI clinics or a mobile clinic established in the vicinity. Women who provided informed consent were interviewed by trained staff about socio-demographic information, sexual risk behaviours, self-perception of risk for STIs, STI history and STI-related health seeking behaviours. Nurses collected venous blood (7 ml) and provided detailed instruction for women to self-collect vaginal swabs.

Specimen storage and transport, and laboratory method and quality control
Whole blood was centrifuged, aliquoted and stored at 4°C at each survey site before being transported to Phnom Penh within 48 hours. All specimens were stored at 4-6°C at the National Clinic for Dermatology and STD (NCDS) laboratory until being tested. Swabs were tested within 4 days of collection. Serum were tested for syphilis using Determine Syphilis TP (Abbott Diagnostics) which is highly sensitive (92.3%) and specific (100%) [13]. This test was done in the field to provide result on-the-spot for offering treatment for women only. Each participant whose syphilis test was positive was offered treatment according to national guidelines. Moreover, all consenting FSWs were offered singledose presumptive treatment for gonorrhea and chlamydia (medication and dosage consistent with national guidelines). Swabs were tested for Neisseria gonorrhoeae (NG) and Chlamydia trachomatis (CT) by the NCDS laboratory using BD Probe Tec™ strand displacement amplification assay. Serum specimens were tested for syphilis with rapid plasma reagin (RPR) and RPR-reactive specimens were confirmed by Treponema pallidum Particle Agglutination (TPPA). It meant that syphilis was defined as those who were both RPR and TPPA positive. Laboratory methods for identifying NG and CT and confirming RPR-reactive specimens were different from those used in the 1996 and 2001 surveys. CT and NG were tested using Ligase Chain Reaction in 1999 whereas Polymerase Chain Reaction was used in 2001. In the two previous rounds of survey, syphilis was tested using RPR and TPHA. Finally all NG-and CT-positive and 10% of negative specimens were retested by real-time Polymerase Chain Reaction by US CDC (Atlanta), considered as the gold standard for comparison.

Statistical analysis
The data were coded and entered into EpiData 3 (Odense, Denmark) and analysis was performed using STATA 10 using the survey commands (Stata Corporation, Texas, USA). The sampling design is self-weighted within provinces [14], but data were weighted to account for differences in the reported numbers of FSWs between provinces. Analysis was performed using survey commands taking into account sampling weight, cluster effect and stratification by province. Descriptive statistics were used to describe variables in terms of frequency, mean and median. Prevalence of NG, CT, syphilis, and "any STI" were estimated with 95% confidence intervals that account for the complex survey design, including provincial sampling weights and primary sampling units or clusters (brothels). Stata software uses the cluster information to calculate design effects and adjusts the standard errors before conducting statistical tests. Socio-demographic characteristics and risk behaviors were cross-tabulated with the variable "any STI," which included any of the following: gonorrhoea, chlamydia, or syphilis. Chi-square test, univariate, and multivariate logistic regression were used to determine factors independently associated with STI prevalence. After completion of the univariate analysis, candidate variables were selected for inclusion in the final regression model based on having a p-value < 0.20 [15] in the univariate analysis or if they were thought to be associated with having an STI. Associations with a pvalue < 0.05 were considered statistically significant. Colinearity between variables in the final model was not observed. Differences in STI prevalence were examined based on comparison of data from this survey with epidemiological data from 1996 [6]

Socio-demographic, risk behaviours, risk self-perception and STI-related health seeking behaviours
Of 1081 FSWs invited to participate in SSS 2005, 1079 consented (99.8% acceptance). Blood was collected from 1079 participants, and 1063 provided vaginal swabs. Mean age of participants was 25 years (Median = 23 years) and 55.4% were younger than 25 years old. Mean years of schooling was 2.4 (Median = 2 years); 44% had never attended school, whereas 9.2% had attended secondary school (≥ 7 years). The women were highly mobile, i.e., about half of them reported living in the current provinces or cities for less than 12 months, 76% of whom had lived in at least two provinces in the past year.
The duration of sex work ranged from 1 month to 12 years, and 60% were "new FSWs" (having sold sex for 12 months or less). More than half of the women had experienced at least one abortion. Although no significant differences between new and longer-working FSWs were observed regarding report of abortion and places where women sought care for their last abortion, a larger proportion of new FSWs reported their last abortion in the past 6 months (63.2% vs. 50.1%) (table 1). Sweetheart relationships (defined in Cambodia as noncommercial, non-marital sexual relationship that possess a certain degree of affection and trust from at least one partner) in the past 12 months were reported by 58% of FSWs. Among these FSWs, 98% reported having sexual relations with their sweetheart in the past three months. The percent of FSWs who reported consistent condom use in the past month with sweethearts (25%) and casual partners (i.e., a non-paying partner other than the sweetheart with whom women neither are married nor living) (34%) was low. In contrast, 80% of FSWs reported always using condoms in the past week with clients. This high level of consistent condom use was achieved despite the fact that 67% of FSWs reported having been forced or convinced not to use a condom by clients in the past week. About one third of FSWs reported intercourse with clients during their menstruation in the past month. No significant differences between new and longer-working FSWs were found in terms of condom use with clients, sweethearts, casual partners, and number of clients per day.

Discussion
As revealed in the 2005 survey, women who were new to sex work contributed substantially to the sustained high prevalence of STIs, i.e., about twice the prevalence of those with longer sex-work experience. The 2005 data did not provide any behavioural explanations to these observed differences in STI prevalence by duration of sex work. Indeed, the number of clients per day, intercourse with clients during menstruation, STI-related treatment seeking behaviours, consistent condom use with clients, sweethearts or other non-paying sex partners and to what extent the FSWs had been forced or convinced not to use condoms by clients did not differ by duration of sex work. Among FSWs new to sex work, 59% were aged 15-24, whereas 50% of longer-working FSWs were aged 15-24. Because of their slightly younger age, FSWs new to sex work may have been more biologically vulnerable to *N = number of specimens tested; number varied because among FSWs for whom duration of sex work was known, 15 FSWs did not submit swabs and 2 swabs were not tested for gonorrhoea **CI = confidence interval **Any STI = number of women with at least one of the bacterial STIs  acquiring STIs than those who had been working longer [3].
To date, the association between STI prevalence and recent entry into the sex trade has not been studied in Cambodia. The association of being new to sex work and having an STI may suggest higher prevalence of high risk behaviour possibly related to the lack of exposure to prevention information, unawareness of STI services, limited access to outreach programmes, and less skill and limited experience in negotiating safer sex with clients. However, our analysis did not reveal any significant differences between new and longer-working FSWs in terms of sociodemographic characteristics and risk behaviours that might explain the significant difference in prevalence of STIs. Therefore, a biological difference between new FSWs and longer-working FSWs is one possible explanation. A similar finding was reported in a report from Vietnam in which the authors suggested that the higher prevalence of STIs among newer FSWs may be attributed to a lack of immunity to chlamydia compared with longer-working FSWs [3]. Other possibilities to consider are that new FSWs may have clients who are different from or at higher risk than clients of other FSWs and that new FSWs may have casual partners or sweethearts that are at higher risk. These issues should be addressed in future surveys or studies.
Reproductive health issues are serious concerns for FSWs as suggested by the large proportion who reported having an abortion in the past 6 months, especially for the newer FSWs. Abortions indicate unwanted pregnancy and unprotected sex which may be the result of less consistent use of condoms with sweethearts or casual partners. Providing sex services to clients during menses (~30%) is likely to put women at higher risk of acquiring STI/HIV infection due to vaginal fragility. A number of studies found an association between sex during menses and increased risk of gonorrhoea and chlamydia [16,17]. Therefore, outreach and peer education programmes for FSWs should address all these reproductive health issues. A low level of education appeared to be a risk factor for STIs. Women with more education may be in a better position to access educational information due to their ability to read and better understand preventive messages. However, many studies show no correlation between level of education and STIs among FSWs [3,18,19].
The majority (80%) of FSWs reported always using condoms in the past week with clients, which seems discordant with the fact that 67% of FSWs reported having been forced or convinced not to use a condom by clients in the past week. Reasons for these seemingly discordant results may be related to interpretation of the question. Some FSWs may have misinterpreted the question about whether a client forced them not to use condoms to mean, "Has a client tried to force you not to use condoms?" Another explanation may be that FSWs who had been forced not to use condoms may not have counted such involuntary episodes as a lapse in condom use and reported that they used condoms consistently in the past week.
Problems with this question were not revealed by pilot testing. This issue should be addressed by future surveys or studies. Although a small proportion of FSWs reported not always using condoms with clients, the fact that a large proportion had unprotected sex with their sweethearts may partially explain the sustained high STI prevalence in 2005. If sweethearts become infected and do not receive treatment they may serve as reservoirs for re-infection of FSWs who were successfully treated as part of the 100% CUP. Over the past 10 years, the proportion of FSWs in Cambodia reporting consistent condom use with sweethearts has never reached 60% [12] although repeated efforts have been made to address this issue in outreach and peer education programmes. However, it is unknown how well the training and outreach and peer education activities have been translated into action that may impact behaviour with non-paying partners. Concerned about the quality of the outreach and peer education programme, NCHADS, in collaboration with many stakeholders including the National AIDS Authority, Ministry of Women Affairs, and nongovernmental organizations, have recently revised the standard operating procedures for the outreach/peer education programme and 100% CUP for sex workers, in an effort to make it more sustainable and efficient and less subject to service disruption [20].  2005). Furthermore, the contradictory attitudes of the government toward sex work (e.g. sometimes endorsing programmes to make sex work safe, sometimes closing sex work establishments) make it difficult for sex workers, and brothel owners and managers to understand and trust the government's intentions and motives [21]. Finally, some brothel owners and FSWs may be uncomfortable collaborating honestly and participating in the 100% CUP because of the involvement of local police in the Condom Use Working Group. On average, about 60% of FSWs visited STI clinics each month in 2005, indicating poor coverage of the 100% CUP.
In addition to quality of interventions, mobility of sex worker and coverage of the programme are main concerns with regards to preventive interventions. Many studies have reported that FSWs often move from place to place in order to maximise income. As they age, some move to more rural settings where competition for clients is less [22,23]. As a result, they may have less access to the outreach/peer education programme, condoms, or STI care services that are mostly concentrated in provincial towns or urban settings. Therefore, coverage of the preventive programme should be considered when interpreting these data. FSWs also move from one type of sex work to another -i.e., from brothel-based to non-brothel-based and vice versa [12] -which disrupts the continuity of prevention programme exposure. This instability is aggravated by the frequent police closure of brothels, which leads women moving to other forms of sex work or to other provinces. For example, 25% of brothel-based FSWs reported previous employment as karaoke workers and 50% of beer promoters reported formerly working in brothels or as dancing girls or karaoke workers. Nonbrothel-based FSWs represent about 60% of the sex worker population in Cambodia [24] and because many deny engaging in sex work they have not been significantly covered as a group by the 100% CUP. A study in Battambang province in north-western Cambodia, found a high STI prevalence among beer promoters, suggesting limited effectiveness of the existing STI programme and limited targeted intervention for non-brothel-based FSWs [25]. Therefore, the effectiveness of the 100% CUP may be limited because of its low overall coverage of the sex industry as a result of targeting only brothel-based FSWs. The 100% CUP works well if implementation is rigorously conducted with high coverage and sufficient intensity as shown in Thailand in the late 1989 and in Sihanouk Ville, Cambodia in late 1998 [7,9].
Our last concern is about the quality of STI services in terms of appropriate diagnosis and treatment of cervical and vaginal infections commonly found among FSWs. Although health care services in developing countries are generally recognized to be poor, the quality of STI care is especially poor for FSWs, a stigmatized group which faces discrimination in many societies. A study in Abidjan, Ivory Coast, showed that the quality of STI services in health facilities visited by FSWs was poor; for example, knowledge among health personnel about the correct treatment of selected STIs was very low in most settings [26]. So far, little is known about the quality of the STI services in Cambodia. Therefore, immediate actions should be taken to evaluate the quality of services and perform a needs assessment. A refresher or in-service training should be provided for staff, steps to reduce attrition of staff should be implemented, technical support and supervision should be increased at the provincial level, and availability of drugs should be ensured.
A number of factors should be taken into account when interpreting comparisons of data by survey year. First, each of the three surveys employed different laboratory techniques and data collection methods. Because of variable test performance (sensitivity and specificity) different test methodologies may have a dramatic effect on the measured prevalence of STI pathogens [27]. In 1996, researchers used ligase chain reaction (LCR) to test urine specimens for CT and NG nucleic acid. Since this technique has a lower sensitivity than tests performed on vaginal swabs in subsequent studies, prevalence may have been underestimated in 1996. Subsequent STI surveys used the polymerase chain reaction (PCR) in 2001 and the strand displacement assay in 2005. Although the techniques were different, all the tests used amplification methods that target nucleic acid sequences and have comparable performance characteristics as reported by test manufacturers: high sensitivity (99%-100%) and specificity (98%-100%) [28]. In 2001, vaginal swabs were collected by medical staff, whereas swabs were self-collected in 2005. However, many studies have confirmed the similarity of the two methods in terms of sensitivity and specificity [29]. Second, varying sampling methods were used. While convenience sampling was used in 1996, the 2001 and 2005 surveys randomly selected participants using cluster sampling. Difference in sampling raises concern about comparing results from the populations sampled in 1996 with those from FSWS sampled in 2001 and 2005. On one hand, convenience sampling in 1996 might have resulted in overestimating the true prevalence since FSWs visiting STI clinics at that time-i.e., those with STI symptoms -may have been more likely to participate in the survey. On the other hand, using LCR to test urine (and not vaginal swabs) may have underestimated CT and NG prevalence in 1996. These biases are in opposite directions, and the magnitude and direction of the resulting bias are difficult to quantify. The combination STIs of short duration of infection (GC and CT) and long duration (TP) may affect the strength of association of covariates with having "any STI." However, given the low prevalence of syphilis, the relatively young age of the women and their short duration of sex work, this effect was probably minimal. Lastly, because most FSWs were familiar with the 100% CUP, social desirability bias may have resulted in FSWs underreporting the frequency of unprotected sex.

Conclusion
STI prevalence among new FSWs was higher than among longer-working FSWs. Because of the high turnover of FSWs, the prevention needs of new FSWs, which may be different than those of longer-working FSWS, should be ascertained and addressed. Despite the implementation of a nationwide 100% CUP, the prevalence of STIs among FSWs in 2005 was comparable to 2001 estimates. The large proportion of FSWs who reported having unprotected sex with non-commercial partners, limited coverage and weaknesses in implementation of the 100% CUP, in addition to questionable quality of STI care services are likely to have contributed to the sustained high prevalence of STIs among FSWs in Cambodia. The 100% CUP should be carefully evaluated, particularly in terms of human resource capacity, sustainable intensity, quality, and coverage. Moreover, the outreach and peer education programmes should include a subcomponent which specifically targets new FSWs in terms of providing preventive messages and support services.