Prevalence and risk factors of syphilis infection among female sex workers in Shenzhen, China: an observational study (2009–2012)

Authors


Abstract

Objective

To investigate prevalence and risk factors of syphilis infection among female sex workers (FSWs) in Shenzhen, China.

Methods

Observational study among (2009–2012) 1653 FSWs recruited by venue-based sampling using questionnaire-based interviews for socio-demographics, behaviours and syphilis testing results. Logistic regression was used to assess risk factors of syphilis infection.

Results

The overall syphilis prevalence was 4.7%, showing a slightly decreasing trend. Factors significantly associated with syphilis infection were inconsistent condom use (OR = 1.87, = 0.015), illicit drug use (OR = 5.45, P < 0.001) and older age in years (OR = 1.08, P < 0.001). Venues where FSWs were recruited and duration of commercial sex work were not significantly associated with syphilis infection (P > 0.05).

Conclusions

Syphilis is still common among FSWs in Shenzhen, China. Current comprehensive prevention programmes (e.g. condom promotion and peer education) should be continued to maintain and increase safe sexual practices and to reduce illicit drug use among FSWs. Expanding point-of-care syphilis screening programmes may be an important strategy for early diagnosis. We recommend timely and effective treatment programmes to be linked to such screening programmes.

Abstract

Objectif

Déterminer la prévalence et les facteurs de risque de l'infection à la syphilis chez les professionnelles du sexe (PS) à Shenzhen, en Chine.

Méthodes

Étude observationnelle de 2009 à 2012 chez 1653 PS recrutées par échantillonnage, basée sur le lieu, au moyen d'interviews par questionnaire pour les caractéristiques sociodémographiques, les comportements et les résultats des tests de la syphilis. La régression logistique a été utilisée pour évaluer les facteurs de risque de la syphilis.

Résultats

La prévalence globale de la syphilis était de 4.7%, montrant une tendance légèrement à la baisse. Les facteurs significativement associés à l'infection par la syphilis étaient l'utilisation irrégulière du préservatif (OR = 1.87; P = 0.015), l'utilisation de drogues (OR = 5.45; P < 0.001) et l’âge plus élevé (OR = 1.08; P < 0.001). Les lieux où les PS ont été recrutées et la durée du commerce du sexe n’étaient pas significativement associés à l'infection par la syphilis (> 0.05).

Conclusions

La syphilis est encore fréquente chez les PS à Shenzhen, en Chine. Les programmes complets actuels de prévention (par exemple, la promotion du préservatif et l’éducation par les pairs) devraient être poursuivis afin de maintenir et accroître des pratiques sexuelles sûres et de réduire la consommation de drogues chez les PS. L'expansion des programmes de dépistage de la syphilis sur les lieux des soins peut être une stratégie importante pour le diagnostic précoce. Nous recommandons que des programmes de traitement opportuns et efficaces soient liés à ces programmes de dépistage.

Abstract

Objetivo

Investigar la prevalencia y los factores de riesgo de la infección por sífilis entre trabajadoras sexuales femeninas (TSFs) en Shenzhen, China.

Métodos

Estudio observacional entre el 2009–2012 de 1653 TSFs reclutadas mediante un muestreo en el lugar de trabajo, utilizando entrevistas basadas en un cuestionario con datos sociodemográficos, de comportamientos y resultados de la prueba de sífilis. Se utilizó una regresión logística para evaluar los factores de riesgo de infección por sífilis.

Resultados

La prevalencia total de sífilis era del 4.7%, con una ligera tendencia a la disminución. Los factores significativamente asociados con la infección por sífilis eran el uso inconsistente del condón (OR = 1.87, = 0.015), el uso ilícito de drogas (OR = 5.45, P < 0.001) y una mayor edad (OR = 1.08, P < 0.001). Los lugares en los que se reclutaron las TSFs y la duración del trabajo sexual comercial no estaban asociados de forma significativa con la infección por sífilis (P > 0.05).

Conclusiones

La sífilis continúa siendo común entre las TSFs de Shenzhen, China. Los programas integrales de prevención (ej. promoción del uso del condón y educación de las TSFs por sus compañeras) deberían continuar para mantener y aumentar las prácticas sexuales seguras y reducir el uso ilícito de drogas entre las TSFs. Expandir lo programas de detección de sífilis en los lugares de atención podría ser una estrategia importante para el diagnóstico temprano. Recomendamos que los programas de tratamiento temprano y efectivos vayan unidos a estos programas de detección.

Introduction

Syphilis has made a remarkable comeback in China since the establishment of the ‘Open-Door Policy’ in 1980 (Lin et al. 2006; Tucker & Cohen 2011). While the rapid economic development has improved public health conditions, the resulting economic disparities, internal migration and re-establishment of the commercial sex industry have greatly facilitated the spread of syphilis and other sexually transmitted diseases (STDs) (Yang & Xia 2006; Poon et al. 2011). After the near eradication of syphilis in the 1970s, the prevalence of syphilis has increased rapidly and nowadays it ranks as the third most prevalent notifiable infectious disease in China, with about 400 000 new reported cases in 2011 (Tucker et al. 2010; Ministry of Health 2012). Congenital syphilis has also re-emerged, with very few cases in the 1990s but 139 reported cases per 100 000 live births in 2009 (Ministry of Health 2012).

Populations at high risk of unsafe sexual practices, such as female sex workers (FSWs), are of particular concern with regard to STDs. It is estimated that there are between one and four million FSWs in China, in both lower-end (i.e. on the street, hair salons and temporary small clubs) and higher-end venues (i.e. karaoke clubs, night clubs and hotels) (Pirkle et al. 2007; Hong & Li 2008; Poon et al. 2011). The prevalence of syphilis is alarmingly high among FSWs, especially in more economically developed provinces where the commercial sex industry is common (Chen et al. 2007; Tucker et al. 2010). The median syphilis prevalence is estimated to be approximately 7% among FSWs nationwide (Poon et al. 2011), with a prevalence as high as 14% in Guangdong province in 2008 (Yang et al. 2010), indicating that unsafe sexual practices are common among FSWs and that syphilis is spreading readily.

As the HIV epidemic in China continues to expand, especially through heterosexual transmission (WHO 2011), syphilis among FSWs is also of concern with regard to the HIV epidemic. Although the HIV prevalence remains relatively low and stable among FSWs (around 0.5%) (Poon et al. 2011; WHO 2011), the high prevalence of syphilis can facilitate the HIV epidemic by increasing about 1 to 5 times of the infectiousness and/or susceptibility through the cofactor effect (Hoare et al. 2012). Also, as a ‘bridging population’, FSWs can increase risks of transmitting HIV in the general population through their clients. Thus, tracking the prevalence and associated risk factors of syphilis infection among FSWs is necessary for improved understanding and control of the syphilis and HIV epidemic.

In China, sentinel surveillance surveys are conducted annually among high-risk populations, including FSWs. The sentinel surveillance system integrates serological surveillance of syphilis and behavioural surveillance of socio-demographic and behavioural characteristics of the targeted populations (Wang & Wang 2010). The surveillance system database provides continuous and comprehensive information for investigating predictors of syphilis infection. As a special economic zone in Guangdong province, Shenzhen, which has over 10 million inhabitants (Shenzhen Statistics 2012), is not only one of the fastest growing and modern cities, but also one of the areas hit hardest by a syphilis epidemic. There are between 60 000 and 100 000 FSWs in Shenzhen (Chen et al. 2008), and the prevalence of syphilis in this group was approximately 9% between 2001 and 2006 (Hong et al. 2009). Most previous studies of syphilis among FSWs in China have a relatively small number of participants and only report single prevalence figures (Ruan et al. 2006; Lu et al. 2009; Chen et al. 2012; Li et al. 2012a; Zhou et al. 2013). We know of no study that synthesised data over time and recorded trends of syphilis among FSWs. In this study, we present the surveillance results for the period from 2009 to 2012 and investigate the risk factors of syphilis infection among FSWs in Shenzhen, China.

Methods

Survey design and participants

Trained public health officials from Shenzhen Center for Disease Control and Prevention (Shenzhen CDC) conducted surveillance surveys with consistent recruiting procedures and questionnaires for the period 2009–2012. Participants were recruited by a venue-based sampling method in three steps as follows: categorising venues; selecting venues; and selecting FSWs. In the first step, all accessible venues were categorised into three subgroups: high-end establishments, low-end establishments and street-based venues, according to the national surveillance guidelines (China 2011). In general, guesthouses/hotels and nightclubs were categorised as high-end establishments; karaoke/dance halls and saunas were categorised as low-end establishments; hair salons/foot bathing shops, temporary sublets/roadside restaurants and streets were categorised as street venues. FSWs may go from one site to another, but usually within a certain subgroup of venues; mobility across subgroups is not common (Chen et al. 2012). In the second step, three venues were randomly selected within each subgroup. When gatekeepers of the selected venues refused to participate, we randomly selected and invited gatekeepers from the remaining venues. In the third step, a maximum of 25 participants were recruited from the selected venues with more than 25 eligible FSWs; from venues with ≤25 eligible FSWs, all were recruited. The second and third steps were repeated until at least 400 valid questionnaires were obtained and then the sampling process stopped. Eligibility criteria required that participants were ≥ 16-year-old (legal age of consent in China) female and reported having had commercial sex in the last 3 months. For validity of results, participants were required to have answered at least 80% of all the questions with logical consistency. To guarantee validity, trained staff explained the purpose of the survey at the venues and explained the questionnaire to FSWs if they were unclear about certain questions. Validity of questionnaires was assessed by trained staff at the venues, and once identified, invalid questionnaires were deleted. Because FSWs were guided by trained staff at venues, invalid questionnaires were rare (<1%). Verbal informed consent was obtained from all participants before the anonymous questionnaire interview and free syphilis testing. All participants were assured confidentiality. The survey protocols were reviewed and approved by the institutional review board of Shenzhen CDC.

Socio-demographic and behavioural measures

We used questionnaire-based interviews to obtain information about socio-demographic and behavioural information. Questionnaires used in our study were the national standard questionnaires used for HIV surveillance, and it took about 30 min to complete. Socio-demographic information was as follows: age, marital status, residency, education, duration of current commercial sex work and location of previous work. Behavioural information was as follows: condom use during the last intercourse, condom use in the last month and illicit drug use in the last year. For illicit drug use, we asked respondents to report if they had ever consumed or ever injected any illicit drugs (e.g. heroin, marijuana and methamphetamine). HIV-related knowledge was assessed from answers to eight core questions (China State Council AIDS Working Committee Office 2007). A high knowledge level was defined as having answered six or more questions correctly.

Serologic measures

After each questionnaire interview, a blood sample was collected for syphilis testing. In accordance with the Chinese national HIV sentinel surveillance guidelines (China CDC 2011), blood samples were first screened for treponemal antibodies using an enzyme-linked immunosorbent assay (ELISA, Wantai Biotechnical Company, Beijing, China). Specimens with positive ELISA were further tested by rapid plasma regain (RPR Rongsheng Biotechnical Company, Shanghai, China) or toluidine red unheated serum test (TRUST, Rongsheng Biotechnical Company, Shanghai, China) to determine positivity and titres of non-treponema-specific antibodies. Active syphilis was defined as ELISA positive and titres of non-treponema-specific antibodies ≥ 1:8 (China 2011; Gupte et al. 2011).

Data analysis

All data from valid questionnaires and laboratory results were double entered into Excel and double checked by two public health officials from Shenzhen CDC. Conflicts were solved by retrieving the original test results and surveys for these cases. We calculated annual prevalence of syphilis and plotted the prevalence figures derived from this study and a different surveillance database from Shenzhen Center for Chronic Disease Control (Shenzhen CCDC) against calendar year to assess any trend. In Shenzhen, surveillance surveys of syphilis prevalence among FSWs were conducted by Shenzhen CCDC from 2001 to 2010 and Shenzhen CDC since 2009. While Shenzhen CCDC recruited FSWs from re-education centres, we recruited FSWs from community venues. Hong et al. published data from Shenzhen CCDC of prevalences of syphilis for the years from 2001 to 2006 only (Hong et al. 2009). In this study, we synthesised all the available data from Shenzhen CCDC and Shenzhen CDC to assess the trend over the years covered by both databases. Data from our surveillance surveys were pooled to investigate risk factors of syphilis infection. As FSWs are highly mobile, and the duration of working as FSWs is generally short, the probability that the same FSWs are sampled repeatedly is small. We first performed univariate logistic regression analysis of syphilis infection. The variables attaining P < 0.20 significance in univariate analysis were included in the multivariate regression analysis. Only variables achieving P < 0.05 significance were retained in the final model using backward stepwise elimination. Age was used as a continuous variable in the univariate and multivariate logistic regression. Categories of age were only used to describe prevalence of syphilis in different age stratums. Impact of risk factors was expressed as odds ratios (OR) with 95% confidence interval (CI). All statistical analysis was performed using Stata software (version 12.0).

Results

A total of 1653 FSWs was enrolled into the consecutive surveillance surveys: 424 in 2009, 429 in 2010, 400 in 2011 and 400 in 2012. Refusal rates of gatekeepers ranged from 8% to 12% over the years. Approximately 5% of FSWs refused to participate in the study, due to the requirement of providing blood samples for testing of syphilis. The prevalence of syphilis infection was 4.5% in 2009, 6.8% in 2010, 5.3% in 2011 and 2.0% in 2012. Thus, the prevalence figures were almost half of those in 2001–2006 (Figure 1).

Figure 1.

Prevalence of syphilis in female sex workers plotted against years. Open symbols indicate the prevalence derived from surveillance data of Shenzhen Center for Chronic Disease Control and Prevention (Shenzhen CCDC) (Hong et al. 2009; Shenzhen CCDC 2011); filled symbols indicate the prevalence derived from data of Shenzhen Center for Disease Control and Prevention (Shenzhen CDC, our study). There was a severe SARS epidemic in Shenzhen between 2003 and 2004. Infected FSWs or those with more risky sexual behaviours may have migrated to other places, so that the prevalences in 2003 and 2004 were low. Comprehensive syphilis intervention refers to interventions for syphilis control and prevention (condom promotion and peer education) among female sex workers, which began in July, 2004 in Shenzhen (Shenzhen CCDC 2012).

Table 1 shows syphilis prevalence by socio-demographic and behavioural characteristics. In univariate analysis, the prevalence increased sharply with age (P < 0.001), with a prevalence of 1.4% among the age group younger than 20 years vs. 9.5% among the age group older than 30 years. The prevalence in FSWs who were married, cohabiting, separated or widowed was higher than that in those who never married (P = 0.016). The syphilis prevalence among persons that had worked as FSWs in Shenzhen for one or more years was more than double the rate among FSWs who had worked in Shenzhen for <1 year (6.9% vs. 3.1%, P < 0.001). FSWs who reported inconsistent condom use in the last month had a syphilis prevalence twice as high as those who reported consistent condom use (7.0% vs. 3.5%, P = 0.004). The higher prevalence among those with inconsistent condom use was particularly pronounced in the 25 to 29-year-old group but less so in other age groups (Figure 2). The syphilis prevalence among FSWs who reported illicit drug use in the last year was nearly five times higher than among those who reported no illicit drug use (19.4% vs. 4.0%, P < 0.001). The difference in syphilis prevalence between FSWs from higher-risk venues and lower-risk venues was below the cut-off level for inclusion in multivariate analysis (P = 0.16). In multivariate analysis, the remaining significant factors associated with syphilis prevalence were older age, inconsistent condom use in the past month and illicit drug use in the past year. Homogeneity testing by including age*condom use in the last month as an interaction term in the final multivariate analysis did not confirm effect modification (P = 0.20).

Table 1. Factors associated with syphilis infection among 1653 FSWs in Shenzhen, China, derived from univariate and multivariate logistic regression
Risk factorSample sizeSyphilis casesPrevalence (%)UnivariateMultivariate
Crude odds ratioP-valueOdds ratioP-value
  1. *High-end establishments include nightclubs and guesthouse/hotels; low-end establishments include saunas and karaoke/dance halls, street-based venues include hair salons/foot bathing shops, temporary sublets/roadside restaurants and streets.

  2. Duration of current sex work refers to the duration of working as a FSW currently in Shenzhen.

  3. The bold values indicate statistically significant (P < 0.05).

All1653774.7    
Venues*
High-end establishments418235.51.00   
Low-end establishments654213.20.570.068  
Street-based venues581335.71.030.92  
Age (categories, years)
<2020831.4    
21–24705223.1    
25–29372174.6    
 >30368359.5    
Age (continuous, in a year)   1.08 <0.001 1.08 <0.001
Marital status     0.016   
Never married998333.31.00   
Married421296.92.160.003  
Cohabiting146106.82.150.040  
Separated/widowed8456.01.850.21  
Missing400.0    
Residency
Guangdong province227114.81.00   
Non-Guangdong provinces1426664.60.950.89  
Education (years)
≤91072514.81.00   
>9581264.50.940.80  
Duration of current sex work (year)
≥1683476.91.00   
<1970303.10.43 <0.001   
Location of previous sex work    0.73  
Non-Guangdong provinces398215.31.00   
Guangdong province1189534.50.840.50  
No previous sex work5823.40.640.56  
Missing8112.5    
Level of HIV-related knowledge
Low458194.11.00   
High1195584.91.180.54  
Condom use during the last intercourse    0.41  
No248135.21.00   
Yes1194584.90.920.80  
Refuse to answer20363.00.550.24  
Missing800.0    
Condom use in the last month     0.009   
Consistent855303.51.00 1.00 
Inconsistent561397.01.870.0041.87 0.015
Refused to answer23083.50.990.981.330.51
Missing700.0    
Illicit drug use     <0.001   
No1564624.01.00 1.00 
Yes671319.45.83 <0.001 5.45 <0.001
Refused to answer2129.52.550.223.110.15
Missing100.0    
Figure 2.

Prevalence of syphilis (with 95% confidence interval) in female sex workers of different age groups by status of condom use in the last month. Open symbols indicate the prevalence of syphilis among those reporting consistent condom use in the last month; filled symbols indicate the prevalence of syphilis among those reporting inconsistent condom use in the last month.

Discussion

The overall prevalence of syphilis among FSWs from 2009 to 2012 was 4.7%, indicating that syphilis is common among FSWs in Shenzhen, but that its frequency has nearly halved compared with the years 2001–2006 (9%). To our knowledge, this is the first study that shows a decreased trend of syphilis among FSWs in China. It confirms risk factors identified in previous studies (Ruan et al. 2006; Lu et al. 2009; Chen et al. 2012; Li et al. 2012a; Zhou et al. 2013): older FSWs, those who used condoms inconsistently and those with a history of illicit drug use were at a higher risk of being infected.

The prevalence of syphilis in Shenzhen reported by our study is higher than many other areas in China, such as Shandong, where the reported prevalence among FSWs is approximately 3% (Liao et al. 2011, 2012). However, it is much lower than the reported prevalence figures by Shenzhen CCDC, but this can partly be explained by differences in the groups of FSWs included in the surveys. The data from Shenzhen CCDC were from FSWs recruited in re-education centres (Hong et al. 2009), while we recruited FSWs from community venues. In China, commercial sex is illegal and FSWs will be sent to a re-education centre when arrested by public security officers. These arrested FSWs may have engaged in more risky sexual behaviours, reflected in a higher prevalence of syphilis. However, if we compare the consistent surveillance surveys from Shenzhen CCDC, the overall mean prevalence prior to 2006 (9%) is still higher than that after 2006 (6%). Thus, it is likely that the prevalence in FSWs in Shenzhen indeed has declined. Comprehensive interventions for syphilis control and prevention among high-risk groups were initiated in July, 2004 in Shenzhen (Shenzhen CCDC 2012). These programmes, which included condom promotion and peer education, may have helped to relieve the burden of the syphilis epidemic among FSWs.

Regardless of the progress made in reducing the prevalence of syphilis in Shenzhen in the past years, the prevalence among subgroups of FSWs remains high, in particular among those over 30 years old (9%), irrespective of condom use status. As our testing method cannot differentiate between recent and old persistent infection, it is possible that part of the positive results among these older FSWs are actually the result of old infections that occurred several years ago, but are still active. The extremely high prevalence among FSWs over 30 years old and the probability that those infections actually occurred several years ago indicate that regular screening with early case-finding programmes should be advocated. As a marginalised population, FSWs have limited access to health services (Hong & Li 2008), apart from the fact that effective syphilis testing is often not available in many basic medical facilities in China (Yang et al. 2011). This situation suggests that priorities should be set to expand point-of-care syphilis screening programmes. Furthermore, it needs to be emphasised that timely and effective treatment programmes should be linked to such screening programmes. Case management also needs to be improved for those already diagnosed as infected FSWs to reduce complications of syphilis and infectiousness and/or susceptibility of HIV.

We found that inconsistent condom use was a significant risk factor of syphilis infection. The overall odd ratio (OR = 1.9) was similar to that reported in Sichuan (OR = 1.7) (Ruan et al. 2006), but smaller than that reported in Guangxi (OR = 5.3) (Lu et al. 2009). As Figure 2 illustrates, the effect of condom use in our study was only high in the age group of 25–29 years: with 8.7% vs. 1.0%, which correspond to an odds ratio of 9.4. For the 21–24 and over 30-year-old group, condom use was shown to be only slightly protective against syphilis infection. For those FSWs younger than 20 years, the prevalence was even slightly higher among those who reported consistent condom use. Information bias or some unstudied factors (e.g. number of clients per day) may explain the ‘non-protectiveness’ of consistent condom use among the younger FSWs. Nevertheless, as the homogeneity test does not confirm any effect modification (P = 0.20), the interaction pattern between condom use, age and syphilis infection is still unclear, and further research is needed. Furthermore, it should be noted that only 52% (855/1653) FSWs reported consistent condom use in the last month, peer education and condom promotion programmes should be continued to help them to maintain or where possible increase safe sexual practices.

In our study, the prevalence of syphilis was strongly associated with illicit drug use. This is consistent with other studies in China (Kang et al. 2011; Liao et al. 2011). The prevalence reached 19% of the 67 FSWs who reported illicit drug use in the past year. This alarmingly high prevalence of syphilis among those who ever used illicit drugs calls for attention from the health authorities. Inconsistent condom use could help explain why the prevalence among those who ever used illicit drugs was higher than that among those who never used illicit drugs. However, even after adjusting for the status of condom use, illicit drug use was still highly correlated with syphilis infection (P < 0.001). Having a larger number of sex partners may play an important role in increasing their probability to contract syphilis infection. Therefore, more detailed studies on the different sexual risk behaviours (e.g. number of clients per day) between FSWs who use illicit drugs and those who do not are needed to explore the reasons for the high syphilis prevalence and to direct interventions.

Moreover, FSWs who use illicit drugs are often more likely to engage in sharing needles have a larger number of sex partners and use condoms more inconsistently than those who never use illicit drugs (Wang et al. 2011). These behavioural characteristics together with co-infection of syphilis will place them at much higher risk of HIV infection. Fortunately, only 4% (67/1653) FSWs in our study reported illicit drug use in the last year, which is much lower than other areas in China, such as Shandong and Yunnan, where percentages as high as 45% have been reported (Wang et al. 2009; Kang et al. 2011; Liao et al. 2011, 2012). The low proportion of illicit drug use places FSWs under a lower risk of HIV infection, and indeed, we did not identify any HIV cases during our study period.

The small proportion of illicit drug users among FSWs in Shenzhen could be due to reporting bias or strong police repression of illicit drug use, but it is, more likely, due to the active outreach and education programmes by the local CDC. At all accessible venues, posters and education materials are provided, and this may have helped to raise the risk awareness of FSWs for using illicit drugs. The proportion of FSWs reported ever used illicit drugs in the last year was reported to be about twice as high in the study by Li. et al. of 2006 (Li et al. 2012b) as in our study of 2009–2012 (8% vs 4%). Peer education may also have been effective in reducing the probability of illicit drug use by FSWs and where needed link them to professional services, for example methadone maintenance treatment programmes.

Several limitations of our study need to be addressed. Firstly, as some of the data collected were based on self-reporting, risk factors such as condom and illicit drug use may have been misclassified, thereby introducing information bias. Secondly, FSWs included in our surveys were from venues whose gatekeepers were cooperative. These gatekeepers may be better educated and provide FSWs with a supportive environment for safe sexual practices. As research has found that gatekeepers' support is highly associated with consistent condom use (Li et al. 2010), results from our studies may be biased and syphilis prevalence may be underestimated. Thirdly, because we used a convenience venue-based sampling method, the representativeness and generalisation of the results to the FSW population at large should be interpreted cautiously. To improve the representativeness, we sampled FSWs from various venues and attempted as much as possible to include hard-to-reach FSWs, such as those based on the streets. However, certain subgroups of underground FSWs may still be undersampled or neglected (e.g. the internet-based FSWs). Fourthly, we recruited all FSWs from venues with <25 FSWs. These small venues are usually low-end establishments or street-based venues, such as foot bathing shops and roadside restaurants. It is likely that we have recruited FSWs from such small venues with a higher probability than those FSWs from bigger high-end establishments. The recruiting methods may thus have introduced selection bias and the syphilis prevalence may be overestimated, as FSWs from low-end establishments and street-based venues usually report more often unsafe sexual practices (Chen et al. 2012; Li et al. 2012b). Finally, this study was cross-sectional, so causal inferences cannot be made.

In conclusion, although our study showed a high prevalence of syphilis among FSWs in Shenzhen, China, it has probably declined compared with the earlier years. The prevalence was particularly high among FSWs who use condoms inconsistently, who have a history of illicit drug use and are in the oldest age groups. Comprehensive interventions (e.g. condom promotion and peer education) should be continued to increase the overall condom use and reduce illicit drug use. Detailed studies focused on FSWs who used illicit drug are needed to explore reasons of the very high prevalence of syphilis and to direct interventions. Point-of-care outreach screening programmes for early diagnosis and timely treatment should be started. These programmes will help to relieve the epidemic of syphilis, also to the benefit of other STDs, in particular HIV.

Ancillary