The links between sex work and injecting drug use have been shown to be important determinants in the spread of an HIV epidemic [21–23]. In Bangladesh, considerable risk behavior among male IDU has been documented through the annual Behavioral Surveillance Survey (BSS)  in whom the risks for HIV are not only through their risky injection practices but also their sexual behaviors. There is very limited information available about female IDU in the South and South East Asian region; most information is obtained through data on injection drug use in sex worker communities [24, 25]. Similarly, in Bangladesh female IDU have been difficult to access and information is largely confined to drug taking behaviors among female sex workers from brothels, streets and hotels obtained from the BSS . The female IDU enrolled in this study were identified through CARE, Bangladesh and through the networks of drug users and sex workers because of which sampling was not random and hence the data are not necessarily representative. Despite this limitation, this study is the first comprehensive report on female IDU in Bangladesh, which provides a comparative analysis of the risks and vulnerabilities of female IDU who do and do not sell sex.
This study revealed high levels of risk behavior and important similarities and differences in injecting and sexual risk behaviors for sex worker and non-sex worker female IDU. The study findings suggest that it is not only their individual behaviors but the circumstances that female IDU live in that can further marginalize and make them more vulnerable. We observed a substantial proportion of sex workers and non-sex workers were living on the streets which is pertinent to HIV as homelessness has been shown to be associated with higher HIV infection rates in IDU [26, 27]. However, non-sex worker IDU were better off than sex worker IDU in this regard as they were more likely to be living with their relatives from whom they were receiving financial support. On the other hand, a higher proportion of non-sex workers were supporting themselves financially by selling drugs, which carries many risks including incarceration and exposure to violence.
The pattern of drug use between sex worker and non-sex worker female IDU described here was similar and this has also been reported from other countries . However, riskier injection practices were documented among sex worker female IDU. Higher percentages of sex workers had shared needles/syringes ever in their lifetime, and more had shared their drug ampoule than non-sex workers. Such higher injection risks among sex worker female drug users have been reported from a study conducted among crack users in Kentucky, USA  but not in another conducted in IDU from Sydney, Australia . Although we report that equal proportions of sex worker and non-sex worker IDU borrowed used needles/syringes in the recent past, sex workers were more likely to lend their used needles/syringes, indicating that sex worker IDU are not only more vulnerable themselves but their injection sharing partners are also at higher risk for blood borne infections.
It is indeed fortunate that in both groups of female IDU no HIV was detected. Although one may argue that we under-sampled higher risk female IDU, but the national HIV surveillance shows that Bangladesh is a low prevalence nation for HIV and that in 2005, there was no HIV detected in IDU (N = 2294) from 13 cities out of the 16 cities from where a total of 3682 IDU were sampled . The reasons for this low prevalence are not clear although the NSEP may have played a role  especially as it commenced before any HIV was detected among IDU. However with rising HIV rates among IDU in Central Bangladesh there is no room for complacency  and it is essential that the harm reduction services are expanded, intensified with broad and active support from all relevant sectors.
We observed high risk sexual behaviors for both sex worker and non-sex worker groups of female IDU, and not surprisingly behaviors tended to be riskier for sex worker IDU. Although more sex worker IDU reported condom use during the last sex act, more reported anal sex, they had concurrent commercial and non-commercial sex partners and a substantial proportion had never used condoms with their non-commercial sex partners. Moreover, sex worker female IDU commonly reported serial sex with multiple partners (group sex), which was not reported at all by non-sex worker IDU. Although we were lacking data on the context of group sex and cannot assume that they were consensual, anecdotal reports suggest that group sex may occur within the context of sex work, with male clients pooling money to share a female sex worker. Such high levels of sexual risk behavior in female IDU are not unique to Bangladesh  and this has also been observed in places where IDU are accessing HIV prevention programs .
Consistent with the high risk sexual behaviors we observed, the prevalence of syphilis was high, especially among sex worker IDU who had a higher lifetime prevalence of syphilis. However, there was no significant difference in the prevalence of active syphilis between the two groups of female IDU which was also comparable to that reported by the national HIV surveillance data from street-recruited female sex workers in Central Bangladesh . Although this study did not measure other STIs, other studies of sex workers from different sites in Bangladesh have recorded very high rates of the different STIs including gonorrhea, chlamydia, trichonomiasis, syphilis and herpes simplex 2 [30, 31]. The sexual risk behaviors we documented among sex worker IDU are similar to those observed among female sex workers from Central Bangladesh reported in the BSS of Bangladesh . However, compared to the BSS, the frequency of reported condom use we observed was higher. The reason for this discrepancy is not clear but we cannot rule out the possibility of socially desirable responding since this study was conducted in collaboration with CARE, Bangladesh.
Female IDU are often more vulnerable to HIV than their male counterparts due to greater overlap between sex and drug use networks . Women who share drugs with their sex partners often share needles/syringes with these partners and may also have unprotected sex with them, compounding their risk of acquiring both blood-borne and sexually transmitted infections . Close to one third of the female IDU in either group studied here reported having unprotected sex with their injection partners. This is of particular concern as male IDU in Central Bangladesh are at the brink of a concentrated HIV epidemic . These women are not only extremely vulnerable to HIV but they may also represent 'transmission bridges' to the general community through commercial sex.
Sexual violence in IDU has been shown to be associated with greater risk of HIV infection and female IDU are more likely to have a history of sexual violence than males [9, 33]. A higher percentage of sex worker IDU reported forced sex as their first sexual experience, compared to non-sex workers. Other studies have shown close associations between childhood sexual abuse, prostitution and early initiation into injection drug use .
In this study, sex worker IDU were more commonly jailed in the last year than non-sex worker IDU as has been shown in other studies . For sex workers, the reason for incarceration was more frequently for selling sex while for non-sex workers it was associated with a special legal clause "section 54" which authorizes the police to arrest anyone on 'reasonable suspicion' that s/he has or is about to commit a crime without a warrant of arrest, or with the requirement to demonstrate any reasonable grounds for such suspicion . Incarceration may pose an additional risk of HIV acquisition in jails where drugs are widely available but needles/syringes are not [27, 36].
In this study, sampling of female IDU was non-random and most were associated with CARE, Bangladesh's intervention programs. This may explain why the overall knowledge about HIV transmission was high amongst both groups of female IDU. However, it was interesting that for prevention of transmission sex worker female IDU were more likely to mention condom use during sex while non-sex worker IDU were more likely to mention not sharing injection equipment as means of prevention. This knowledge pattern reflects the nature of services accessed by the two groups of female IDU, with sex workers more commonly availing of services for sex workers (i.e. condoms and STI management) whereas non-sex workers were more likely to access the NSEP. The services available for female IDU in Bangladesh are limited. The data here show that of the female IDU using the NSEP services several were getting their needles/syringes from other additional sources and the reasons provided for this indicated restrictions in access to NSEP. It has been well documented that effective harm reduction services including NSEP can reduce the spread of HIV in IDU [37, 38] and with restricted access to the NSEP continued needle sharing has been recorded . Modeling data obtained to assess the effectiveness of CARE Bangladesh's NSEP in Dhaka city suggests that the epidemic may have been blunted by the NSEP . However, this view is controversial, as BSS data on IDU from Central Bangladesh do not show safer injection behaviors over time despite the presence of the NSEP .