DRAFT. HIV/AIDS in South Asia: Understanding and Responding to a Heterogeneous Epidemic

Transcription

1 DRAFT HIV/AIDS in South Asia: Understanding and Responding to a Heterogeneous Epidemic University of Manitoba Human Development Department South Asia Region (SAR) The World Bank Global HIV/AIDS Program The World Bank Han Kang, Princeton University, USA James F. Blanchard, University of Manitoba, Canada Stephen Moses, University of Manitoba, Canada Faran Emmanuel, National AIDS Control Program, Pakistan Sushena Reza Paul, University of Manitoba, Canada Marissa Becker, University of Manitoba, Canada David Wilson, Global HIV/AIDS Program, The World Bank Mariam Claeson, South Asia Region, The World Bank

2 TABLE OF CONTENTS A. EXECUTIVE SUMMARY... 4 B. BACKGROUND AND RATIONALE... 7 B.1 INTRODUCTION... 7 B.2 RATIONALE... 7 C. ANALYTIC FRAMEWORK C.1 INTRODUCTION C.2 EPIDEMIC POTENTIAL C.3 EPIDEMIC PHASE C.4 BIOLOGICAL FACTORS C.5 BEHAVIORAL AND STRUCTURAL FACTORS D. HIV AND RELATED SURVEILLANCE IN THE SAR REGION D.1 INTRODUCTION D.2 INDIA D.3 NEPAL D.4 PAKISTAN D.5 BANGLADESH D.6 SRI LANKA D.7 AFGHANISTAN, BHUTAN AND THE MALDIVES E. SEXUAL AND INJECTING DRUG USE BEHAVIORS E.1 INTRODUCTION E.2 FEMALE SEX WORKERS E.3 CLIENTS OF FEMALE SEX WORKERS E.4 MEN WHO HAVE SEX WITH MEN E.5 GENERAL POPULATION E.6 INJECTING DRUG USERS E.7 FEMALE SEX WORKERS, CLIENTS, AND MSM WHO ALSO INJECT DRUGS E.8 IDUS WHO ALSO HAVE RISKY SEX F. HIV PREVALENCE AND SPREAD F.1 FEMALE SEX WORKERS F.2 MEN WHO HAVE SEX WITH MEN F.3 INJECTING DRUG USERS F.4 HIGH RISK MALE GROUPS F.5 GENERAL POPULATION... 31

3 G. COUNTRY-SPECIFIC ANALYSES G.1 INDIA G.1.1 Epidemic overview...33 G.1.2 A case study in heterogeneity Bagalkot District, India...40 G.1.3 Analysis of the HIV epidemic in India...44 G.1.4 Special considerations...45 G.1.5 India s response...46 G.2 NEPAL G.2.1 Epidemic overview...52 G.2.2 Nepal s response...52 G.3 PAKISTAN G.3.1 Epidemic overview...54 G.3.2 Pakistan s response...54 G.4 BANGLADESH G.4.1 Epidemic overview...56 G.4.2 Bangladesh s response...56 G.5 AFGHANISTAN G.6 BHUTAN G.7 THE MALDIVES H. COUNTRY ANALYSIS SUMMARIES I. SUMMARY OF RECOMMENDATIONS J. ANNEXES J.1 ANNEX 1: THE WORLD BANK RESPONSE TO AIDS IN SOUTH ASIA J.2 ANNEX 2: DEFINITION OF TARGETED AND GENERAL POPULATION INTERVENTIONS K. REFERENCES

4 A. EXECUTIVE SUMMARY Introduction: South Asia s HIV epidemic is severe in magnitude and scope, with India alone having at least 60% of all people living with HIV in Asia. The HIV epidemic is highly heterogeneous understanding the diversity of the epidemic between and within countries is a prerequisite for informed, prioritized and effective responses. This review was undertaken to provide a basis for rigorous, evidence-informed HIV policy and programming in South Asia. Focus: This review focuses on five South Asian countries for which significant data are available: India, Nepal, Pakistan, Bangladesh and Sri Lanka. Although data limitations preclude a detailed analysis for Afghanistan, Bhutan and the Maldives, data from these countries are cited where available. The paper focuses on prevention, while acknowledging and affirming the important and complementary role of treatment. Heterogeneity: South Asia s most severe epidemic is in parts of India, particularly in a cluster of southern and western states, including Tamil Nadu, Karnataka, Andhra Pradesh, Maharashtra and some northeastern states, including Mizoram, Nagaland and Manipur. Within these states, there is remarkable variability in HIV prevalence between districts and within districts, variability has been observed between blocks/tahsils/talukas and even between some villages in the same block/tahsil/taluka. The HIV epidemic may be as severe in parts of Nepal, where transmission occurs largely through sex work and injection drug use, among the sexual partners of those engaging in injection drug use. Significant numbers of both men and women have HIV. Both Pakistan and Bangladesh face growing epidemics, particularly among injecting drug users, but HIV rates remain relatively low among sex workers in those countries, and there is still an opportunity to avert a major heterosexual epidemic. Sri Lanka remains in an early epidemic phase, and its epidemic potential may be related to the likelihood of the growth of injecting drug use. Although there are limited HIV data for Afghanistan, it must act urgently to limit rapidly growing HIV infection in its large population of injection drug users, and especially where injection drug use and the sex trade intersect. Other South Asian countries that have too little data to form a core focus of this review Bhutan and the Maldives - have sufficient data to suggest they still have low prevalence epidemics. Structural amplifiers: All countries in South Asia have a diverse range of structural factors which amplify HIV-vulnerability and risk, including: widespread poverty and inequality; illiteracy; low social status of women; trafficking of women into commercial sex; a large, structured, sex work industry; porous borders; widespread rural-urban and inter-state and international migration; stigma and cultural impediments to sexual discussion; high rates of sexually transmitted infections; and, limited condom use. These amplifiers differ across the region. For example, trafficking of women into sex work is limited in Bhutan and Sri Lanka; taboos regarding sexual discussion are minimal in Bhutan; and Sri Lanka has high literacy. A preventable epidemic: South Asia s HIV epidemic is severe, but further spread is preventable. The future size of South Asia's epidemics will depend on an effective two-pronged approach: firstly, and most critically, on the scope and effectiveness of HIV prevention programs for sex workers and their clients, injection drug users and their sexual partners, and men having sex with men and their other sexual partners; and secondly, to support these programs, on the effectiveness of efforts to address the underlying socio-economic determinants of the epidemic, 4

5 and to reduce stigma and discrimination towards people with high risk behaviors, often marginalized in society, as well as people living with HIV. Prevention works: HIV prevention programs for sex workers, injection drug users and men having sex with men in South Asia work to a large extent, we do know what to do and how to do it. Results have been achieved through targeted interventions aimed at reducing risk behaviors and exposure. Furthermore, countries such as India have made major strides in tackling stigma and discrimination, although much remains to be done. We also know how to work though key sectors other than health, such as the transport sector, to reach potential clients of sex workers effectively. Coverage is the greatest challenge high coverage of high impact interventions among populations engaged in high risk behaviors, and their sexual partners, are essential to reduce HIV transmission. Prevention is cost-effective: HIV prevention among sex workers and clients, injection drug users and their sexual partners, and men having sex with men and their sexual partners, is relatively inexpensive and provides a high return on investment. Urgent, effective programs for sex workers, injection drug users, men having sex with men and the sexual partners of these communities can still prevent HIV becoming widely established in the general population, and such action will greatly reduce the costs of HIV prevention, care, treatment and other costs. HIV priorities and investments should closely address these transmission patterns and their key structural determinants. Rural epidemics: There is evidence of significant rural epidemics in parts of India and Nepal, and we lack an understanding of HIV prevention in rural contexts in South Asia. Understanding rural epidemics and how to respond effectively to them constitutes a major challenge. Country priorities: Major country priorities are summarized below: India: The future size of India's HIV epidemic will depend above all on the scope and effectiveness of programs for sex workers and clients, but also on the scope and effectiveness of programs for men having sex with men and their other sexual partners, and injection drug users and their sexual partners, the latter particularly in the northeast. Throughout India, it remains vital to tackle stigma and discrimination towards people at risk and people living with HIV. In high prevalence states, districts and blocks/tahsils/talukas, it is also important to tailor and apply focused strategies to reduce HIV transmission into vulnerable segments of the general population. HIV prevention and AIDS treatment have potential reciprocal benefits: HIV prevention makes AIDS treatment more affordable and AIDS treatment creates important opportunities for enhanced HIV prevention. Nepal: The future size of Nepal's HIV epidemic will depend above all on the scope, coverage and effectiveness of programs for sex workers and clients, and injection drug users and their sexual partners. Cross-border migration, particularly involving women migrating into sex work, particularly to Mumbai, increases HIV transmission. Sex between men constitutes a further risk, which must also be addressed. Nepal s continuing political difficulties makes civil society s already critical role even more vital. Tackling stigma remains a priority, as elsewhere in the region. 5

6 Pakistan: Pakistan's HIV epidemic is currently largely centered within networks of injecting drug users, with evidence of epidemic expansion among MSM and hijra communities. Effective prevention programs among these communities may avert a wider epidemic. If HIV spreads from injection drug users to sex workers, the epidemic will become even more serious, and a major prevention opportunity will have been lost. HIV infection among sex workers is still at a low level in general, and can be kept low through intensive programs for sex workers and clients, including a major focus on sex workers who inject drugs or whose sexual partners inject drugs. Stigma reduction is essential in order to achieve high-quality, high coverage programs. Bangladesh: Similarly, Bangladesh s HIV epidemic is currently largely driven by injecting drug users and is growing among MSM and hijra communities. Effective prevention programs among these communities may still avert a wider epidemic. If HIV spreads from injection drug users to the large sex networks, the epidemic will become even more serious, and a major prevention opportunity will have been lost. HIV infection among sex workers is still at a low level, and can be kept low through intensive programs for sex workers and clients, including a major focus on sex workers who inject drugs or whose sexual partners inject drugs. Stigma reduction is essential in order to achieve high-quality programs, with high coverage. Afghanistan: There is evidence of considerable HIV transmission among some members of Afghanistan s injection drug use community. Afghanistan must act urgently to limit HIV infection in its large population of injection drug users. Sri Lanka: HIV remains low even among high risk groups in Sri Lanka. Early, effective, affordable programs for injection drug users and their sexual partners, sex workers and clients, and for men having sex with men and their other sexual partners, can ensure that HIV remains at very low levels the country has an opportunity it must not lose. Bhutan and the Maldives: Despite limited data, and for very different reasons, these disparate countries may have low HIV prevalence and relatively small numbers of injection drug users, sex workers and clients. Programmatic implications and conclusion: South Asia requires a dual approach to HIV prevention: (a) most importantly, effective large-scale programs for sex workers and clients, injection drug users and their sexual partners, and men having sex with men and their other sexual partners; and (b) to support these programs, widespread HIV prevention and stigma reduction campaigns for the general population, and for individuals and groups who can facilitate or impede program implementation at a structural level. Given the enormous scale and epidemic heterogeneity in South Asia, investment is required to build a comprehensive information base regarding the priority locations for focused prevention programs. In addition, substantial capacity building is needed to improve the scale and quality of program implementation. Multisectoral responses, which address both the immediate practices and the underlying socioeconomic factors contributing to transmission, are essential. Large-scale, high coverage of targeted interventions and programs based on these principles can: greatly reduce the size of South Asia's HIV epidemic; prevent HIV from becoming widely established in the general population; and, markedly reduce AIDS treatment and other costs, providing a high return on investment. 6

7 B. BACKGROUND AND RATIONALE B.1 Introduction The South Asia Region (SAR) includes: Afghanistan, Bangladesh, Bhutan, India, the Maldives, Nepal, Pakistan and Sri Lanka. It is a large and diverse region, which is experiencing a complex, heterogeneous HIV epidemic, with considerable variation within and between countries. HIV was probably introduced into the region in the early 1980s, and the first AIDS cases were recognized in most countries by the end of the decade. In spite of similar times of HIV introduction, the epidemics in the various countries have played out in remarkably different ways. Indeed, this has even been the case within individual countries, particularly India, which is better understood as a continent, with individual states and even smaller geographies being considered as countries in their own right, with unique epidemic patterns, requiring particular responses. Indeed, a major lesson from Africa - a continent with approximately half the population of India alone - is the need to understand the remarkable diversity and heterogeneity of HIV, both between and within regions and countries. This lesson has received insufficient emphasis in South Asia. This report, which is based on existing data, focuses on five countries in the region for which there is adequate secondary data to support an analysis: India, Nepal, Pakistan, Bangladesh and Sri Lanka. Brief comments are also offered on the HIV epidemics in Afghanistan, Bhutan and the Maldives, for which there are intriguing strands of information, but too few data to support an in-depth analysis. B.2 Rationale In light of the complexity and diversity of the region s HIV epidemic, it is vital to ensure that policy and programming are informed by rigorous analysis. In particular, it is important to understand HIV epidemiology, transmission dynamics and the behavioral and socioeconomic determinants, potential evolution, response priorities and gaps. After three decades of the global HIV epidemic, we have a greater understanding of the global distribution of HIV and of the underlying sexual behavior and injection behavior determinants of HIV transmission. The contours of the epidemic are increasingly apparent and there are sufficient continental similarities to speak broadly of Asian epidemic patterns. Within the Asian epidemic, there are important variations. An understanding of both the underlying similarities and variations is central to an informed understanding of the epidemic. In three decades, we have also learned how critical it is to ensure that HIV responses are based on a rigorous and objective understanding of the bio-behavioral determinants of HIV transmission and the underlying structural and contextual factors. Global HIV responses have been undermined by generic approaches, which do not address the major local drivers of epidemics in each context. It is vital to ensure that responses are tailored and prioritized to address the major drivers of transmission. Alongside an understanding of the importance of transmission dynamics and priority responses, we have learned how important it is to identify and invest in effective, proven HIV responses. Crucially, effective approaches must be undertaken on a large-scale and reach a majority of those 7

8 at risk of infection. These principles are admirably encapsulated in a time-honored public health mantra: do the right thing, do it right and do enough of it. Nowhere are these principles more important than in Asia, whose size, complexities and disparities within and between countries compels an intelligent epidemiological reading and effective, focused response. Despite the diversity of HIV in Asia, there are sufficient commonalities to speak of broad Asian HIV epidemic patterns. What are the central features of the Asian HIV epidemic? In Asia, the epidemic is driven largely by high risk practices: injection drug use and unprotected commercial sex and anal sex among subsets of the population. The sexual partners of those engaging in the above practices are also at elevated risk of acquiring and transmitting HIV. The overall size of the Asian epidemic thus depends on the prevalence of HIV within these networks, their size, number of sexual or injecting partners, sexual relationships with the wider community, and the extent of preventive measures. In some Asian countries, such as Thailand, Cambodia and much of India, the scale and frequency of commercial sex has been sufficient to ignite sexual epidemics among sex workers, their clients and a growing number of their sexual partners. In many countries, such as Indonesia, Vietnam, and China, injection drug use ignites epidemics that spread to sex workers, then to their clients and beyond. Throughout Asia, unprotected anal intercourse constitutes a significant source of HIV transmission, which needs to be better understood. HIV spreads among groups at high risk to their immediate sexual partners and thereafter into the wider community, presenting a major development challenge. An understanding of these sexual and injecting determinants in each context is central to an informed response. Such an understanding is required to reach those at risk with appropriate, large-scale interventions. In addition, widespread stigma makes it harder to reach groups at high risk and to implement proven approaches stigma reduction is thus essential. An analysis and understanding of socio-cultural and economic determinants is also important, since South Asia exhibits several socio-cultural characteristics that may contribute to increased HIV transmission, including poverty, inequality, gender inequity, migration, human trafficking and proximity to the golden crescent, or nerve centre of the global opium trade. There is a growing corpus of biological and behavioral studies in South Asia, and these studies provide the basis for a better understanding of South Asia s epidemic. However, these studies have seldom been analyzed and interpreted in an integrated, analytical manner. Indeed, the literature on Asia s epidemics is largely dominated by East Asian evidence and discourse. There is thus a need for a rigorous analysis and synthesis of the major bio-behavioral determinants and trends in South Asia s HIV epidemic, reinforced by an equally rigorous review of the evidencebase for various interventions, and a review of the scope and reach of existing interventions. It is vital to examine the heterogeneity of the epidemic across and within South Asian nations. The notion of a national epidemic in a country as vast and diverse as India let alone the wider South Asian region - belies the reality of multiple, variegated local epidemics. This report seeks to undertake such an analysis in South Asia. The report s objectives are: to provide a state-of-the-art assessment of South Asia s existing HIV epidemic, its major transmission dynamics and potential evolution; to propose a rigorous, evidence-based, practical HIV response strategy for the region; and, to highlight priorities for greater emphasis in South Asia s HIV responses. 8

9 We also hope that this analysis will stimulate a discussion about priorities, strategic focus and allocation of resources, among all partners involved in mounting an effective response to curb the HIV epidemic in South Asia by 2015: governments, development partners, nongovernmental organizations, the private sector and civil society. 9

10 C. ANALYTIC FRAMEWORK C.1 Introduction It is clear that the economic, social and cultural diversity in South Asia is associated with substantial heterogeneity in HIV epidemiology and transmission dynamics, both between and within countries in the region (MAP, 2005). Describing and understanding this diversity is perhaps the central challenge in formulating strategies to curb HIV. Although there are large gaps in the data necessary to do this comprehensively, it is our view that analysis of the extant data is sufficient to provide broad strategic guidance. It is important to note that to do this properly it is necessary to use an analytic approach that is both descriptive and anticipatory. Accordingly, our analytic approach is based on two epidemiological concepts: epidemic potential and epidemic phase. C.2 Epidemic potential Based largely on an analysis of the size and distribution of key high risk sub-populations, e.g. sex workers (SWs), injection drug users (IDUs), and men who have sex with men (MSM), and the nature of sexual networking, an assessment of the epidemic potential focuses primarily on the extent to which an HIV epidemic can be maintained and amplified beyond the spread within networks directly linked to high risk sub-populations. For this analysis, we consider three types of epidemics, based on potential: Truncated epidemic: This describes a situation where HIV transmission will be confined to individuals who participate in high-risk networks such as commercial sex networks or IDU sharing networks, and where these high-risk networks are not localized. An example would be a rural area where there are a substantial number of out-migrants who are sex clients at an urban migration destination point, but there is a low level of high-risk behavior locally. In this circumstance, transmission is largely confined to the sexual partners of returning migrants, and not further amplified by local transmission. Prevention strategies should thus focus primarily on interrupting transmission at the migration destination points, with reinforcement at origin and transit points in areas where there are very large concentrations of out-migrants. At the locations of origin, greater emphasis will be required on HIV counseling and testing, and care and support services. The following figure schematically depicts a truncated epidemic. Truncated Epidemic High risk network (distal) Bridge Population Local Partners 10

11 As we discuss later in this document, truncated epidemics are likely to be present in many rural locations of Pakistan and India where the local sexual structure does not support much local HIV transmission, but where there is substantial out-migration of men to urban areas where significant commercial sex exists. Transmission in the local rural area is largely restricted to the sexual partners of returning migrants who have been infected at their migration destination. Local concentrated epidemic: This refers to an epidemic where there are substantial enough spread networks locally (e.g. sex work or IDU) to initiate local transmission within high-risk sub-populations, and to the wider local population through bridge populations. The size of such epidemics would be largely determined by the size of the high-risk sub-populations and the other sexual networks in the local area, but HIV transmission dynamics remain driven by the high-risk networks. Prevention strategies in this type of epidemic would need to focus on interrupting transmission within both distal and local high-risk transmission networks. Schematically, a local concentrated epidemic is depicted as follows. Local Concentrated Epidemic High risk network (distal) Bridge Population High risk network (local) Local Partners In our view, this is the most important epidemic pattern in South Asia. It is epitomized by the many locations where there are substantial high-risk sexual and IDU networks, but although the HIV prevalence within those networks reaches high levels, the overall epidemic growth is constrained, and the prevalence in the general population does not reach levels beyond 1-3%. Generalizing epidemic: Such epidemics begin in local high-risk networks, but due to the risk behavior patterns in the wider community, the HIV epidemic spreads beyond the highest-risk networks, and ultimately transmission becomes somewhat independent of easily defined highrisk groups. Strategic responses to such situations should include both targeted interventions for high-risk groups, and an early emphasis on reducing the potential for transmission in the more general population through enhanced STI services, broader behavior change programs, and aggressive condom promotion. Schematically, a generalizing epidemic is conceptualized as follows: 11

12 Generalizing Epidemic High risk network (distal) Bridge Population High risk network (local) Local Partners Generalizing epidemics are seen in many countries of sub-saharan Africa, but as we discuss subsequently, we do not believe that there are many locations within the South Asia Region that have truly generalizing epidemics in the sense that the transmission dynamics are such that epidemics are sustained and amplified without dependence on high-risk networks. Possible exceptions would be some localized areas in southern India, particularly in the states of Andhra Pradesh, Karnataka and Maharashtra, as well as parts of northeast India, particularly in the states of Manipur, Mizoram and Nagaland. It should be noted that while the concentrated and generalizing epidemics will generally result in higher HIV prevalence, these typologies are not defined strictly by that measure, but rather on behavior patterns and networks. Indeed, local concentrated epidemics could result in HIV prevalence in the general adult population exceeding one per cent (districts in Manipur may constitute a relevant example) if the sizes of the high-risk sub-populations are large enough. Thus, an epidemic is concentrated if it is driven primarily by high risk groups and if effective programs for high risk groups would reduce overall HIV transmission, and generalized if transmission occurs primarily outside high risk groups and would continue despite effective programs for high risk groups. In short, an epidemic is concentrated if stopping high risk group transmission would control the epidemic, and generalized if stopping high risk group transmission would not control the epidemic. C.3 Epidemic phase Conceptually, the epidemic phase describes the extent to which the HIV epidemic has progressed along its expected trajectory in terms of its sub-population distribution. This trajectory will be determined by the epidemic potential described above. The main purpose of understanding the epidemic phase is for diagnostic purposes, since judging solely from HIV prevalence, an early phase generalizing epidemic will not be much different from a later phase concentrated epidemic. Assessing the epidemic phase can be difficult, since it relies on an understanding of how long ago HIV was introduced into various sub-populations and a measurement of the HIV prevalence in different sub-populations. Each type of epidemic can be at an incipient, growth, plateau, or decline phase as depicted below. 12

13 Generalizing Concentrated Truncated Incipient Growth Plateau Decline C.4 Biological factors In addition, there may be biological factors related to the virus or the population which may mitigate the spread of HIV infection. Many of these are inadequately understood, and it is beyond the scope of this paper to discuss them in detail, but the most important such factor, which does appear to influence the pattern of HIV spread in South Asia, is male circumcision. The following comments are not intended as an argument for male circumcision as a recommended public health intervention in South Asia, but as a heuristic to understand epidemic potential in South Asia. Scientists have noted an association between male circumcision and HIV rates since the 1980s (Bongaarts et al, 1989), including in India (Reynolds et al, 2004). For years, the evidence was considered plausible, but many observers have argued that it is difficult to disentangle other factors, such as religion, culture, sexual behavior and geography as potential confounding factors. However, the weight of evidence has grown stronger. A meta-analysis of 38 studies from Africa concluded that uncircumcised men were more than twice as likely to have HIV as uncircumcised men (Weiss et al, 1999). A longitudinal study of male sexual partners of HIV-positive women in Rakai, Uganda, found that 40 out of 137 uncircumcised men and 0 out of 50 circumcised men acquired HIV (Grey et al, 2000). Ecological evidence demonstrates an increasingly close geographic association between lower male circumcision rates and higher HIV prevalence rates. A major UNAIDS multi-country comparison of high and low prevalence African cities concluded that male circumcision was the major predictor of disparities in HIV levels (Auvert et al, 2001). No Asian country with widespread male circumcision has a significant heterosexual HIV epidemic. The gold standard of public health programs is of course a randomized trial. In mid-2005, a randomized trial of male circumcision of 3,035 men in Orange Farm, South Africa was halted when an interim analysis demonstrated a protective effect so large that it would have been unethical to continue the trial. The analysis showed that male circumcision reduced HIV incidence by 60%, from 2.2% to 0.77% (Auvert et al, 2005). There are plausible biological explanations for the relationship between male circumcision and HIV infection. The intact foreskin has far more Langerhans target cells than other genital tissue. The internal foreskin has a soft mucosal surface, unlike the hardened skin-like surface of the external foreskin. Circumcision results in keratinisation, or toughening of the glans. An intact foreskin provides a warm, moist environment for infectious agents (Patterson et al, 2002, Szabo et al, 2000). 13

14 The implications of these data for the South Asia region may be summarized as follows. Male circumcision is widespread in Pakistan, Bangladesh and Afghanistan, and uncommon elsewhere in the region. Thus, Pakistan, Bangladesh and Afghanistan may have a more limited potential for heterosexual HIV epidemics. However, injection drug use may ignite otherwise dormant epidemics, particularly if there is a nexus between injection drug use and sex work. HIV transmission among men having sex with men may also play a proportionately greater role in Pakistan, Bangladesh and Afghanistan, because of greater transmission efficiency related to anal intercourse, even among circumcised men. Conversely, the absence of extensive male circumcision may increase the relative epidemic potential in other South Asian countries, particularly where it coincides with other behavioral and structural factors, as discussed below. C.5 Behavioral and structural factors In addition to biological factors, behavioral and structural factors influence epidemic potential. Perhaps the most important behavioral factors are rate and patterns of sexual partner change. While there is a robust association between the number of sexual partners and HIV infection in many contexts, patterns of partner change may be at least as important (Halperin and Epstein, 2004). Growing biological evidence shows that HIV viral load, and thus, infectivity, is far higher during acute HIV infection, that is, in the initial weeks after HIV infection (Chao et al, 1994; Quinn et al, 2000). This leads to the important distinction between serial and concurrent sexual patterns (Halperin and Epstein, 2004). In serial partnerships, one typically has one ongoing sexual relationship at a time. In concurrent partnerships, one may be in a sexual network with more than one ongoing sexual relationship at a time. Whereas serial partnerships limit exposure to a partner with acute HIV infection (who has higher infectivity), concurrent partnerships expose everybody in an ongoing sexual network to greater risk. Mathematical models suggest that concurrent sexual partnerships may increase HIV transmission tenfold projections which are firmly supported by growing biological evidence of variability in viral load and infectivity (Morris et al, 1997). Structural factors, often inter-related - including poverty, gender inequality, trafficking, large structured sex work industries, and oscillating migration - influence sexual behaviors and networking patterns. What is clear is that the size of HIV epidemics in South Asia depends largely on rates and patterns of partner change inside and outside commercial sex. Injection drug use may ignite heterosexual HIV transmission in contexts where it may otherwise have been unlikely, including in Pakistan, Bangladesh and Afghanistan, and amplify it where the potential already exists. A nexus between injection drug use and sex work may play a particularly important role in igniting and amplifying HIV transmission. The golden crescent, which is the nerve centre of the global opium trade, straddles South Asia, and trafficking routes transect the entire region. The map below depicts major drug producing areas, with the golden crescent and golden triangle encircled. Four countries in South Asia are directly affected by these production areas Afghanistan and Pakistan by the golden crescent and India and Bangladesh by the golden triangle (UNODC, 2004). It is clear from this map that HIV risk transcends artificial regions and requires trans-regional programming linking Afghanistan and parts of Pakistan more closely to Iran, and central Asia and parts of India and Bangladesh more closely Burma and East Asia. This has important implications for multi-sectoral, regional and trans-regional programming. 14

15 Golden Crescent 15 Golden Triangle UNODC, 2004

16 D. HIV AND RELATED SURVEILLANCE IN THE SAR REGION D.1 Introduction By the late 1990s, all five SAR countries included in the core review had established some form of sentinel serological surveillance adhering to the World Health Organisation s recommended practice of collecting and screening anonymous and unlinked blood samples. In addition, India, Pakistan and Bangladesh have initiated second-generation surveillance, and have conducted at least one round of behavioral surveillance. Nepal also intends to launch second-generation surveillance activities, but requires more preparation. The content and quality of surveillance varies by country and time. D.2 India India began surveillance for HIV infection and identification of AIDS cases through 62 public health and nine reference centres in After its establishment five years later, the National AIDS Control Organisation (NACO) assumed responsibility for HIV surveillance, gradually expanding the network to comprise 180 sites, the majority involving antenatal (ANC) women, but including 77 for sexually transmitted infection (STI) patients and 9 for intravenous drug users by One year later, NACO added more sites for STI patients, IDUs and antenatal women, but also started HIV surveillance among female sex workers (FSW, with one site in Mumbai) and men who have sex with other men (with three sites in Mumbai, along with sites in Goa and Tamil Nadu). While the number of sites for FSWs and MSM have remained at about the same level since 2000, those for STI patients, pregnant women and (to a lesser extent) IDUs have increased in each annual round (NACO, 2005). As part of second-generation surveillance, as well as the national monitoring and evaluation (M&E) framework, India conducted a baseline behavioral surveillance survey (BSS) among FSWs, clients of FSWs, MSM, IDUs and the general population in 2001 (NACO, 2002). This was the largest behavioral survey ever undertaken, comprising 22 sampling units from 34 states and territories, and included 84,478 people (42,125 urban, 42,263 rural, 42,631 female, 41,847 male, mean age 29 for females and 30 for males). NACO also conducted behavioral surveillance among 5,572 FSW and 5,468 clients. The M&E framework calls for follow-up behavioral surveillance in the middle and at the end of NACO s second phase of the National AIDS Control Program (NACP II). Although only one national round has been completed, several states have conducted their own state-specific BSS, which are cited in this report. All of the behavioral and HIV prevalence data in this report are drawn from available national and state surveillance reports, and from other sources where available. For brevity, most subnational data cited generally includes only the high- and moderate-prevalence states. NACO (2005) defines the former as a state in which HIV prevalence exceeds 5% among high-risk groups and 1% among pregnant women, and the latter as a state in which HIV prevalence among high-risk groups also exceeds 5%, but does not exceed 1% among pregnant women. However, the adequacy of surveillance in low-prevalence states is of great concern and requires urgent improvement

17 HIV Sentinel Surveillance Sites in India Year HIV Sentinel Surveillance Sites ANC / ANC-R STDIDUMSMFSWNo. of HIV sentinel sites /

18 D.3 Nepal Nepal attempted to establish sentinel HIV surveillance in 1991 (National Centre for AIDS and STD Control/Family Health International, 2003). The first round was planned to cover FSWs, IDUs, STI and tuberculosis (TB) patients, as well as antenatal women, from seven sites across the country. However, follow-up surveillance has not occurred systematically. The National Centre for AIDS and STD Control (NCASC) has managed to continue surveillance among STI patients, but has not collected data for the past two years. The unreliability of surveillance data has prompted bilateral donors to conduct a series of cross-sectional studies, contributing to the current knowledge of Nepal s HIV situation. After recently reviewing its strategy, NCASC plans to re-establish surveillance with a second-generation system covering STI and TB patients, military service personnel, pregnant women and blood donors. D.4 Pakistan Pakistan launched sentinel HIV surveillance in The current site network consists of voluntary counselling and testing (VCT) centres in all four provinces, all public sector blood banks in three provinces (Balochistan, Punjab and Sindh) and STI clinics in all tertiary-level hospitals in two provinces, Punjab and Sindh (personal communication with HIV/AIDS programs at the national and provincial levels). More recently, supported by the Canadian International Development Agency, Pakistan has launched a second generation surveillance program that includes: enhanced mapping to determine the locations and sizes of key high risk networks of sex workers, IDUs, MSM and hijras; and integrated biological and behavioural surveillance (IBBS) among high risk groups to better understand HIV transmission dynamics and epidemic potential. Pakistan has now completed mapping in several key cities and IBBS activities have been initiated, with initial successful pilots in Karachi and Rawalpindi having been completed in In addition, Pakistan has commissioned additional broader mapping studies, and biological and behavioural studies, with a strong focus on high risk male populations. D.5 Bangladesh Bangladesh began surveillance for HIV infection in 1998, covering FSWs, STI patients, truck drivers, IDUs, MSM and the general population (albeit to a lesser extent), through its National AIDS and STD Programme established in the same year (National AIDS/STD Programme, 2004). Most sentinel sites, however, are in urban areas and locations with HIV prevention programs. Bangladesh has five rounds of behavioral surveillance and is now pioneering the use of respondent driven sampling among high risk groups in South Asia. Bangladesh has made major strides in surveillance in recent years

19 D.6 Sri Lanka In 1993, Sri Lanka started sentinel surveillance for HIV infection among FSWs as well as STI and tuberculosis patients. Similar to the evolution of India s sentinel surveillance system, the National STD and AIDS Control Program (NSACP) in Sri Lanka gradually expanded its coverage, adding blood donors in 1998, pregnant women in 2000, military service personnel in 2003, and transport workers and civil service candidates in Following WHO s guidelines to tailor surveillance activities according to the country-specific epidemic, Sri Lanka, with its low-level epidemic, is further expanding coverage of high-risk groups while discontinuing surveillance of the general population (personal communication with the National AIDS Control Programme). To date, most HIV-related behavioral research has been in the form of crosssectional studies. Like Pakistan, however, Sri Lanka is planning to launch baseline behavioral surveillance among FSWs, military service personnel, police, transport workers, internal migrant laborers and beach boys within two years. Preparation for this task has influenced and enhanced the conduct of sentinel surveillance, shifting sample collection away from clinical facilities and adding new groups, towards the eventual integration of both serological and behavioral surveillance. D.7 Afghanistan, Bhutan and the Maldives HIV surveillance is largely limited to incomplete case reporting in each of these three very disparate countries. Afghanistan will undertake behavioral HIV surveillance, perhaps including HIV testing, among injection drug users in In Bhutan, an important objective of the World Bank financed HIV/AIDS prevention and control project, which became effective in August 2004, is to support improved strategic information systems, including HIV sero and behavioral surveillance and STI surveillance, for HIV/AIDS prevention. The 2004 round of sentinel surveillance included 10 population groups, including ANC attendees, blood donors, STI patients, armed forces, and drivers, prisoners and female sex workers. The Royal Government of Bhutan has recruited ICDDRB (which is implementing the sero and behavioral surveillance in Bangladesh) to provide TA in surveillance and M&E. The government plans to start presurveillance assessment (identification and mapping of high risk populations) in It has also revised the sero surveys, which had in the past serious limitations due to small sample sizes of high risk populations, such as sex workers. The Maldives largely relies upon case reporting

20 E. SEXUAL AND INJECTING DRUG USE BEHAVIORS E.1 Introduction In this chapter, behavioral trends are analyzed across countries in South Asia. A subsequent country analysis distils major trends within countries and links them to biological trends and program data. E.2 Female sex workers Behavioral surveillance survey (BSS) data from India (NACO, 2001b) and from a crosssectional study in Sri Lanka (Saravanapavananthan, 2002) highlight how client load and consistent condom use may differ according to the type of sex work. It is estimated that there are 3-5 million female sex workers in India, with considerable variation among states. In the 2001 NACO BSS in India, FSWs overall reported a mean of 11 paying clients in the past seven days, with considerable variation among states. Brothel-based FSWs reported a much greater paying client load than their non-brothel-based counterparts. Likewise in Sri Lanka, which has an estimated 30,000 FSWs and 4,800 brothels country-wide (2,000 massage parlor-based FSWs and 68 brothels in Colombo), client load varies by the site of sex work. A higher proportion of brothel-based FSWs than of those based in massage parlors or on the street in Colombo have two to five clients per day. Within India, the mean number of paying clients in one high-prevalence state (Maharashtra) and two moderate-prevalence states (Goa and Gujarat) exceed the national mean, reflecting in part the importance of brothels as sex work foci in these states. In India, a greater proportion of brothel-based (57%) than non-brothel-based (46%) FSWs reported using condoms consistently with clients in the past 7 days. Similarly in Sri Lanka, a higher proportion of brothel-based (38%) than massage parlor-based (10%) FSWs in Colombo report consistent condom use. In contrast to the behavioral surveillance survey in India, however, condom use in Sri Lanka does not differ by the type of sex partner. Data from India show consistent gaps in consistent condom use between paying clients and non-paying partners, with the widest gap observed in Maharashtra (73% for paying clients versus just 7% for non-paying partners) and Tamil Nadu-Pondicherry (54% versus 9%). Hence, the lower proportions of all FSWs in Colombo reporting consistent condom use may be attributable to infrequent or no condom use with their non-paying partners. Such behavior suggests that FSWs misperceive non-paying partners as less prone to transmit HIV. This is a common theme throughout the region, and suggests an important area in which prevention programs among FSWs should focus, particularly as they mature, emphasising correct and consistent condom use with both paying clients and non-paying partners. Nepal and Bangladesh have completed mapping and condom use studies among FSWs. Nepalis refer to FSWs as bhiringi (a colloquial term for syphilis) girls (Pike, 1999). Many are members of the Badi caste from the Far Western region. Somewhat similar to the Devadasi tradition in northern Karnataka, Badi women historically worked as entertainers, but the scope of their occupation has evolved to include sex work. Mapping studies estimate a high concentration of FSWs (9,600) in the highway districts, with a substantial number in the Kathmandu Valley (4,000) and some in Pokhara of the Western region (300) as well (NCASC/FHI, 2003). The client load among FSWs from the Kathmandu Valley resembles that among FSWs from Colombo: only 10% report more than five clients per week (Furber, Newell and Lubben, 2002). The proportions of FSWs from the highway districts (52%) and Pokhara (72%) reporting

21 consistent condom use resemble those across India (FHI, 2002). Bangladesh has about 50, ,000 FSWs. In contrast to Sri Lanka and Nepal, however, Bangladeshi FSWs report higher client loads and low levels of consistent condom use. Hotel-based FSWs report the highest client load (44 per week), while their brothel-based (18 per week) and street-based (17 per week) counterparts have similar but lower client loads. Although 85% of brothel-based FSWs report that they have participated in some form of NGO-delivered prevention intervention, consistent condom use remains low among new clients (5%) and is even lower among regular clients (2.8%) (NASP, 2004). Pakistan has recently completed a mapping study to estimate the number of FSWs in the country, which shows that there are significant concentrations of sex work networks in some major cities (e.g. Karachi, Lahore and Multan) (NACP, 2005). In these cities, a substantial proportion of FSWs operate from hotels or homes. The home is generally the predominant site for sex work in other cities, except in Hyderabad, where three times as many home-based FSWs operate from the street. While some cities have established red-light districts, the illegal status of FSWs, inter alia, has driven them to work out of homes and/or other private facilities. The predominance of home-based sex work in many cities across Pakistan may pose a barrier to traditional prevention interventions which are designed to reach FSWs who are gathered together in well-defined, readily identifiable, public places. E.3 Clients of female sex workers Estimating numbers of clients of FSWs is even more problematic than estimating numbers of FSWs, but in India alone, they clearly must number at least in the tens of millions. Crosssectional studies in South Asian countries of males whose occupations entail travel and/or extended stays away from their households reveal significant proportions who report sex with FSWs. Sri Lanka has an estimated 700,000 clients of FSWs, including students, police, truck drivers, dockworkers and sailors (Saravanapavananthan, 2002). In Bangladesh, one study found that 54% of truck drivers and rickshaw pullers had sex with at least one FSW in the past year (Gibney et al, 2002). Another study in Nepal showed that the proportions of transport workers and migrant labourers who had sex with an FSW increased by 20% for each group within a oneyear period (i.e. 42% to 62% for transport workers and 10% to 30% for migrant labourers between ) (FHI, New Era and SACTS, 2002a). Both the study in Bangladesh (Gibney et al, 2002) and the NACO behavioral surveillance data in India (NACO, 2001b) estimate the number of FSWs with whom clients had sex. According to the former, truck drivers and rickshaw pullers report sex with a mean of almost two FSWs in the past month and of about five in the past year. In India, clients of FSWs report sex with a mean of nearly four FSWs in the past three months, with the highest number reported in Gujarat (6.2 FSW partners). Furthermore, Indian clients report a greater mean of non-brothel-based FSW partners (4.2) than of those based in brothels (3.3). Consistent condom use varies by country, and in India, by type of sex partner. In Bangladesh, the same cross-sectional study (NASP, 2004) noted above involving transport industry workers found low reports of consistent condom use: 2.3% of rickshaw pullers and 4.1% of lorry drivers. In contrast, in the same cross-sectional study from Nepal noted above, 60% of transport workers and 45% of migrant labourers reported using condoms consistently (FHI, New Era and SACTS, 2002a). These figures resemble those observed across India, ranging from as low as 54% in Manipur to as high as 77% in Maharashtra, with a slight difference between clients who report

22 sex with brothel-based FSW (60%) and those who report sex with non-brothel-based FSWs (56%). Reports of consistent condom use reported by clients generally match those reported by FSWs, except in Andhra Pradesh, Karnataka and Tamil Nadu-Pondicherry, where clients reports exceed FSW reports by about 10%, suggesting possible recall biases. Social desirability bias is often cited as an explanation for the high rates of reported condom use reported by FSWs in India in various settings, but these data suggest that this might not be as great a problem as might be suspected. Behavioral surveillance data from India again highlights a variation in consistent condom use by the type of sex partner. Generally, clients of FSWs are less likely to use condoms consistently with more regular non-fsw female partners, perhaps because they perceive them as less risky. This issue is analogous in nature to that noted above, of FSWs using condoms less frequently in interactions with non-commercial partners. E.4 Men who have sex with men Mapping studies from Pakistan demonstrate that like FSWs, there are large concentrations of men who have sex with men (MSM) in large urban areas. These suggest that Karachi has almost 5,000 male sex workers (MSWs) and 7,626 hijras (Ghauri et al, 2003), while Lahore has 7,500 MSWs and 2,000 hijras (NACP and Naz Foundation International, 2005). These kind of sites are therefore appropriate for HIV and second generation surveillance among MSM, as has been undertaken in India, where populous cities have been selected for both serological and behavioral surveillance. Behavioral surveillance data from India (NACO, 2001c) highlight that among male sex partners of MSM, a greater proportion are MSW, as the mean number of MSW partners exceeds that of non-msw male partners in all sites. Additionally, contrary to the trend observed among FSWs and their clients, a higher proportion of MSM report consistent condom use with non-msw partners than with MSW partners in every site except Delhi, where the difference is minimal. This may perhaps be a function of partner regularity, i.e. MSM may be more likely to have MSW than non-msw males as regular partners. Findings from a cross-sectional study among MSM in Sri Lanka (Saravanapavananthan, 2002) indicate that partner load varies by site. Generally, MSM identified at STI clinics are more likely than their counterparts in gay support groups and other locations to report fewer lifetime sex partners (45% of MSM STI patients report less than two lifetime sex partners versus 69% of MSM from other locations, who had more than 10 such partners) and less same-sex partners in the past 12 months (65% of MSM STI patients report less than two male sex partners, versus 44% of MSM from other locations who report sex with more than 10 males). The study also points out that in each site-specific MSM group, a proportion of them self-identify as bisexual. Bisexual behavior manifests in two ways. First, cross-sectional studies from Bangladesh (Gibney et al, 2002) and Pakistan (Baqi et al, 1999; Khan, 1996; Mirza and Hasnain, 1995) as well as behavioral surveillance from India (NACO, 2001c) reveal that clients of FSWs also have sex with men. In Bangladesh, 7% of rickshaw pullers and truck drivers report sex with a MSW in the past year, while 21% of them report sex with an MSW at least once in their lifetime. Similarly, in Pakistan, 39% of clients of FSWs are also clients of MSWs, and 30% of male STI patients have sex with males. Across India, 29% of clients of FSWs report sex with at least one male partner in the past 12 months; a greater proportion of those who are clients of brothel-based FSWs (37%) versus non-brothel-based FSWs (26%) report such behavior. By state, percentages range from 16% in Manipur to over 45% in Maharashtra, Goa and Gujarat. Reports of consistent condom use also vary. In Andhra Pradesh, Karnataka and Manipur, clients of FSWs who have sex with male partners do not use condoms consistently with them. In Maharashtra, Tamil

23 Nadu-Pondicherry and Gujarat, about a third of FSW clients report consistent condom use with their male partners. Behavioral surveillance data from India also indicate that about a quarter or more of MSM in all sites except for Karnataka report sex with at least one female in the past six months, with a mean of about one to three partners. Just as MSM are less likely to use condoms consistently with MSWs than with non-msw male partners, they also report lower consistent condom use with female partners, not exceeding 26% in any site. The low reports of consistent condom use by MSM with female partners may suggest that for MSM in India, female partners are perceived as less risky for HIV transmission. In any case, this represents a potential bridge for the transmission of HIV and other sexually transmitted infections to the general population. E.5 General population Behavioral surveillance in the general population in countries other than India is generally lacking or not very informative. As part of the NACO BSS in India, men and women in the general population in urban and rural areas were asked about the extent to which they had sex with non-regular partners and used condoms consistently (NACO, 2001a). Across the country, 7% reported having sex with non-regular partners in the past 12 months, 12% among men and 2% among women. In two of the five high-prevalence states, Andhra Pradesh and Maharashtra, and two moderate-prevalence states, Goa and Gujarat, sex with non-regular partners is reported more frequently than the national mean. There is considerable variability among states and between rural and urban areas. Overall, a roughly similar proportion of urban males report nonregular partners compared to rural males, and the same is true for females, although at a lower level overall. In Maharashtra, a greater proportion of urban males than their rural counterparts report having sex with non-regular partners (23% versus 10%), but the opposite urban-rural divide is generally the case in other states, although the differences are small. In any case, the high levels of risky sex reported from rural areas is clearly important in considering the epidemic potential of HIV in India. Throughout India, less than a third of the general population who report sex with non-regular partners report using condoms consistently with them, with notable exceptions in Maharashtra and Goa-Daman-Diu. Both urban and rural dwellers in these two states report comparatively high levels of consistent condom use with non-regular partners. In urban areas, well over half of the general population report such condom use, although reported condom use is in general at least 20% higher among males than females. In Goa-Daman-Diu, male reports of consistent condom use with non-regular partners in rural areas is relatively high (70%), while no females report such condom use. In Maharashtra, a slightly greater proportion of rural females (42%) than rural males (38%) report using condoms with non-regular partners. E.6 Injecting drug users Drug use patterns vary greatly across the South Asia Region. Sri Lanka has about 40,000 50,000 drug users (author unknown, 2003). One cross-sectional study highlights that among individuals arrested for drug-related offences, almost three-quarters of males and all females preferred heroin, and nearly 60% of males and over 95% of females had used drugs for at least six years. Among those imprisoned for such charges, only 2% of heroin users injected the drug. Monitoring usage patterns and choices of drugs in Sri Lanka is therefore very important. For example, most Bangladeshi IDUs began using drugs through other modes and eventually adopted injection as their primary one (NASP, 2004). Heroin was introduced to Nepal in the

24 1960s, but buprenorphine (commonly known as Tidigesic) replaced it as the drug of choice in the 1990s because of its lower price (one-eighth of the cost of heroin) and comparable ability to induce a high after injection (Reid and Costigan, 2002). While the Kathmandu Valley has an estimated 5,000 IDUs, Pokhara in the Western region has about 600, and Jhapa in the Eastern region has approximately 2,300 (personal communication with FHI). The drug supplies for the latter two regions are imported from bordering towns in India. Those along India s boundary with Bangladesh could also have played a role in the emergence of IDUs there, estimated at 20,000 in Dhaka, Rajshahi and other towns in the border areas (personal communication with the Department of Narcotics Control). Straddled between Afghanistan and India, Pakistan has about 500,000 chronic heroin users, of whom 15% report injection as their primary mode of use (UNODC, 2002). Mapping studies show that like FSWs and MSM, IDUs are highly concentrated in Karachi, with as high as 3,200 home-based IDUs and almost four times as many street-based IDUs (NACP, 2005a). Cross-sectional studies from Pakistan (Altaf, 2003; Ghauri, 2003; Zafar, 2003) and behavioral surveillance from India (NACO, 2001c) show that most IDUs use and exchange drugs and equipment, increasing their HIV risk. A substantial proportion of IDUs in Pakistan (between 40 77%) and in India (between 50 70%, except in West Bengal) used a needle/syringe previously used by someone else. Between about 60 80% inject drugs in a group setting, where group members more likely share drugs and equipment. Well over half of IDUs, except in West Bengal, lent, rented and/or sold a used needle/syringe in the past one month. A consistently greater proportion of IDUs in Manipur and Tamil Nadu also reported using pre-filled syringes (almost 30% in both versus 20% overall), using syringes squirted from another (slightly over 40% in Manipur and 50% in Tamil Nadu versus 31% overall) as well as sharing cookers, vials, containers, filters and/or rinse water (about 65% in both versus 47% overall) in the past one month. A high proportion of IDUs in Manipur (68%) and Tamil Nadu (62%), but especially in Maharashtra (82%), reported drawing drug solution from a common container in the past one month. Both Bangladesh and Pakistan feature street doctors or professional injectors who receive payment from IDUs for injections but typically use the same needle/syringe for multiple IDUs. In Pakistan, a substantial proportion of IDUs from as low as 37% to as high as 80% received injections from street doctors or professional injectors. E.7 Female sex workers, clients, and MSM who also inject drugs While FSWs and MSM are considered high-risk for HIV transmission in their own right, they can also increase their risk for HIV transmission by injecting drugs. One cross-sectional study examined sex work and drug use in Pakistan (NACP, 2005b), where 3% of the FSW sample were also IDUs, probably through exposure to drug use through their sex partners: 21% reported sex with IDU clients, and 15% sex with non-paying IDU partners. From the NACO BSS in India, about 6% of FSWs overall reported having ever tried any ( illicit ) drug, and about onethird of these (2% overall) report injecting drugs in the past year. Throughout India, about 2% of FSW clients also reported injecting drugs in the past year. The corresponding figures for both female sex workers and clients are much higher of course in the northeast: in Manipur, they are about 19% for FSW and 11% for clients. For FSWs and their clients, injection drug use appears to be more common among sex workers outside than inside brothels. By state, Manipur has the highest proportion of FSWs and clients who report injecting drugs. Among MSM, those in Karnataka and Tamil Nadu-Pondicherry accounted for most of the IDU behavior observed overall (NACO 2001b for FSWs and their clients; NACO, 2001c for MSM)

25 E.8 IDUs who also have risky sex Conversely, while IDUs constitute a high-risk group due to their drug and equipment sharing behaviors, they can also increase their HIV risk by practicing unsafe sex. One cross-sectional study in Pokhara, Nepal found that one in three IDUs had sex with at least one FSW in the past 12 months, and 70% of these had sex with more than two (FHI and NCASC, 2002; FHI, New Era and SACTS, 2003). A cross-sectional studies from Pakistan (Altaf, 2003; Ghauri, 2003; Zafar, 2003), at least one-fifth of the samples reported having sex with FSWs. Except for one study, well over half of the IDU participants had never used condoms with any previous sex partner. India s behavioral surveillance collected similar data by different types of partners (NACO, 2001c). Greater proportions of IDUs in every state reported having sex with regular partners than with FSW or non-regular partners. Possibly because IDUs also attribute a lower risk to greater regularity of their particular sex partners, they are less likely to report using condoms consistently with regular partners than with FSW or non-regular partners

26 F. HIV PREVALENCE AND SPREAD F.1 Female sex workers Data on HIV prevalence among this sub-population underscore the remarkable heterogeneity of HIV in the South Asia region. In India, HIV prevalence among FSWs sampled in Bangalore, Karnataka was 14% in 2003 and 22% in 2004, but it is not clear if the populations were comparable. A recent survey of FSWs in Mysore, also in Karnataka, found an HIV prevalence of about 25% (Karnataka Health Promotion Trust, 2004). In Mumbai, HIV prevalence among FSWs has ranged from 45-60% from (NACO, 2005). Some segments of Nepal have an advanced HIV epidemic among FSWs. The 2001 HIV prevalence among FSWs in Kathmandu Valley (16%) is slightly higher than the figure among Karnataka s FSWs in 2004 (NCASC and FHI, 2003). It is, however, nearly three times higher for Nepali FSWs who had worked in India other than Mumbai (41%) and almost five times higher for those who had worked in Mumbai (72%) (FHI, New Era and SACTS, 2002b). Every year, about 5,000 7,000 Nepali girls are trafficked to India, where at least 20,000 of the FSWs come from Nepal (Furber, Newell and Lubben, 2002). Given the high HIV prevalence among FSWs in India, especially Mumbai, as well as the additional marginalisation that Nepali girls and women would face, these data suggest that for working in India greatly increases their risk for HIV infection. HIV infection rates of 0.5% and 0%, respectively, were reported among sex workers in Lahore and Karachi, Pakistan (NACP, 2005). With an emerging epidemic, Bangladesh disaggregated its surveillance of FSW by the type of site of sex work. In the sixth surveillance round in 2005, HIV prevalence was 0.2% among brothel-based FSWs, 0-0.2% among street-based sex workers, 0-0.6% among hotel-based sex workers and 0-1.7% among casual sex workers limited to the North-West-K district (NASP, 2005). The prevalence among all categories of sex workers in Bangladesh is much lower than among comparable populations in India (NACO, 2005). Classified as having a low-level epidemic, Sri Lanka s annual rounds of sentinel surveillance in each province between 2000 and 2004 found that HIV prevalence among FSWs has generally not exceeded 1% (NACSP, 2005). Similarly, HIV among FSWs in Pakistan remains low, at 0% in Karachi and 0.5% in Lahore (NACP, 2005). Moreover, a significant nexus between FSWs and IDUs means that injection drug use may have contributed to the HIV infections observed. F.2 Men who have sex with men Data on HIV prevalence among men who have sex with men in the South Asia region are scarce, but more data are emerging. Bio-behavioral surveillance in Pakistan revealed that HIV prevalence was 2% among hijra and 4% among MSWs in Karachi and 0% among hijra and 0.5% among MSWs in Lahore (NACP, 2005). Sentinel surveillance data from Bangladesh indicate that, similar to FSWs, the HIV epidemic remains at a low level among MSM, with a prevalence of 0% among MSM, 0.3% among male sex workers and 0.8% among hijra (NASP, 2005). Much higher rates are seen from sentinel surveillance in Goa, Mumbai and Tamil Nadu in India, ranging from 2% to over 50% in populations at different times, but there are extreme fluctuations in prevalence rates from year to year, likely due to a combination of small sample

27 sizes and different populations being sampled. Much more additional data are needed to understand the extent to which the HIV epidemic has reached the MSM community in general. F.3 Injecting drug users The efficiency of HIV transmission through IDU behaviors and networks can accelerate the growth of HIV epidemics among IDUs. For example, during 2004, serological surveillance data collected by provincial AIDS control programs in Pakistan found an increase in HIV prevalence among IDUs from 0.4% at the beginning of the year to 10% by December. This is an indication of how rapidly and explosively HIV can spread through injection drug use. One study in Nepal also documented rapid rises in HIV prevalence among IDU between 1991 and 2002: from 2% to 68% in the Kathmandu Valley and from 0 to 22% in Pokhara of the Western region. Jhapa, a town in the Eastern region with a significant IDU community, had an HIV prevalence of 30% among IDUs in The graph below shows HIV increasing from 0% in 1994 to 50% in 1998 in Kathmandu. HIV infection among injecting drug users in Kathmandu, Nepal Sentinel surveillance from the Bangladesh HSS showed that the prevalence of HIV grew for several years among injecting drug users in central Bangladesh, while remaining low elsewhere in the northwest, illustrating both the potential and heterogeneity of transmission in the country NASCP, 2005). However, in the recent sixth round, it has started to rise in both the southeast and northwest, indicating that low infection rates among injection drug users offer at best a brief window for prevention programming

28 HIV Infection among Injecting drug users in Bangladesh Round II Round III Round IV Round V Round VI Central A SE D NW F1 HIV infection occurs largely among male injection drug users at present, though elevated rates were also observed among male heroin smokers. 10 HIV infection among male and female injecting drug users and male heroin smokers in Central-A district Male IDU Female IDU Male heroin smokers Central A1 Central A2 Central A HIV infection was highly concentrated in two neighborhoods, underscoring the local heterogeneity of HIV transmission, the potential for rapid diffusion once HIV is injected into drug-using neighborhoods, and also the opportunity for concentrated prevention and treatment programs

29 HIV infection among neighborhoods in Central-A district Area (n) HIV, n (%) Area - 1 (157) 14 (8.9) Area - 2 (94) 2 (2.1) Area - 3 (59) 0 Area - 4 (25) 0 Area - 5 (24) 0 Area - 6 (32) 0 Area - 7 (13) 0 Total (404) 16 (4.0) Female injecting drug users currently have low rates of HIV infection, which may be a major reason why HIV transmission in sex work remains limited. There is clearly a major nexus between injecting drug use and sex work. Over 60% of female injecting drug users were sex workers and they reported an average of 10 clients in the past week. Significant numbers were also married or had boyfriends. Female injecting drug users in Bangladesh Sex worker 61.8% Married 47.8% Currently living with regular sex partner 50% Living area is slum 50% Living on their own 22.1% Principal source of income in last six months is from selling sex 44.9% Mean number of clients (new/regular) 10.5 last wk (among those with clients) By 2000, India s sentinel surveillance program had established its northeastern region as experiencing IDU-driven HIV epidemics, with HIV infection rates among IDUs in Manipur of over 50%, and high rates in Nagaland and Mizoram as well. Prevalence rates in Manipur have declined since then, but are still over 20%. As shown by behavioral surveillance, a substantial proportion of IDUs in Manipur inject with used needles/syringes and pre-filled/squirted syringes, share equipment and engage in other risky network drug use practices, fuelling HIV epidemic growth. In some northeastern districts bordering Burma, which has Asia s most severe HIV epidemic among injecting drug users, antenatal HIV rates of 8% have been reported, which illustrates how injecting drug use can ignite a wider HIV epidemic, quite literally injecting HIV into sexual networks. HIV prevalence rates are also high among IDUs in Mumbai, Tamil Nadu and Delhi (NACO, 2005)

30 F.4 High risk male groups HIV infection among truckers in several sites in India rose steeply from 1994 to 1997, suggesting that truckers are at greater risk than the general male community and demonstrating the potential for a significant increase among men at elevated risk. While truckers continue to be at greater risk, large-scale targeted interventions that have promoted safer sexual practices may subsequently have reduced their risk, particularly in Tamil Nadu (APAC, 2005). HIV among truckers in selected Indian sites Tamil Nadu Tiruchirapalli Chennai Namakkal In Bangladesh, HIV remains low among male risk groups. As noted above, 0.5% of male heroin smokers and 0-0.8% of men-having-sex-with-men were HIV-positive. Among large samples of male truckers, rickshaw pullers and dockworkers, HIV has remained relatively stable, ranging from 0.4 to 0.6% (NACP, 2005) HIV prevalence among male risk groups in Bangladesh Round I Round II Round III Round IV Round V Round VI

31 F.5 General population Most estimates of HIV prevalence in the general population derive from sentinel surveillance among women attending antenatal clinics. Such data may be biased in both directions. For example, in many places in India, tubal ligation is common among women, and the mean age of tubal ligation may be as low as 25. ANC attenders may therefore represent a younger age distribution than the general population, which could result in bias towards higher HIV rates than in the general population. On the other hand, sex workers, particularly older ones, may be underrepresented in ANC populations, which may result in bias towards lower HIV rates than in the general population. Similarly, antenatal sites tend to over-sample urban areas, which may lead to higher estimates. Moreover, men are excluded from such estimates, and they may have higher HIV rates. Global experience derived largely from Africa, suggests that antenatal data may overestimate general population prevalence, partly because HIV rates are higher among women than men. However, it is uncertain whether such trends also apply in Asia, where HIV rates may be higher among men. In at least two Indian sites, it is possible to compare antenatal and population-based data. In Tamil Nadu, a comparison of several years of antenatal data and one round of populationbased data suggests that the antenatal and population-based data may be broadly comparable, when combined male and female population-based prevalence is compared. However, a gender analysis shows that HIV is far higher among males, which suggests that antenatal data may underestimate infection rates among populations comprising men and women (APAC, 2005). ANC and population-based HIV prevalence in Tamil Nadu, India ANC ANC ANC ANC ANC ANC ANC POP

32 Population-based HIV prevalence among men and women in Tamil Nadu, India Total Men Women Data from a high prevalence taluka in northern Karnataka suggest that antenatal and populationbased HIV prevalence levels may be broadly comparable (Blanchard et al, 2005). HIV prevalence in Bagalkot, Karnataka, India ANC-district ANC sub-district Pop-male Pop-female In India, there are a wealth of HIV prevalence data from ANC clients, collected over several years through the NACO HIV sentinel surveillance system described above, but there are few true population-based estimates (NACO, 2005). In the four southern states where the HIV epidemic has traditionally been considered most advanced (Andhra Pradesh, Karnataka, Maharashtra and Tamil Nadu), HIV prevalence among ANC attenders has generally ranged in the 1-2% range since 2000, with prevalences in Tamil Nadu seeming to decline in recent years

33 A recent analysis has suggested that HIV prevalence as measured through ANC attender populations in the southernmost states may be declining, or at minimum stabilizing, whereas it may be increasing slightly in northern states, although the epidemic there still remains at a relatively low level. In the northeastern states of Manipur and Nagaland, where injection drug use is a key driver of the HIV epidemic, HIV prevalence in ANC populations has also ranged from 1-1.5%. Because of changes over time in surveillance sites and sampling issues, ascertainment of true trends is difficult. In a population-based survey conducted in 2003 in Bagalkot District in northern Karnataka, which is largely a rural district, HIV prevalence among males was 3.3% and among females 2.4% (see Figure above). The prevalence among ANC attenders in the district was 2% among women sampled in the district headquarters hospital and 4.2% in a sub-district government hospital. A very important consideration is that there is considerable variability in HIV prevalence among districts within states, and even at sub-district and village level. This is illustrated, for example, by the enormous variability in HIV rates in the state of Karnataka, with HIV prevalence of under 1% in roughly half of the districts, but with some districts in the northern part of the state having prevalences of over 3%. In the populationbased survey in Bagalkot District, HIV prevalence in the three sub-districts sampled ranged from 1.2% to 4.9%, and in 10 villages sampled from 0 to 8.2%, with four villages having an HIV prevalence under 2% and three villages over 6%. HIV prevalence among rural residents (3.6%) was higher than among urban residents (2.4%). There are no general population HIV prevalence data available from Nepal, which vitiates a confident interpretation of the country s HIV epidemic. In Pakistan, professional blood donors constitute the closest approximation to a general population samples. Notwithstanding the obvious potential upward bias, HIV infection rates among professional blood donors remain close to zero. In Bangladesh, no HIV infections were found in studies conducted in recent years in populations of rural men and women living in the Matlab region and among a general population of women attending a health care clinic in Dhaka, (NASCP, 2005). Data from Sri Lanka reinforce it being categorized as a country with a low-level epidemic, where in 2004, HIV prevalence among ANC attenders was 0.14%, despite significant levels of HSV-2 (NACSP, 2005). As noted above, Bhutan s 2004 round of sentinel surveillance included 10 population groups, including ANC attendees, blood donors, STI patients, armed forces, and drivers, prisoners and female sex workers. The sample size of sex workers and drivers was too small to draw any conclusion on prevalence. For other general population groups, the prevalence rate among ANC attendees was 0.04% and among blood donors 0%. There are no general population data for Afghanistan or the Maldives. G. COUNTRY-SPECIFIC ANALYSES G.1 India G.1.1 Epidemic overview With approximately 40% of Asia s population, India has over 60% of the continent s estimated HIV infections (UNAIDS, 2004)

34 HIV in Asia Other 10% Thailand 11% Burma 5% China 13% India 61% The heterogeneity of the HIV epidemic in India is critically important. Given India s size and complexity, it is best analyzed as a continent, some of whose states are larger than many African countries, and many of whose districts are larger than some African countries. When its size and diversity are acknowledged in this manner, the heterogeneity of its epidemic becomes easier to recognize. India s epidemic is largely concentrated in seven states with over 1% antenatal prevalence in the south, west and northeast: Tamil Nadu, Karnataka, Andhra Pradesh, Maharashtra, Goa, Manipur and Nagaland (NACO, 2005)

35 W HIV prevalence by state in India A M N M M K G I P A A I H A O N N G Z N A D I A These seven states have less than 30% of India s population and almost 70% of its HIV cases. As Kumar et al. s graph below shows, HIV levels in the high prevalence states in the south and west and northeast are at least fivefold higher than the rest of India (Kumar et al, 2005)

36 HIV prevalence in state clusters in India TN-AP-KN-M H NORTH-EAST REST As noted above, theses differences in HIV prevalence are mirrored by differences in sexual behavior. As the following graphs from Kumar et al. (2005) indicate, men and women in the high prevalence states tend to report a higher number of sexual partners in the last year. Sexual partners in last year among men in state clusters in India TN-AP-KN-MH NORTH-EAST REST

37 Sexual partners in last year among women in state clusters in India TN-AP-KN-MH NORTH-EAST REST Further analysis of India s epidemic suggests that HIV infection is also concentrated in a number of high prevalence districts in the high prevalence states identified above. The following NACO map identifies high prevalence districts in the south and west and northeast, although the paucity of district level data from northern India is of concern (NACO, 2005)

38 HIV prevalence by district in India HIV prevalence at ANC sites < 1.0% HIV prevalence at ANC sites % HIV prevalence at ANC sites % HIV prevalence at ANC sites % HIV prevalence at ANC sites % HIV prevalence at ANC sites > 3.0%

39 The above district map indicates that there are three major foci for HIV transmission in India, specifically: 1. The Northern Karnataka-Maharashtra corridor 2. Coastal Andhra Pradesh 3. High drug using northeastern states bordering Burma These trends are illustrated by the following district level ANC data (and one voluntary blood donor sample) from each of the above high transmission areas (NACO, 2005): HIV prevalence among pregnant women in Karnataka-Maharashtra corridor, India Bagalkot Belgaum Dharwad Pune HIV prevalence among pregnant women in Coastal Andhra Pradesh, India E Godavari W Godavari Guntur Prakasam HIV prevalence among pregnant women and voluntary blood donors in North-Eastern States, India Churachandpur Chandel CHC Hospital, Hospital, Manipur Manipur Tuensang Hospital, Nagaland Aiz awl Ukhurl Hospital, Voluntary Blood Manipur Donors

40 G.1.2 A case study in heterogeneity Bagalkot District, India As emphasized earlier, a hallmark of the HIV epidemic in India is its heterogeneity, with large observed differences in HIV prevalence among states, and among districts within states. One possible explanation is that epidemics are at different phases due to later introduction of HIV into some areas, although with the high level of mobility in India, this explanation is becoming increasingly implausible. An alternative explanation is that differences in the underlying sexual structure (sexual behaviors and networks) between locations drive the observed epidemiological heterogeneity. To better understand possible sources of heterogeneity in HIV transmission dynamics, we have undertaken a detailed analysis of sexual structure and HIV epidemiology in Bagalkot district, Karnataka State, in south India. Bagalkot is a district in northern Karnataka with a population of 1.65 million, with 71% living in rural areas. Its main industries are agriculture and mining, and there are substantial differences in economic activity within the district, since some districts are irrigated and others are more drought-prone. Sixty-five percent of workers in the district are agricultural cultivators or laborers. Bagalkot is divided into six talukas or sub-districts, which are the main administrative sub-units within the district. The present analysis focuses on three adjacent talukas with a combined population of 810,100 distributed across 241 villages and six urban centres (see map)

41 Bagalkot District, Karnataka State, India Bagalkot district Bagalkot District Karnataka State Three sources of data were used: 1) a detailed mapping and enumeration of female sex workers (FSWs) in all 247 villages and urban areas; 2) a set of polling booth surveys of sexual behaviors in a random sample of village-dwelling men; and 3) a population-based sample survey of HIV prevalence. HIV Epidemiology: Based on routine sentinel surveillance data from (n=2400), the overall HIV prevalence among women attending antenatal clinics was 3.1%. Interestingly, the prevalence did not differ significantly between women in rural areas (3.4%) and urban areas (3.0%). A community-based HIV prevalence study was conducted in 2003 in the three focus talukas based on a random cluster sampling of 10 villages and 20 urban blocks (total sample size of 4,007). The overall HIV prevalence was 2.9%, and was higher among males than females (p=0.10), and significantly higher in rural areas than in urban areas (p<0.05). HIV prevalence in a community based survey of 3 talukas in Bagalkot, Karnataka, India HIV prevalence (%) Total Male Female Rural Urban