Secondary Logo

Journal Logo

Rapid Communication

Risk Factors for HIV Infection in Injection Drug Users and Evidence for Onward Transmission of HIV to Their Sexual Partners in Chennai, India

Panda, Samiran MD, DTM&H*; Kumar, M Suresh MD, DPM, MPH*; Lokabiraman, S MSW, MBA*; Jayashree, K MBBS, PhD*; Satagopan, M C MSc; Solomon, Suniti MD; Rao, Usha Anand MSc, PhD§; Rangaiyan, Gurumurthy MA, PhD; Flessenkaemper, Sabine MD, MSc, DTM&H; Grosskurth, Heiner MD, PhD; Gupte, Mohan D MD, DPH

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: May 1, 2005 - Volume 39 - Issue 1 - p 9-15
doi: 10.1097/01.qai.0000160713.94203.9b
  • Free


By the end of 2003, 2.2-7.6 million people including adults and children have been estimated to be living with HIV in India,1 a number rating second only to South Africa. Continuously increasing infection rates and documented diffusion of injection drug use in various parts of India2 have raised concerns against this backdrop about the effect of injection drug use on the larger HIV epidemic in the country,3 which is predominantly heterosexual. However, HIV intervention in injection drug users (IDUs) is yet to receive adequate research attention in the country. The only exception to this is the northeastern states of India having common border with Myanmar. The Indian Council of Medical Research4 documented explosive HIV epidemics among IDUs in 2 of these states, namely Manipur and Nagaland in the early 1990s. This triggered a series of targeted interventions in the region. Most of these interventions, however, focused on safer injection practices and little attention was paid to the sexual risk for HIV infection in regular and casual sex partners of IDUs. HIV risks among IDUs resulting from unsafe sexual practices also did not get much attention in these initiatives, probably reflecting the long-held belief that injectors have low levels of sexual activity.

Experiences accumulated over the past few years, however, underscore that the primary risk behaviors that should be included in HIV surveillance studies among IDUs are “sharing” of drug injection equipment, the potential for rapid partner change among risk partners, and sexual risk behavior.5 A study from northern Thailand adjacent to the Golden Triangle recorded high HIV prevalence among IDUs who had had sex with commercial sex workers compared with those who had not (35% vs. 23%; odds ratio [OR] 1.8, 95% CI 1.2-2.7), which corroborates the assertion that the risk of IDUs getting HIV through risky sex could be high.6 Conversely, a behavioral study from Indonesia found the potential for the sexual spread of HIV from IDUs to noninjectors to be extremely high.7 Studies examining the transmission of HIV from IDUs to their sex partners are, however, sparse and the one that came with convincing evidence in this regard in the late 1990s was from Manipur, the northeastern state of India bordering Myanmar where 45% of the wives of HIV-positive IDUs were found to be infected with the virus.8 None of the women in this cohort of wives was an IDU. We recently conducted a cross-sectional study in central Chennai among IDUs and their regular sex partners to determine the HIV-discordant partnerships and to collect information on their drug use and sexual practices that could inform future intervention design. The present article describes the findings of this study.



The study population comprised IDUs and their regular sex partners including spouses who lived in central Chennai. Anybody ever injecting drugs was termed “IDU” for the study. Those who reported having injected drugs within the previous 6 months were considered as current IDUs. Peer outreach workers and field researchers recruited the study participants over a period of 3 months from April-July 2003. Drug-using clusters were mapped in central Chennai using snowballing technique. This and the involvement of the outreach workers from different parts of central Chennai ensured that almost all the married IDUs and IDUs in live-in relationships in central Chennai were enrolled.

Nature of the Study

After informed consent was obtained, the study participants were interviewed at field-based clinics. Information was collected on their sociodemographic profile, initiation of drug use, switching to injection drug, injection equipment sharing practices, sexual practices in and outside marriage, risk perception, and knowledge of HIV/AIDS by using a field-tested, semistructured questionnaire. The study received ethical approval from the institutional review board of the Population Council, New York, and the ethical committee of the local nongovernmental organization SAHAI Trust, in Chennai. Each study participant donated blood for anonymous HIV testing and those willing to know their HIV status were supported in access to free HIV voluntary confidential counseling and testing (VCCT) services.

Laboratory Tests and Statistical Analysis

Serum samples were tested for HIV antibody by enzyme-linked immunosorbent Assay (ELISA) (Biomerieux BV, Boxtel, The Netherlands) and Vironostika HIV Uni-Form II antigen/antibody (Vironostika USA, Hazelwood, MO) at YRG Care, Chennai. Any first positive HIV test result was confirmed by a second ELISA system that used different antigen coating on the wells. Ten percent of the stored serum samples were sent for quality control to the reference laboratory at ALM Post Graduate Institute of Basic Medical Sciences, Chennai.

Software packages Epi-Info (version 6.4b, Centers for Disease Control, Atlanta, GA, in collaboration with World Health Organization, Geneva, Switzerland) and Systat (version 8.0 SPSS, Chicago, IL) were used for data analysis. Although the present study was cross-sectional in nature, the study participants provided information on the past exposures. We calculated the ORs for different risks factors for HIV infection, taking HIV-positive IDUs as cases and HIV-negative IDUs as controls. Factors with biologic plausibility and having significance (P ≤ 0.1) in univariate analysis were subsequently entered in a multivariate logistic regression model to identify the variables associated with HIV infection in IDUs. Detailed statistical analysis for determinants of HIV infection in women could not be conducted due to the low number of HIV infections in them.


A total of 260 IDUs and their regular female sex partners were enrolled for interviews and collection of blood samples at the field-based clinics. Data from the 34 couples were incomplete. While in some of these 34 cases, IDUs were intoxicated and withdrew themselves from the middle of the interviews, on some occasions female sex partners of IDUs did not give consent for blood collection. Analysis was performed on the data from 226 IDUs (87% of the enrolled population; 144 were current and 82 ex-IDUs) and their regular sex partners (202 wives and 24 in live-in relationships). No difference was observed between sociodemographic profile of those who participated in the study and those who withdrew.

IDUs recruited in the study had a mean age of 36 years (SD ± 6; range 22-55 years; median 36). The mean age of the regular sex partners of the IDUs including spouses was comparatively lower (30 years, SD ± 6; range 18-48 years, median 30). The mean duration of marital relationship of couples was 12 years (SD ± 7; range 1-31 years, median 12 years). While 27% of women (62/226) were illiterate and 34% (78/226) had primary or lower level of education, the figures were 8 and 41%, respectively, for IDUs. A little over 70% of the women were Hindus and one-fourth were Christians, the rest Muslims. While 29% (65/226) of the women reported living in their own concrete houses and 11 in their own mud houses, the majority were in rented residences; 3% (7/226) reported being homeless. Ninety percent of the women during study were living with their IDU partners and the rest reported intermittent stay or stay during weekends due to family feuds. Fewer women participants (17/226; 7%) compared with IDUs (184/226; 81%) reported having sex outside marriage, of whom the majority (14/17; 82%) had sex with only one extramarital partner whereas 63% of the IDUs having sex outside marriage had sex with ≥5 different partners. A little over one-third of the women were housewives and a similar proportion worked as housemaids. The next most common occupations of the female participants in the study were flower selling, sweeping streets, and carpentry work; only one reported dealing drugs.

All the IDUs except 2 had reported ever using heroin and the mean age at onset of heroin use was 24 years (SD ± 6; range 12-43 years; median 23); one-fifth of the IDUs started using heroin before they were in their 20s. Ninety percent of those who had reported ever using heroin had used it first through smoking and the rest had injected heroin when they were using it for the first time. Many of the IDUs who had started using heroin through smoking subsequently injected heroin too (208/224; 93%) at a mean age of 29 years (SD ± 6; range 14-45 years, median 28). As many as 90% of the IDUs (204/226) had reported injecting buprenorphine purchased from the streets. This indicates that almost all the heroin injectors had also injected buprenorphine at some time. Only 18 IDUs were exclusive buprenorphine injectors among our study population. None of the regular sex partners or wives in our study had ever injected drugs.

The overall HIV prevalence was 30% (68/226) in male participants (all IDUs) and 5% (11/226) in their female sex partners, all being sexual partners of infected men. While 36% (82/226) of the IDUs were ex-users who had last injected drug 6 months previously, 38% (54/144) and 22% (32/144) of the current IDUs reported “not injecting daily” and “injecting drug once a day,” respectively. While 23% of the current IDUs injected drugs twice a day, 17% (25/144) of the current IDUs had reported injecting drug more frequently. Borrowing injection equipment from others while taking drugs during the previous 2 weeks was reported by 29% (42/144) of the current IDUs, of whom 25 were HIV negative. More than 50% of those (24/42) who had reported borrowing injection equipment from others within the previous 2 weeks borrowed it from ≥3 different persons. Of 57 IDUs (57/226; 25%) who had reported lending injection equipment to others during the last injecting episode, 22 were HIV infected.

Sixteen percent of the regular sex partners of HIV-positive IDUs in our study were HIV positive (11/68). In 25% of the couples (57/226) recruited in the study, the HIV test results were discordant and only the male partners were HIV positive. While 5% (11/226) of the couples were concordant for HIV-positive test results, the remaining couples (158/226; 70%) were all HIV negative. However, HIV risk perception was very low in the study population and this was independent of their HIV status. In fact, while 57% (39/68) of the HIV-positive IDUs thought that they had “no chance” or “very little chance” of getting HIV, 45% of the HIV-infected women (5/11) had a similar perception. Overall, 48% (109/226) of the IDUs and 52% of their regular sex partners (117/226) did not perceive that they had any chance of being infected with HIV. When asked about HIV risk perception, only 5 of 68 HIV-infected IDUs (7%) and 1 of 11 HIV-infected women (9%) had volunteered that they were HIV infected. None of the HIV-negative study participants volunteered their HIV status. This low self-reported HIV serostatus, taken together with low HIV risk perception, probably reflects a low HIV test uptake in the study population as well.

Univariate and Multivariate Analysis Results

Univariate analysis among wives of all the HIV-positive IDUs (68), considering HIV-positive women in this subgroup as cases (11/68) and the rest as controls (57/68), identified only one factor as associated with HIV infection in women, which was “duration of relationship.” Women reporting >6 years of relationship had a greater probability (P = 0.02) of being HIV positive (6/16; 37%) compared with their counterparts (5/52; 10%) who had reported lesser duration of relationship in this subgroup analysis. Because of the relative infrequency of HIV infection in the women participants (11/226), we limited the remaining statistical analysis to the determinants of HIV infection in IDUs only.

Univariate association between key risk factors and HIV-positive test result among IDUs recruited in the study is shown in Table 1. Neither age nor educational status nor age at onset of drug use was associated with HIV-positive test result (P > 0.05). No significant difference was observed (P = 0.1) between HIV prevalence in the ex-IDUs (19/82; 23%) and that in the current IDUs (49/144; 34%). Although nearly 60% of the IDUs (137/226) reported ever having sex with female commercial sex workers, and 30% (22/137) reported having had commercial sex as their first sexual intercourse in life, these practices did not have any association with their current HIV status. Conversely, 63% of the IDUs (62/98) who had tattoos were HIV positive, compared with 5% HIV prevalence (6/128) among those who did not have tattoos (P < 0.001; Table 1).

Comparison of risk factors between HIV=positive and HIV=negative IDUs in Chennai, India: univariate analysis results
(Continued) Comparison of risk factors between HIV=positive and HIV=negative IDUs in Chennai, India: univariate analysis results

The statistically significant association of “tattoos” with HIV infection in IDUs prompted subsequent investigation that revealed that the persons applying tattoos carry out their business from roadside pavements or by roaming the streets in Chennai. They also display their wares in festivals and weekly bazaars. Instead of battery-operated machines used by male tattoo makers, women belonging to nomadic tribes who are also in such businesses manually prick the skin repeatedly with solid sharp needles dipped in green dye or ink along the lines of the design drawn on the chosen part of the body. The time taken to complete a tattoo varies from 15 minutes to 1 hour, depending on the size of the design. Some wipe off the oozing blood with a strip of cloth, which has seen many usages. Some even advise not to wipe off the blood and let the wound heal by itself. Talking with an injector revealed more details about tattooing in the city. He recalled being the 5th person in the line. Every one took their turn and once the above-mentioned process was completed, the person applying the tattoo did not change the needle or clean it at all. The needle was in fact not even kept immersed in any kind of solution. This appeared to be the common practice in this business followed by most of the tattoo makers.

HIV prevalence was higher among IDUs who were “ever in jail” (48/129; 37%) than among those who were never incarcerated (20/97; 21%; P = 0.01). While 36% (47/132) of the injectors who had reported “ever injecting drug in drug-selling places” were HIV infected, only 22% (21/94) of the IDUs who did not report such practice contracted the virus-a statistically significant difference (P = 0.04). IDUs having injecting frequency of more than twice a day also had higher HIV prevalence compared with the relatively infrequent injectors (Table 1). Moreover, HIV prevalence was significantly higher among the IDUs who were recruited from the northern part of central Chennai (47/103; 46%) compared with that in IDUs recruited from the eastern (13/58; 22%; P < 0.01) or western parts of the city (8/65; 12%; P < 0.001). Attributable risks for HIV infection among IDUs were 14% for injecting practices in drug-selling places, 58% for tattooing, 5% for those who had been injecting drug more than twice a day, and 16% for “ever being in jail.” Marginally higher HIV prevalence was seen in IDUs who had ever been admitted to addiction treatment centers (Table 1) compared with those who had not, but this was not statistically significant. Similarly, current IDUs had higher HIV prevalence compared with that in ex-IDUs but this did not achieve statistical significance; it was noticed, however, that more current IDUs had reported being ever in addiction treatment centers (72/144; 50%) compared with ex-IDUs (29/82; 35%; P = 0.03).

In a multivariate model (Table 2), IDUs in the northern part of central Chennai had 6 times the odds of being HIV positive (OR 5.72; 95% CI 2.40-13.63) compared with those recruited from the western part of central Chennai (used as reference site), and IDUs who had “ever injected drug in drug-selling places” had twice the odds of being HIV positive than their counterparts (OR 2.11; 95% CI 1.08-4.12). However, the relationship between “ever being in jail and having higher HIV prevalence” and “having tattoo and higher proportion of them being HIV-positive” was weakened (P = 0.07) slightly in the multivariate model compared with the univariate analysis findings. Injecting frequency was not picked up as an independent variable for HIV infection among IDUs in the multivariate model. The multivariate analysis findings together with the HIV seroprevalence data in couples highlighted immediate HIV intervention needs as discussed below.

Adjusted OR and 95% CI for variables predicting HIV infection in IDUs in Chennai, India


The number of the IDUs and their regular sex partners that could be enrolled from central Chennai was low compared with the expected number from an earlier estimated size of IDUs for the whole of Chennai,9 which was one of the limitations of the present study. The HIV prevalence in women also did not allow a detailed examination of the determinants of transmission of HIV infection from IDUs to their regular sex partners. In spite of these limitations, we believe the study has generated valuable data on HIV in IDUs and their regular sex partners, an area in which there is little published information. The study has also provided insights for future intervention design.

Once HIV prevalence among a high-risk IDU population reaches 20%, HIV epidemics can become self-perpetuating, with even modest levels of risk behavior leading to substantial rates of infection.10,11 The HIV prevalence among IDUs in our study was 30% and borrowing as well as lending of injection equipment was also high. Thirty percent of the current IDUs reported borrowing injection equipments from other IDUs within the last 2 weeks, of whom about two-thirds were HIV negative. Conversely, one-fourth of the IDUs, 22 of whom were HIV positive, reported lending injection equipment to others during the last injecting episode. This calls for an immediate intervention initiative. A high level of sharing of injection equipment is a feature of IDUs in many other cities of south Asia12,13 such as Kathmandu (in Nepal) and Lahore (in Pakistan) that is mostly guided by the poor economic situation of the IDUs. It is worth noting that one syringe and needle costs a little less than one-tenth of a dollar (US) in India. At times, law enforcement also prevents IDUs from carrying their own syringes. A situation assessment in the city of the present study earlier recorded that the police identified drug users by possession of syringes; as a result many drug users stopped carrying syringes when they went out in search of drugs. This compelled sharing practices at places outside of IDUs' houses and, in particular, at dealers' locations.14 Ensuring safer injection practices through targeted and environmental intervention should therefore gain a priority in the city of Chennai.

It is important not to overlook the social and contextual factors that may be directly or indirectly related to HIV transmission among IDUs.15,16 In the present study, we analyzed that the practice of injecting drugs in drug-selling places was associated with a 14% attributable risk of HIV infection among IDUs (adjusted OR 2.1, Table 2), which was not surprising in view of the increased chance of sharing of injection equipment by an IDU in such settings. Also, IDUs recruited from the northern part of central Chennai had a 6 times higher risk of HIV infection compared with those who were from the western part of the city. The difference between HIV prevalence among IDUs from the western compared with the eastern part of the city was, however, not significantly significant. Although it is not surprising to find such variations in large cities, the factors that could probably explain the observed variation included relatively higher poverty in the northern part of the city, more slums, and larger networks of IDUs, all of which were noticed during the fieldwork.

It is worth noticing that within approximately a 5-year period, heroin smokers reported switching to injecting in our study. This kind of change in drug use practice is generally influenced by several environmental determinants such as sudden heroin drought on the streets due to police seizure, peer influence, and pharmacodynamic factors such as increased tolerance to opiates among heroin smokers that force them to either increase the frequency of smoking or to use cheaper and more effective forms of drug intake such as injecting. Reaching out to the heroin smokers and averting this switch could therefore be another important area for intervention.

Marginally higher HIV prevalence among IDUs who had ever been admitted to addiction treatment centers (Table 1) prompted us to examine the types of drug users who went for treatment and the nature of addiction treatment centers in the city; no association was found between admission rate and earlier onset of injection drug use. However, it was revealed that a higher proportion of our study population currently injecting drugs reported ever being admitted to drug treatment centers compared with the ex-injectors, and HIV prevalence was also comparatively higher among current IDUs compared with the ex-users. This could probably explain the higher HIV prevalence in IDUs who had ever been admitted to drug treatment centers and also reflect that the current IDUs, a higher proportion of whom had ever sought admission to treatment centers, had more relapses and were at greater risk of HIV infection compared with the ex-IDUs. While 4 government-run centers presently offer free addiction treatment to drug users in Chennai, there are 6 private paying treatment centers.

The 2 additional areas for HIV intervention that stood out prominently during analysis of the study findings were “practice of tattooing” and “history of incarceration” among IDUs. While use of unsterilized needles during tattooing on multiple persons in a row as described above in the “Results” section could explain higher HIV prevalence among IDUs who had tattoos, drug injecting in jail has been least researched in India. Literature exists, however, on an HIV outbreak in a Scottish prison17,18 due to multiperson use of injection equipment by IDUs while in jail. Failure to preempt such possibilities in Indian jails and jails in the neighboring countries and inability to launch an appropriate intervention could seriously plague the national and regional AIDS control programs.

Apart from documenting a high HIV prevalence among IDUs in Chennai, we also recorded an onward transmission of HIV in their regular sex partners. While in 25% of the couples, the HIV test results were discordant and only the male partners were HIV positive, 5% of the couples were concordant for HIV-positive test results. Given the low level of knowledge and individual risk perception in both IDUs and their sex partners and high background HIV prevalence in IDUs and already recorded transmission of HIV in their regular sex partners, the HIV-negative women as well as IDUs are at risk for acquiring the infection. It is also important to note that 16% of the regular sex partners of HIV-infected IDUs tested HIV positive in the current community-based study. While this prevalence is lower compared with the documented high HIV prevalence figure of 45% among noninjecting wives of HIV-positive IDUs in Manipur,8 the need for immediate launching of intervention cannot be overemphasized. Interventions in this regard should particularly focus on individual risk appreciation and negotiation skill for safer sex practices through individual- (for women as well and not just for IDUs) and couple-oriented sessions. Making means of behavior change available to them should go hand in hand with this approach.

Finally, HIV risk perception was very low in both IDUs and their regular sex partners in our study. While 57% of the HIV-positive IDUs thought that they had “no chance” or “very little chance” of getting HIV, 45% of the HIV-infected women had a similar perception. Overall, 48% percent of the IDUs and 52% of their regular sex partners did not perceive that they had any chance of contracting the virus. When asked about HIV risk perception, <10% of the HIV-infected IDUs and women reported that they were HIV positive and none of the HIV-uninfected study participants volunteered their HIV status. This low self-reported HIV status, taken together with low HIV risk perception, probably reflects a low HIV test uptake in the study population as well. While such a low awareness about one's own HIV status has been recorded among IDUs from the countries observing recent and explosive HIV epidemics,19 similar findings from IDUs in a country like India, which has been living with the epidemic for the past 2 decades, is of concern. Presently 12 government-run and 3 private HIV VCCT centers cater to the HIV test needs in the city. None of these centers, however, is tuned to the special needs of IDUs, and drug user-friendly HIV VCCT services for street-based IDUs are run by only 2 organizations, one of which participated in the present study and could draw funding support for interventions following dissemination of the present research findings. Promoting HIV VCCT should therefore be an integral part of the comprehensive HIV intervention package for IDUs in India and other south Asian countries.


1. UNAIDS. 2004 Report on the Global AIDS Epidemic (4th Global Report). Geneva-27, Switzerland. 2004.
2. United Nations Office on Drugs and Crime (UNODC). Rapid Assessment Survey of Drug Abuse in India. New Delhi: Regional Office for South Asia and MSJE, Government of India. 2002.
3. National AIDS Control Organization (NACO). National AIDS Prevention and Control Policy; New Delhi: Ministry of Health and Family Welfare, Government of India. 2001.
4. Pal SC, Sarkar S, Naik TN, et al. Explosive epidemic of HIV infection in northeastern states of India, Manipur and Nagaland. Center of AIDS Research (CAR) Calling, ICMR. 1990;3:2-6.
5. Des Jarlais DC, Dehne K, Casabona J, et al. HIV surveillance among injecting drug users. AIDS. 2001;(Suppl 3):S13-S22.
6. Razak MH, Jittiwutikarn J, Suriyanon V, et al. HIV prevalence and risks among injection and noninjection drug users in northern Thailand: need for comprehensive HIV prevention programs. J Acquir Immune Defic Syndr. 2003;33:259-266.
7. Pisani E, Dadun, Sucahya PK, et al. Sexual behavior among injection drug users in 3 Indonesian cities carries a high potential for HIV spread to noninjectors. J Acquir Immune Defic Syndr. 2003;34:403-406.
8. Panda S, Chatterjee A, Bhattacharya SK, et al. Transmission of HIV from injecting drug users to their wives in India. Int J STD AIDS. 2000;11:468-473.
9. Dorabjee J, Samson L. A multi-center rapid assessment of injecting drug use in India. Int J Drug Policy. 2000;11:99-112.
10. Des Jarlais DC, Marmor M, Friedmann P, et al. HIV incidence among injection drug users in New York City, 1992-1997: evidence for a declining epidemic. Am J Public Health. 2000;90:352-359.
11. Holberg S. The estimated prevalence and incidence of HIV in 96 large US metropolitan areas. Am J Public Health. 1996;86:642-654.
12. Pokharel BR, Aryal S, Bhattarai A, et al. Situation Analysis of HIV/AIDS in Nepal. Teku, Kathmandu: National Center for AIDS and STD Control; 2000.
13. United Nations Office for Drug Control and Crime Prevention (UNODCCP) and UNAIDS. Baseline Study of the Relationship Between Injecting Drug Use, HIV and Hepatitis C Among Male Injecting Drug Users in Lahore. Lahore, Pakistan; 1999.
14. Kumar SM, Mudaliar S, Daniels D. Community based outreach HIV intervention for street-recruited drug users in Madras, India. Public Health Rep. 1998;113(Suppl 1):58-66.
15. Koester SK. Coping, running and paraphernalia laws: contextual variables and needle risk behavior among injection drug users in Denver. Hum Organ. 1994;53:287-295.
16. Latkin C, Mandell W, Vlahov D, et al. People and places: behavioral settings and personal network characteristics as correlates of needle sharing. J Acquir Immune Defic Syndr. 1996;13:273-280.
17. Gore SM, Bird AG, Burns SM, et al. Drug injecting and HIV prevalence in inmates of Glenochil prison. BMJ. 1995;310:1264-1265.
18. Yirrell DL, Robertson P, Goldberg DJ, et al. Molecular investigation into outbreak of HIV in a Scottish prison. BMJ. 1997;314:1446-1450.
19. Rhodes T, Lowndes C, Judd A, et al. Explosive spread and high prevalence of HIV infection among injecting drug users in Togliatti city, Russia. AIDS. 2002;F25-F31.

HIV; injection drug user; regular sex partners of injection drug users; India

© 2005 Lippincott Williams & Wilkins, Inc.