Secondary Logo

Journal Logo

Epidemiology and Social Science

Declining HIV Incidence Among Patients Attending Sexually Transmitted Infection Clinics in Pune, India

Mehendale, Sanjay M MD, MPH*; Gupte, Nikhil PhD; Paranjape, Ramesh S MSc, PhD*; Brahme, Radhika G BSc, MCM*; Kohli, Rewa MSc, PhD*; Joglekar, Neelam MSc*; Godbole, Sheela V MD*; Joshi, Smita N MBBS*; Ghate, Manisha V MBBS, DCH*; Sahay, Seema MSc, PhD*; Kumar, B Kishore MSc, PhD*; Gangakhedkar, Raman R MBBS, DCH, MPH*; Risbud, Arun R MD, MPH*; Brookmeyer, Ron S PhD; Bollinger, Robert C MD, MPH

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: August 15, 2007 - Volume 45 - Issue 5 - p 564-569
doi: 10.1097/QAI.0b013e3180d0a6ba
  • Free


India has more than 5 million people infected with HIV, the largest burden of HIV-infected people in Asia.1,2 Some expect that transmission rates are likely to remain high in India and that the epidemic is likely to follow a similar course as seen in some sub-Saharan African countries.3 Others expect that the Indian epidemic is likely to be much more limited.4 A recent study has demonstrated a decline in HIV prevalence rates among young men and women in southern India,5 suggesting for the first time that community-based HIV prevention efforts may be decreasing the risk of HIV infection in some regions of India. The results of this report have been questioned for study methodology, reliance on seroprevalence, and unwarranted data aggregation, however.6-8

The Government of India initiated the first National AIDS Control Plan (NACP-I) in 1991 and has recently announced phase 3 (NACP-III). To identify appropriate strategies for HIV prevention and control under NACP-III, it is important to confirm whether declines in HIV seroprevalence reflect actual declines in HIV transmission rates and to examine which risk factors may be associated with declines in HIV incidence. Understanding the dynamics of HIV transmission over time and confirming such findings in other settings should improve the generalizability of the data, reinforcing existing effective HIV prevention programs and scaling up prevention efforts in other regions. To date, there have been no direct estimates of HIV transmission (HIV incidence) over time from any population in India. To determine if there was evidence for a decline in HIV incidence among high-risk groups over 10 years, we analyzed HIV incidence rates among 3 specific high-risk groups of patients attending sexually transmitted infection (STI) clinics in Pune, India.


Study Design

This study was designed as a prospective, clinic-based, cohort study.

Study Population

Between 1993 and 2002, men and women attending 2 STI clinics and a women's health clinic in Pune, India were offered serologic screening for HIV-1 and HIV-2 infections. HIV-seronegative persons were invited to participate in a prospective study of HIV acquisition. Details of this prospective study have been described previously.9 Written informed consent was obtained from all the study participants, and the project was approved by the Johns Hopkins Medical School and National AIDS Research Institute Ethical Review Boards. After receiving informed consent, a structured questionnaire was used to collect data on demographics, STIs, medical history, sexual behavior, reproductive history, HIV/AIDS knowledge, and HIV prevention practices. Clinical diagnoses of genital ulcer disease (GUD), herpes simplex virus-2 (HSV-2), cervicitis, or vaginitis were based on a detailed physical examination made by the physician without knowledge of the HIV serostatus of the study participants. Patients were treated with standard therapy for STIs (diagnosed on the basis of clinical impression and microbiologic tests) using guidelines from the Centers for Disease Control and Prevention and the World Health Organization.

After patients had received counseling, sera were screened with commercially available enzyme-linked immunosorbent assay (ELISA) kits for detection of HIV-1 and HIV-2 antibodies. Serum samples were screened initially using a commercially available ELISA kit for detection of HIV-1 and HIV-2 antibodies (HIV EIA, LABSYSTEMS OY, Helsinki, Finland; Detect-HIV, BioChem ImmunoSystems, Montreal, Quebec, Canada; or Innotest HIV-1/HIV-2 Ab sp, Innogenetics NV, Zwijndrecht, Belgium). Specimens positive by ELISA were confirmed using the Rapid Test Device (Immunocomb II, HIV-1 & 2 BiSpot, Orgenics, Yavne, Israel; HIV TriDOT, J. Mitra and Company, New Delhi, India; or Capillus HIV-1/HIV-2, Cambridge Diagnostics, Galway, Ireland). Specimens that were discrepant by these assays were confirmed using a third ELISA or commercially available HIV-1/HIV-2 Western blot (INNO-LIA HIV-1/HIV-2 Ab, Innogenetics NV, Zwijndrecht, Belgium or HIV BLOT 2.2 Western Blot Assay, Genelabs Diagnostics SA, Geneva, Switzerland). Western blot results were interpreted according to Centers for Disease Control and Prevention criteria.10 For the purpose of this analysis, individuals with evidence of HIV-1 and/or HIV-2 infection were considered to be HIV infected. An identical HIV testing algorithm was used throughout the study period.

Statistical Analyses

HIV incidence was calculated by dividing the number of HIV seroconverters by the person-years (PYs) of follow-up for each of the 3 specific gender risk groups that attended our STI study clinics in Pune, namely, women who reported a history of commercial sex work (FSW), men, and other women who denied a history of commercial sex work. These non-sex worker women were almost all married and monogamous and were mostly referred to the STI clinic by their husbands, who were attending STI clinics at the advice of the study staff. Because men who have sex with men (MSM) and heterosexual men may have inherently different infection patterns, we calculated HIV incidence separately for these 2 groups of male patients with STIs. To analyze trends in HIV transmission rates over time, HIV incidence was calculated for the 3 risk groups by calendar year. We also calculated incidence rates by years of follow-up to account for cohort effects that might result from HIV risks that vary according to how long persons have been in the study. We then fitted a Poisson regression model to the HIV incidence rates (with calendar year, age, marital status, education, religion, and years of follow-up as independent variables) to obtain adjusted relative risks (RRs) of HIV incidence by calendar time. Additionally, HIV incidence estimates for each year were standardized to the 1995 population demographic mix by using the modeled HIV incidence rates from the Poisson regression and obtaining directly standardized incidence rates in each calendar year (standardizing for age, marital status, education, religion, and years of follow-up).

To determine if changes in HIV incidence over time were associated with changes in specific demographics or risk behavior, Poisson regression analyses that included sociodemographic, behavioral, and clinical explanatory variables were also performed. Through systematic univariate and multivariate modeling (data not shown), we selected age, marital status, education, religion, condom use with FSWs, number of multiple sex partners, prevalence of GUD, and prevalence of HSV-2 infection as risk factors to adjust and explain the trends in HIV incidence rates. We fitted several models to identify the risk behaviors that could explain the changing trends in acquisition of HIV infection. Model 1 included such variables as calendar year of HIV diagnosis, completed years of follow-up, age, marital status, education, and religion. Model 2 added condom use with FSWs to model 1. Model 3 added GUD to model 1. Model 4 added HSV-2 infection to model 1. Model 5 added condom use, GUD, and HSV-2 to model 1. Model 6 added number of sex partners to model 5. Model 7 included multiple sex partners but not condom use, GUD, and HSV-2 infection. Data were sufficient in each calendar year to conduct the risk factor analysis in the Poisson model for male patients with STIs but not for FSWs or other women.


Between 1993 and 2002, 14,147 patients attending the STI clinics were screened for HIV, of whom 3185 (22.5%) were HIV infected. A total of 3268 HIV-uninfected individuals agreed to participate in the prospective cohort study, and the incidence estimations were done in the 3 study cohorts with different risk behaviors. In all, 274 seroconversions were identified over a period of 10 years. Of these 86.9%, 0.36%, and 7.30% were positive for HIV-1, HIV-2, and HIV-1 and HIV-2, respectively, with incidence rates of 3.94 per 100 PYs (95% confidence interval [CI]: 3.45 to 4.47), 0.02 per 100 PYs (95% CI: 0.00 to 0.12), and 0.39 per 100 PYs (95% CI: 0.24 to 0.60). With available testing kits, 15 (5.47%) could not be classified as positive for HIV-1 or for HIV-2.

In all, 197 of 2641 men contributing 4907 PYs of follow-up over the study period became HIV infected, resulting in a cumulative HIV incidence estimate of 4.01 per 100 PYs (95% CI: 3.47 to 4.62). Because only 0.9% (24 of 2665) men reported bisexual behavior, they have been excluded from this analysis. HIV infection was identified in 174 of 2420 heterosexual men with 4460 PYs of follow-up, resulting in a cumulative HIV incidence estimate of 3.9 per 100 PYs (95% CI: 3.34 to 4.53). The HIV incidence estimate of 5.15 per 100 PYs (95% CI: 3.26 to 7.72) among MSM was a result of 23 seroconversions among 221 such men with 447 PYs of follow-up. We estimated an HIV incidence rate of 3.66 per 100 PYs (95% CI: 2.76 to 4.75) among young male patients with STIs (between 20 and 25 years of age), which is not significantly different from the incidence rate of 4.74 per 100 PYs (95% CI: 3.41 to 6.44) among male participants in the age group >35 years. Of 232 FSWs contributing 612 PYs of follow-up, 57 became HIV infected, resulting in a cumulative HIV incidence estimate of 9.31 per 100 PYs (95% CI: 7.05 to 12.07). In FSWs who were younger than 20 years of age, the incidence rate of 23.88 per 100 PYs (95% CI: 6.5 to 61.1) was significantly higher (P = 0.03) than that observed in the age group older than 35 years (5.85 per 100 PYs [95% CI: 3.11 to 9.10]). Among 395 non-sex worker women contributing 609 PYs of follow-up, 20 HIV seroconverters were identified, yielding a cumulative HIV incidence estimate of 3.28 per 100 PYs (95% CI: 2.01 to 5.07). The HIV incidence rate was 5.59 per 100 PYs (95% CI: 1.52 to 14.32) among young non-sex worker women (age <20 years), which was not significantly different from the incidence rate of 2.01 per 100 PYs (95% CI: 0.42 to 5.89) observed in the age group >35 years. An analysis of trends in HIV incidence over the 10-year study period among each gender risk group is presented in Table 1.

HIV-1 Seroincidence by Calendar Year Among STI Clinic Patients, Pune, India

A strong cohort effect was demonstrated, and a longer duration of follow-up was associated with a decrease in risk of HIV seroconversion for individuals enrolled in each calendar year of the study (data not shown). Accounting for the potential impact of changes in demographics over time and cohort effect, RRs and HIV incidence were calculated standardized to the 1995 population demographic mix. The RR of HIV infection decreased significantly, by approximately 80% (P < 0.001) among male patients with STIs and by approximately 70% (P = 0.02) among FSWs, over the 10-year study period (Fig. 1). In contrast, there was no significant change in the RR of HIV acquisition among non-sex worker women (P = 0.7).

The RR of HIV acquisition over time was estimated for each risk group. The analysis was standardized to the 1995 population mix, and RRs were calculated using a Poisson regression analysis that included age, marital status, education, religion, and completed years of follow-up.

A significant decline in HIV incidence among male patients with STIs and FSWs but not among women denying sex work was observed (Fig. 2). Similar trends were demonstrated when the analyses were standardized to the population demographic mix of other years of the study period. Sufficient prospective data were available to analyze the impact of changes over time in risk factors among male patients with STIs. Because heterosexual men and MSM did not differ in their sociodemographics and risk of acquiring HIV over a period of 10 years, we decided to study the impact of risk factors in a single group of male patients with STIs. Figure 3 shows that the percentage of men presenting to the STI clinic who reported any history of condom use with FSW contact increased over the 10-year period (P < 0.001). There was also a decrease over time in the percentage of men presenting with a clinical diagnosis of GUD (P < 0.001) and HSV-2 infection (P < 0.001). The percentage of men presenting to the STI clinic who reported a history of multiple sex partners remained unchanged over 10 years (P = 0.94), however. Similarly, Venereal Disease Research Laboratory (VDRL) test reactivity also remained unchanged over the study period.

HIV incidence rates were estimated for each risk group by calendar year, and a separate estimate of HIV incidence was calculated based on number of years of follow-up of individuals for each calendar year. HIV incidence estimates shown were standardized to the demographic mix of 1995 by age, marital status, education, religion, and completed years of follow-up.
Percentages of men attending STI clinics in Pune, India who reported a history of multiple sex partners, condom use with FSW sexual contact, history of HSV-2, and GUD are shown. These data are shown for new study participants enrolled in the prospective cohort study in each calendar year.

We therefore investigated whether such factors as demographic background, years of follow-up, reported condom use, GUD, presence of HSV-2 infection, or history of multiple sex partners could account for the significant decline over time in risk of HIV acquisition among men presenting to the STI clinics in Pune. Poisson regression model 1, shown in Table 2, demonstrated that the decline in the RR of HIV infection remained statistically significant over 10 calendar years (RR = 0.90; P = 0.002) when completed years of follow-up, age, marital status, education, and religion were included in the regression model. In contrast, models 2 and 3 demonstrated that the RR of HIV infection showed no significant decrease over calendar time with inclusion of condom use with FSW contact (RR = 1.00; P = 0.93) or GUD (RR = 0.97; P = 0.51). Model 4 demonstrated a significant decrease in the RR of HIV infection over calendar time when HSV-2 infection was included in the model (RR = 0.90; P = 0.002). Model 5, which included reported condom use with FSWs, GUD, and presence of HSV-2 infection, demonstrated an insignificant change in the RR of HIV acquisition over calendar years (RR = 1.07; P = 0.19). Model 6, constructed by adding multiple sex partners to model 5, demonstrated no significant change in the RR of HIV infection (RR = 1.10; P = 0.09). Finally, model 7, which included the number of multiple sex partners but not reported condom use, GUD, or presence of HSV-2 infection, demonstrated a significant decrease in the RR of HIV acquisition over calendar years (RR = 0.90; P = 0.004). Taken together, the Poisson regression models suggest that the decrease in HIV risk over time among men was associated with an increase in their reported condom use with FSWs and a decrease in the prevalence of GUD.

Poisson Regression Analysis of HIV Seroconversion Among Male Patients Attending STI Clinics in Pune, India: May 1993 Through August 2002


Our data provide the first direct evidence of a decline in risk for HIV acquisition over time in India, through systematically conducted cohort studies among FSW and male patients with STIs in Pune, and support the conclusion of a recent sentinel surveillance study from southern India reporting declining HIV incidence in southern India based on seroprevalence rates.5 Increased condom use by men with FSW contact and a decrease in prevalence of GUD seem to be the key factors associated with the decreasing HIV transmission risk among men in our study. The decline in risk for HIV infection among FSW may also be attributable to the significant increase in reported male condom use. Although men with high-risk behavior were more likely to use condoms with FSW sexual contact, we found that women not engaged in sex work reported a low rate of condom use with their husbands, and this did not change over the 10-year study period. Studies suggest that the prevalence of HIV infection in the same population has remained unchanged.11 Relatively unchanged prevalence and decreasing incidence over 10 years could be suggestive of increased survival;12 however, the reasons for the same and the possible association with access to antiretroviral therapy (ART) need to be investigated. The lack of change in HIV infection risk for female spouses of male patients with STIs, who reported no other risk factors than sexual exposure to their husbands, suggests that Indian women married to husbands with high-risk behavior require additional targeted HIV prevention interventions. The observed incidence rates reported here for different gender risk groups were lower than the ones reported earlier by the same investigators.9 The primary reason for this difference is that earlier incidence rates were based on a short follow-up and this report is based on a larger cohort and longer follow-up for nearly 10 years.

Although a reduction in HIV infection rates has been documented in some high-risk communities,13,14 only a few studies from countries facing the largest burden of HIV infection have demonstrated success in reducing HIV incidence in high-risk groups.15-17 The lack of direct HIV incidence estimates has also led some to question the conclusions of the recent analyses of seroprevalence surveys suggesting declining HIV rates among young men and women in southern India.5-8 Limited data demonstrating success in reducing HIV transmission in India may have contributed to pessimistic assumptions by some about the future impact of the HIV epidemic.3,18 Use of HIV prevalence data may also result in overestimation of the projected number of new infections.19 More accurate estimates of changing HIV incidence rates over time are likely to facilitate more accurate estimation of sample sizes required for future intervention studies, including vaccine and microbicide efficacy trials, for prevention of HIV infection in India. Our results from Pune are important because they indicate that significant declines in HIV infection rates in high-risk individuals in India can be achieved. Use of HIV incidence estimates coupled with HIV prevalence data generated through surveillance might also improve estimates of HIV disease burden and projections of the trajectory of the HIV epidemic in India.

Although the demonstration of declining risk for HIV infection among high-risk communities in Pune supports the conclusions that HIV incidence may be declining in southern India, it is important to recognize that generalization of these findings should be done with caution. Transmission dynamics among high-risk urban communities in Pune may not be applicable to rural communities, the general population, or other cities in India. The HIV epidemic in India is geographically diverse. Declining HIV incidence in men is expected to be followed by a decrease in HIV incidence among married monogamous women, with a gap of some years.

The finding of declining HIV rates among high-risk men and FSWs is good news for Pune, however, and may reflect the impact of years of concerted HIV prevention efforts and interventions like the extensive counseling employed at our site for risk reduction in behavior, condom promotion, and aggressive management of sexually transmitted diseases. Additionally, efforts of the Maharashtra State AIDS Control Society guided by the National AIDS Control Organization and those of a large number of other dedicated community-based voluntary organizations, medical colleges, and research institutes might have resulted in behavioral change leading to in the decline in HIV incidence in the study population. Because the epidemic in India is younger than that in Africa, we may not have reached a stage when lower incidence could be attributed to exhaustion of the susceptible population. Therefore, it is important to sustain these efforts, and it is desirable to strengthen and increase HIV prevention efforts that are targeted to protect the spouses of high-risk men, who remain at high risk.20 It is important that men are also involved extensively in research on female controlled options for HIV prevention to ensure better acceptance eventually. More focused and concentrated efforts in the high-risk population of FSWs and the bridging population of men visiting them are likely to arrest the spread of the HIV epidemic and restrict it from becoming widespread in the low-risk general population in India.


Informed consent was obtained from all study participants. This project was reviewed and approved by the Institutional Review Boards of the Johns Hopkins Medical School and the National AIDS Research Institute.

The authors thank the staff of the National AIDS Research Institute and Johns Hopkins University who provided help in the conduct of this study. They especially thank Mary Shepherd for setting up the data management in the initial stages and B.J. Medical College, Pune, and Pune Municipal Corporation for facilitating clinical collaborations. They thank the staff of Preparation for AIDS Vaccine Evaluation (PAVE), HIV Network for Prevention Trials (HIVNET), Acute Pathogenesis Study, and HIV Prevention Trials Network (HPTN) for counseling, clinical care of study participants, community work, laboratory work, data management, and statistical support.


1. Joint United Nations Program on HIV/AIDS (UNAIDS) and World Health Organization (WHO). AIDS Epidemic Update-December 2006. Global Summary of the AIDS Epidemic. Geneva, Switzerland. 2006:1-2.
2. National AIDS Control Organization, Ministry of Health and Family Welfare, Government of India. HIV/AIDS Epidemiological Surveillance and Estimation Report for the Year 2005. India: National AIDS Control Organization; New Delhi, India, 2006.
3. National Intelligence Council. The next wave of HIV/AIDS: Nigeria, Ethiopia, Russia, India, and China. Intelligence community assessment. September 2002. Available at: Accessed October 22, 2006.
4. Rao JV, Ganguly NK, Mehendale SM, et al. India's response to the HIV epidemic. Lancet. 2004;364:1296-1297.
5. Kumar R, Jha P, Arora P, et al, for the International Studies of HIV/AIDS (ISHA) Investigators. Trends in HIV-1 in young adults in south India from 2000 to 2004: a prevalence study. Lancet. 2006;367:1164-1172.
6. John T. HIV-1 prevalence in young adults in south India. Lancet. 2006;368:113-114.
7. Soman C, Mathew A. HIV-1 prevalence in young adults in south India. Lancet. 2006;368:113.
8. Gupte M, Mehendale S, Panda C, et al. HIV-1 prevalence in young adults in south India. Lancet. 2006;368:114.
9. Mehendale SM, Rodrigues JJ, Brookmeyer RS, et al. Incidence and predicators of human immunodeficiency virus type 1 seroconversion in patients attending sexually transmitted disease clinics in India. J Infect Dis. 1995;172:1486-1491.
10. Centers for Disease Control. Interpretation and use of the Western blot assay for serodiagnosis of human immunodeficiency virus type 1 infection. Morb Mort Wkly Rep. 1989;38:1-7.
11. Brahme R, Metha S, Sahay S, et al. Correlates and trends of HIV prevalence among female sex workers attending sexually transmitted disease clinics in Pune, India (1993-2002). J Acquir Immune Defic Syndr. 2006;41:107-113.
12. Brookmeyer RS, Mehendale SM, Peltz R, et al. Estimating the rate of occurrence of new HIV infections using serial sero-prevalence surveys: the epidemic in India. AIDS. 1996;10:924-925.
13. Valdiserri RO, Ogden LL, McCray E. Accomplishments in HIV prevention science: implications for stemming the epidemic. Nat Med. 2003;9:881-886.
14. Joint United Nations Program on HIV/AIDS (UNAIDS). HIV Prevention Needs and Successes: a Tale of Three Countries. An Update on the Prevention Success in Senegal, Thailand and Uganda. Geneva, Switzerland: UNAIDS; 2001. Available at: Accessed October 11, 2006.
15. Celentano DD, Nelson KE, Lyles CM, et al. Decreasing incidence of HIV and sexually transmitted diseases in young Thai men: evidence for success of the HIV/AIDS control and prevention program. AIDS. 1998;12:F29-F36.
16. Stoneburner RL, Low-Beer D. Population-level HIV declines and behavioral risk avoidance in Uganda. Science. 2004;304:714-718.
17. McClelland RS, Hassan WM, Lavreys L, et al. HIV-1 acquisition and disease progression are associated with decreased high-risk sexual behaviour among Kenyan female sex workers. AIDS. 2006;20:1969-1973.
18. Nagelkerke NJ, Jha P, de Vlas SJ, et al. Modelling HIV/AIDS epidemics in Botswana and India: impact of interventions to prevent transmission. Bull World Health Organ. 2002;80:89-96.
19. Ghys PD, Kufa E, George MV. Measuring trends in prevalence and incidence of HIV infection in countries with generalised epidemics. Sex Transm Infect. 2006;82(Suppl 1):i52-i56.
20. Gangakhedkar RR, Bentley ME, Divekar AD, et al. Spread of HIV infection in married monogamous women in India. JAMA. 1997;278:2090-2092.

high-risk groups; HIV; incidence; risk factors; patients with sexually transmitted infections; trends

© 2007 Lippincott Williams & Wilkins, Inc.