Modelling and estimation of HIV prevalence and number of people living with HIV in India,2010

Abstract

This paper provides HIV estimation methodology used in India and key HIV estimates for 2010–2011. We used a modified version of the Spectrum tool that included an Estimation and Projection Package as part of its AIDS Impact Module. Inputs related to population size, age-specific pattern of fertility, gender-ratio at birth, age and gender-specific pattern of mortality, and volume and age–gender distribution of net migration were derived from census records, the Sample Registration System and large-scale demographic health surveys. Epidemiological and programmatic data were derived from HIV sentinel surveillance, large-scale epidemiological surveys and the programme management information system. Estimated adult HIV prevalence retained a declining trend in India, following its peak in 2002 at a level of 0.41% (within bounds 0.35–0.47%). By 2010 and 2011, it levelled at estimates of 0.28% (0.24–0.34%) and 0.27% (0.22–0.33%), respectively. The estimated number of people living with HIV (PLHIV) reduced by 8% between 2007 and 2011. While children accounted for approximately 6.3% of total HIV infections in 2007, this proportion increased to about 7% in 2011. With changing priorities and epidemic patterns, the programme has to customise its strategies to effectively address the emerging vulnerabilities and adapt them to suit the requirements of different geographical regions.

Keywords

Epidemiology human immunodeficiency virus (HIV)AIDS prevention Asia prevalence Spectrum tool India

Introduction

One of India's National AIDS Control Programme strategies is to strengthen the evidence base for informed decision making. The estimation and projection of HIV is one of the critical pieces of information required to formulate strategies and interventions to control the epidemic in the country.^1,2 India has been producing annual HIV estimates since 1998, following globally recommended and validated methods at every round of estimation.^3–6 Tools have also been updated and customised to the country's context for improved accuracy of the estimates.

One of the key sources of information for HIV estimation and projection in India comes from the HIV sentinel surveillance.⁷ India has expanded the HIV sentinel surveillance network from 176 sentinel sites in 1998 to 1359 sites in 2011 to ensure coverage of all districts of the country, which are the lowest administrative unit for implementation of the HIV prevention programme. This expansion was done to comprehensively understand the epidemic's pattern amongst high-risk groups (female sex workers [FSWs], men having sex with men [MSM], transgender [TG] people, people who inject drugs, single-male migrants and long-distance truckers) and pregnant women attending antenatal clinics (ANC). In 2010–2011, the HIV sentinel surveillance was conducted at 696 sites for women attending antenatal clinics, 194 sites for sexually transmitted infection (STI) patients, and 479 sites for high-risk groups, including 39 sites for migrants and long-distance truckers. A total of 427,559 samples were collected and tested during HIV Sentinel Surveillance 2010–2011.^7,8

Apart from geographical expansion, the HIV sentinel surveillance was also strengthened by improving sample selection at sentinel sites, data management and laboratory support. Key strategies for improving the quality and comprehensiveness of data included: changing sampling methodology from consecutive sampling to random sampling for high-risk groups, putting special focus on technical validation of new sites, mechanisms for ensuring high quality of data collection and data management using an e-platform (SIMS), and specimen collection and processing, which included a four-tier supervisory structure.⁸

This paper provides key results along with details of the methodology used to generate HIV estimates for India for 2010–2011. The HIV estimation exercise aimed not only to provide estimates of HIV prevalence, but also HIV incidence, AIDS-related mortality, and programme needs for the years 2010 and 2011, as well as back calculate comparable estimates for previous years.

Methods

The model

The estimation process included a deterministic modelling technique to arrive at robust estimates of HIV prevalence in the country. A modified version of the Spectrum tool (version: 4.53 Beta19) was used that included an Estimation and Projection Package (EPP)^9,10 as part of its AIDS Impact Module (AIM).^10–12 The deterministic model used by Spectrum divides the 15–49 age group into three groups: (1) a not-at-risk group; (2) an at-risk group; and (3) an infected group. Three differential equations describe the changes in these groups and hence changes in HIV prevalence over time.¹¹ In these equations, four parameters determine the shape of the epidemic curve: start year of the HIV epidemic, force of infection, initial fraction of the adult population at risk of infection, and adjustment in size of the high-risk groups in response to behavioural changes.^9,10 The AIM module uses the Bayesian melding approach¹² which refers to combining information about the inputs and outputs of a deterministic model to estimate the incidence and their 95% confidence intervals.

The model allows the entire population to be divided into several subgroups according to their relevance to the socio-cultural setting and the nature of the local epidemic. Considering the history of the HIV epidemic in India, it was defined as being a concentrated epidemic amongst three high-risk groups, namely, FSWs, MSM and TG, as well as injecting drug users (IDUs). The epidemic was specified as being a non-IDU concentrated epidemic in all states/Union Territory – excluding Manipur and Nagaland – based on existing data.² Thus, we divided the total population into three high-risk groups (FSWs, MSM and IDUs) and the remaining population. For each state/Union Territory, population sizes of high-risk groups (obtained from mapping and size estimation) were used. The trajectory of year-wise growth in high-risk groups' population size was aligned with the growth of the general population.

The model also allowed movement of individuals from high-risk groups to the general population by calculating the reassignments based on the average duration of being a FSW and/or IDU based on available evidence (FSW: eight years and IDU: 15 years).^13–15 Further, for IDUs, the non-AIDS mortality was assumed to be 7% higher compared to non-IDU populations. This assumption was made to account for the higher risk of mortality experienced by all IDUs regardless of HIV status.¹⁶ This allowed adjustments in HIV prevalence in the general population due to the movement of FSWs and IDUs. However, additions to the number of HIV-positive people in the general population were not considered to be new infections and hence this did not alter the estimated incidence of HIV in the general population.

A wide range of datasets were used to fit the model. This included data from the HIV sentinel surveillance, census data for India, Sample Registration System, National Family Health Survey,¹⁷ Behavioural Sentinel Surveillance Surveys,^18–20 Integrated Behavioural and Biological Assessments^13,14 and programme data related to antiretroviral therapy (ART) and Prevention of Mother-to-Child Transmission (PMTCT).

Using demographic, epidemiological and programme data, along with assumptions (based on available data) for transmission probabilities and epidemic patterns, the model provided the required HIV estimates. The inbuilt EPP in the AIM module generated the estimated trend of HIV prevalence amongst adults for each population group. Based on demographic projections, estimated trend of adult HIV prevalence and epidemiological assumptions, the number of people living with HIV in different age groups were determined. The entire process was repeated separately for each state/Union Territory of the country. The process is presented in Figure 1.

Figure 1.

Schematic diagram showing the structure and process for generating HIV estimate. ART, antiretroviral therapy; HRG, high risk groups; PLHIV, people living with HIV; PMTCT, prevention of mother-to-child transmission.

Inputs

Demographic projection

Population size for each year of the projection period (1981–2017) was estimated using the DemProj module²¹ of Spectrum. The age–gender distribution of the population from the 1981 Census was used, along with age-specific patterns of fertility, gender-ratio at birth, age and gender-specific pattern of mortality, and volume and age–gender distribution of net migration as inputs for the demographic projection. These values were obtained from censuses, Sample Registration System bulletins, and other large-scale demographic health surveys in the country. The projected age–gender distribution of the population was matched with the age-gender distribution derived from the four census years (1981, 1991, 2001, and 2011) included in the projection period to ensure consistency. The process was repeated separately for each state/Union Territory.

Epidemiological inputs from sentinel surveillance

Surveillance data inputs from HIV sentinel surveillance, coupled with Integrated Behavioural and Biological assessments^22,23 were used to derive state-specific prevalence. To ensure appropriate reflection of epidemic levels and to avoid bias due to small sample size, we only included those sentinel sites for estimation where at least 75% of the assigned sample was achieved. Specifically, ANC sites were included only if 300 or more women tested positive out of a total assigned sample size of 400. Similarly, for high-risk groups and bridge population sites the cut-off was 188 out of the total assigned sample size of 250.⁸ To account for the change in sampling methodology from consecutive sampling to random sampling in 2010–2011 for high-risk groups⁸ the data points for 2010–2011 were regarded as ‘new sites’. Sentinel surveillance data were cleaned for outliers by dropping data points if the prevalence of the site for a particular year was assessed to be too high or too low compared with other years. HIV prevalence amongst high-risk groups from two rounds of behavioural and biological assessments were included as independent data points for sites.

Programme statistics on coverage for vertical HIV transmission and ART coverage

Programme data on the number of mothers receiving single-dose nevirapine and number of adults and children receiving ART from 2003 to 2011 were available, while values of these indicators for the period of 2012–2017 were projected following the targets (calculated based on analysis of past performance, intended outcomes, scale-up strategies and plans, as well as available resources) of the National AIDS Control Programme. The trend in eligibility criteria for receiving ART was also incorporated in the estimation process (up to 2008: CD4 count < 200 cells/mm³; 2009–2011: CD4 count < 250 cells/mm³; 2012–2017: CD4 count < 350 cells/mm³).

Epidemiological assumptions

Age–gender pattern of HIV incidence

Assumptions were made on the trend in the ratio of female-to-male HIV incidence amongst those aged 15–49 years, and the trend in the distribution of HIV incidence by age for both men and women. Three sets of distribution patterns of HIV incidence by age for both men and women were provided in the AIM module.¹¹ These patterns were estimated for following three types of epidemics: generalised epidemic, concentrated non-IDU epidemic and concentrated IDU epidemic. For all states in India, a concentrated non-IDU pattern was used. For Manipur and Nagaland, the concentrated IDU epidemic pattern was used considering the nature of epidemic in these states.

Transmission parameters

Assumptions related to mother-to-child transmission were made to account for the fact that children who are infected in utero, peripartum and intrapartum die sooner than children who are infected after birth through breastfeeding. The probability of HIV infection at birth for a child born to a HIV-positive mother was assumed to be 20% in the absence of prophylaxis, and 11% with single-dose nevirapine. We also adjusted the effect of breastfeeding on mother-to-child transmission by specifying transmission probabilities related to the duration of breastfeeding. The probability of infection through breastfeeding was assumed to be 1.5% per month for mixed feeding during the first six months, 0.75% per month for exclusive feeding during the first six months, 0.75% per month for months seven and later, and 0.3% per month when the mother is on triple therapy.²⁴ These rates were applied to the expected number of births to HIV-positive women to calculate HIV-positive children. The number of HIV-infected children was linked with the inputs on the children's survival to estimate the number of AIDS-related deaths amongst children.

Fertility of HIV-positive women

The ratio of age-specific fertility rates between HIV-positive and -negative women was derived from the analysis of 20 demographic and health surveys conducted during 2003–2007. Based on the evidence, it was assumed that HIV-positive women in the age group of 15–19 have a 20% (ratio: 1.2) higher fertility rate than HIV-negative women in this age group. However, HIV-positive women were assumed to have lower fertility rates in older age groups as evident from the assumed ratios for the different age groups (20–24: 0.76; 25–29: 0.71; 30–34: 0.65; 35–39: 0.59; 40–44: 0.53; 45–49: 0.47).

Curve fitting, and estimation of adult prevalence and incidence

With a plausible range of the four parameters (prior distributions), a set of possible epidemic curves were generated. The Sampling Importance Resample (SIR) algorithm^25,26 was used to randomly select a large number of a combination of the four parameters. For each combination, a curve was generated and compared with the observed HIV prevalence. If the generated curve was very different from the observed HIV prevalence, it was assigned a low or zero weight. If the curve resembled the observed HIV prevalence, it was assigned a higher weight. The epidemic curves and input parameters were resampled such that probability of being selected was proportional to the assigned weights. The 95% confidence interval was given by the lower 2.5th and upper 97.5th percentile of the prevalence for that year within the sample.

Calibrating the ANC prevalence curves

State-specific data from the population-based household surveys were used to calibrate the trend of ANC HIV prevalence because of the differential levels between ANC prevalence through sentinel surveillance and survey estimates. For five high HIV prevalence states (Andhra Pradesh, Karnataka, Maharashtra, Manipur and Tamil Nadu) the population estimates were derived using the 2005–2006 National Family Health Survey (NFHS-3)¹⁷ data while for Nagaland it was determined from a specific study undertaken by National AIDS Control Organization (NACO).²⁷ While the survey estimates were used directly in the model for states where survey estimates were available, for rest of the states, calibration factors were derived as ratio of national level prevalence estimates from surveillance and household surveys.

Estimation of HIV incidence for all ages

Based on the information about estimated trends in HIV prevalence and number of people receiving ART, Spectrum determined trends in HIV incidence amongst the adult population that are needed to keep the prevalence at the estimated levels. All cases of children living with HIV were calculated as resulting from mother-to-child transmission. From the ratio of male-to-female incidence of infections amongst adults, Spectrum calculated the proportion of women living with HIV at all ages. A specific pattern of fertility rate amongst HIV-positive women was applied to estimate the number of HIV-positive children. The probability of transmission from mother to child with or without treatment was then applied, considering also the duration of breastfeeding and the probability of transmission linked to it. Once the final incidence trend was determined, with a breakdown pattern for all age groups, Spectrum recalculated the HIV incidence from the beginning of the epidemic. The recalculated HIV incidence was applied over the total population size of the respective groups to derive the estimated HIV prevalence. Thus, all other parameters were recalculated in Spectrum using the estimated HIV incidence since the start of the epidemic.

Ethics statement

The study was part of the national HIV estimation and projection of India. The process, methodology and results of HIV estimation and projection were reviewed by the Technical Resource Group on Surveillance and Estimations, constituted by the Government of India. The ethical issues were also reviewed and approved by the ethics committee of National Institute of Medical Statistics. The results were accepted and published by Government of India as the official HIV estimates for the country.⁸

Results

Adult HIV prevalence

Adult HIV prevalence was calculated as a percentage of the total adult population living with HIV. Figure 2 shows that estimated adult HIV prevalence retained a declining trend in India, following its peak in 2002 at a level of 0.41% (within bounds 0.35–0.47%). By 2010 and 2011, it levelled at estimates of 0.28% (0.24–0.34%) and 0.27% (0.22–0.33%), respectively.

Figure 2.

Estimated adult HIV prevalence in India, 2000–2011 with uncertainty bounds.

HIV prevalence amongst children, young men and young women

Age and gender-wise, the HIV prevalence was respectively calculated by aggregating the number of people within the specified age group living with HIV divided by the total population for these age groups. Estimated HIV prevalence amongst children remained almost constant during 2007–2011 at 0.04% (0.03–0.05%) (Figure 3). HIV prevalence amongst the young male population declined gradually from an estimated 0.15% (0.11–0.21%) in 2007 to 0.11% (0.07–0.17%) in 2010 and 2011 (Figure 4). A similar level and trend was estimated for the young female population: 0.15% (0.11–0.19%) and 0.11% (0.07–0.16%) in 2007 and 2011, respectively (Figure 5).

Figure 3.

Estimated HIV prevalence amongst children (<15 years) in India 2007–2011, with uncertainty bounds.

Figure 4.

Estimated HIV prevalence amongst young male population (15–24 years) in India, 2007–2011, with uncertainty bounds.

Figure 5.

Estimated HIV prevalence amongst the young female population (15–24 years) in India, 2007–2011, with uncertainty bounds.

Number of people living with HIV by age and sex

The estimated number of people (adults and children) living with HIV (PLHIV) in 2011 was 2.09 million (1.72 million–2.53 million), compared to the estimated 2.25 million (1.92 million–2.53 million) PLHIV in the country in 2007 (Figure 6). A comparison between 2007 and 2011 estimates reflects an approximate 8% decline in the total number of PLHIV. The decline in number of PLHIV annually has been at a steady pace; approximately 3% from 2007 to 2008, 2.5% from 2008 to 2009, 1.5% from 2009 to 2010 and nearly 1% from 2010 to 2011.

Figure 6.

Estimated number of people living with HIV (all ages) in India, 2007–2011, with uncertainty bounds.

Estimates of the number of children living with HIV increased from 2007 to 2009 before declining from 2009 to 2011 (Figure 7). The number of children living with HIV was estimated at about 145,000 (116,000–183,000) in 2011. This was at a slightly lower level than the estimated 142,000 (111,000–186,000) children with HIV in 2007. The proportional contribution of the number of children living with HIV out of the total PLHIV population is estimated to have marginally increased from 6.3% in 2007 to 7.0% in 2011.

Figure 7.

Estimated number of children (<15 years) living with HIV in India, 2007–2011, with uncertainty bounds.

Figure 8.

Estimated number of adults (15 + years) living with HIV in India, 2007–2011, with uncertainty bounds.

The number of adults living with HIV retained a declining trend, similar to the trend in prevalence. The estimate for this indicator in 2011 was 1.94 million (1.59 million–2.37 million) which is fewer than the 2.10 million (1.79 million–2.36 million) adults estimated to be living with HIV in 2007 (Figure 8). The proportional contribution of adults to the total PLHIV population is slightly declining. This group accounted for approximately 94% of total infections in 2007 and 93% of total infections in 2011.

Out of the total adult PLHIV population, women are estimated to have accounted for approximately 39% of infections, whilst men accounted for approximately 61% of infections in 2011.

Discussion

The method of HIV estimation in India followed global recommendations and adapted the tools to incorporate characteristics of the local HIV epidemic. The current methodology addresses the social, behavioural and epidemiological aspects of the HIV epidemic in India, and hence it is more likely to have yielded accurate results compared to methods used in the past. The complexities and multidimensional nature of the epidemic requires such a comprehensive approach to get robust estimates. For instance, reassignment of people from high-risk groups to the general population is a reality and it has to be addressed in the methodology to arrive at the close estimates. Similarly, the rigorous demographic projections are essential to calculate the meaningful indicators described in this paper. Use of the Bayesian technique provides precise point estimates and their uncertainty bounds. The increased availability of data from cross-sectional surveys and research made it possible for India to use this advanced methodology.

The estimated trend of HIV prevalence in India reflects a declining trend in the epidemic, which is consistent with other studies stating a decline in HIV prevalence and increase in practices focusing on reducing its spread in general as well as in high-risk groups in different parts of the country.^15,28–30

Information collected from the HIV-positive cases detected at integrated counselling and testing centres across the country shows that unprotected hetero-sexual contact is the main route of HIV transmission, accounting for around 88% of all cases. Transmission through contaminated blood and blood products has decreased significantly to less than 1%. While mother-to-child transmission accounts for around 5% of HIV infections, infections through injecting drug use and homosexual sex are increasing. Considering that the HIV epidemic in India remains concentrated in nature and heterogeneous in its distribution, it is important to have more sub-national (state and district) level analysis to consistently adapt and respond to the changing epidemic patterns.

One of the key strengths of India's AIDS Control Programme is its evidence-based approach to addressing the vulnerabilities. Results of the HIV estimation process are analysed in greater detail to interpret the diverse patterns in different epidemiological indicators in different states. These results are interpreted in light of evidence from other data sources to formulate appropriate recommendations for the programme. For more effective use of estimation and projection exercises, it is recommended that similar exercises should be undertaken for specific regions or groups of districts within states to better characterise the epidemics.

There are a few limitations to the methodology used by India for providing HIV estimates. These are mainly related to the lack of data related to age-gender distribution of HIV incidence, transmission probabilities, and fertility of HIV-positive women. Because of this, we had to assume values of these parameters based on evidence from different parts of the globe. More research is needed to validate these assumptions for the country and also their possible effect on HIV estimates.

In conclusion, India has successfully managed to control its HIV epidemic. The programme has increasingly protected people from getting infected with HIV, saved many people from dying due to AIDS-related causes, improved the quality of life of those who are infected, empowered marginalised communities by enhancing their access to health and social services, and created an enabling environment where persons and families affected with HIV are treated with dignity and respect. With changing priorities and epidemic patterns, the programme has to customise its strategies to effectively address the emerging vulnerabilities and adapt them to suit the requirements of different geographical regions.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

National AIDS Control Organization (NACO). National AIDS control programme, Phase II. Ministry of Health & Family Welfare, Government of India, New Delhi, 1999.

National AIDS Control Organization (NACO). NACP-III – to halt and reverse the HIV epidemic in India. Ministry of Health & Family Welfare, Government of India, New Delhi, 2006.

National Institute of Medical Statistics, National AIDS Control Organization. Technical Report: India HIV Estimates. New Delhi: Ministry of Health and Family Welfare, 2010.

Pandey

Reddy

Ghys

. Improved estimates of India's HIV burden in 2006. Indian J Med Res 2009; 129: 50–58.

Pandey

Sahu

Bakkali

. Estimate of HIV prevalence and number of people living with HIV in India 2008–2009. BMJ Open 2012; 2: 1–8.

Pandey

Thomas

Reddy

. Process of estimating the number of people living with HIV in India. Indian J Public Health 2007; 51: 7–13.

National Institute of Medical Statistics, National AIDS Control Organization. Technical Report: India HIV Estimates. New Delhi: Ministry of Health and Family Welfare, 2012.

National AIDS Control Organization. HIV Sentinel Surveillance 2010–11: a technical brief. New Delhi: Ministry of Health and Family Welfare, 2012.

Brown

Grassly

Garnett

. Improving projections at the country level: the UNAIDS Estimation and Projection Package 2005. Sex Transm Infect 2006; 82: iii34–40.

10.

Brown T, Peerapatanapokin W and Nalyn D. Improved process models for estimation and projection of HIV in Asian countries. East-West Center Honolulu, Hawaii, 2009.

11.

Stover

Brown

Marston

. Updates to the Spectrum/Estimation and Projection Package (EPP) model to estimate HIV trends for adults and children. Sex Transm Infect 2012; 88: i11–i16.

12.

Bao

Salomon

Brown

. Modelling national HIV/AIDS epidemics: revised approach in the UNAIDS Estimation and Projection Package 2011. Sex Transm Infect 2012; 88: i3–i10.

13.

Indian Council of Medical Research, Family Health International. National Interim Summary Report-India, Integrated Behavioral and Biological Assessment (IBBA), Round 1. Indian Council of Medical Research, New Delhi, India. http://www.nari-icmr.res.in/ibba.html (2007, accessed 10 July 2014).

14.

Indian Council of Medical Research, Family Health International. National Interim Summary Report - India, Integrated Behavioral and Biological Assessment (IBBA), Round 2 (2009–2010). Indian Council of Medical Research, New Delhi, India. http://www.nari-icmr.res.in/ibba.html (2011, accessed 27 May 2014).

15.

Mishra

Dube

Sahu

. Changing epidemiology of HIV in Mumbai: an application of the Asian Epidemic Model. Global J Health Sci 2012; 4: 100–112.

16.

Mathers

Degenhardt

Bucello

. Mortality among people who inject drugs: a systematic review and meta-analysis. Bull World Health Organ 2013; 91: 102–123.

17.

International Institute for Population Sciences (IIPS), Macro International (2007) National Family Health Survey (NFHS-3), Mumbai, India, 2005–2006.

18.

National AIDS Control Organization. National Behavioral Surveillance Survey, 2006, Female sex workers and clients of female sex workers. Ministry of Health & Family Welfare, Government of India, New Delhi. http://naco.gov.in/NACO/Quick_Links/Publication/ME_and_Research_Surveillance/Reports_and_Surveys/National_BSS_20062/ (2006, accessed 15 March 2014).

19.

National AIDS Control Organization. National Behavioral Surveillance Survey, 2006, Men who have Sex with Men (MSM) and Injecting Drug Users (IDUs). Ministry of Health & Family Welfare, Government of India. http://naco.gov.in/NACO/Quick_Links/Publication/ME_and_Research_Surveillance/Reports_and_Surveys/National_BSS_20062/ (2006, accessed 19 August 2014).

20.

National AIDS Control Organization. National Behavioral Surveillance Survey: General Population. Ministry of Health & Family Welfare, Government of India. http://naco.gov.in/NACO/Quick_Links/Publication/ME_and_Research_Surveillance/Reports_and_Surveys/National_BSS_20062/ (2006, accessed 19 August 2014).

21.

Stover J and Kirmeyer S. DemProj-Version 4: a computer program for making population projections. Policy Project, USAID, 2005.

22.

Saidel

Adhikary

Mainkar

. Baseline integrated behavioural and biological assessment among most at-risk populations in six high-prevalence states of India: design and implementation challenges. AIDS 2008; 22: S17–S34.

23.

Paranjape

Ramakrishnan

George

. The integrated behavioural & biological assessments – tool for analysis & evaluation of trends in HIV epidemic. Indian J Med Res 2012; 136: 1–3.

24.

Stover

Johnson

Hallett

. The Spectrum projection package: improvements in estimating incidence by age and sex, mother-to-child transmission, HIV progression in children and double orphans. Sex Transm Infect 2010; 86: ii16–ii21.

25.

Rubin

. Comment on “The calculation of posterior distributions by data augmentation,” by M Tanner and W H Wang. Journal of American Statistical Association 1987; 82: 543–546.

26.

Rubin DB. Using the SIR algorithm to simulate posterior distributions. In: Bernardo JM, et al. (eds). Bayesian Statistics. Oxford University Press, 1988, pp.395–402.

27.

Bachnai

Sogarwal

Rao

. A population based survey on HIV prevalence in Nagaland, India. SAARC J Tuberculosis Lung Dis HIV/AIDS 2009; 6: 1–11.

28.

National AIDS Control Organization. Annual Report 2009-10. New Delhi: Ministry of Health and Family Welfare, 2010.

29.

Thomas

Mimiaga

Kumar

. HIV in Indian MSM: reasons for a concentrated epidemic & strategies for prevention. Indian J Med Res 2011; 134: 920–929.

30.

Gakidou

Levin-Rector

. Assessment of population-level effect of Avahan, an HIV-prevention initiative in India. Lancet 2011; 378: 1643–1652.

Modelling and estimation of HIV prevalence and number of people living with HIV in India,2010–2011

Abstract

Keywords

Introduction

Methods

The model

Inputs

Demographic projection

Epidemiological inputs from sentinel surveillance

Programme statistics on coverage for vertical HIV transmission and ART coverage

Epidemiological assumptions

Age–gender pattern of HIV incidence

Transmission parameters

Fertility of HIV-positive women

Curve fitting, and estimation of adult prevalence and incidence

Calibrating the ANC prevalence curves

Estimation of HIV incidence for all ages

Ethics statement

Results

Adult HIV prevalence

HIV prevalence amongst children, young men and young women

Number of people living with HIV by age and sex

Discussion

Footnotes

Declaration of Conflicting Interests

Funding

References