University of California, San Francisco Logo

University of California, San Francisco | About UCSF | Search UCSF | UCSF Medical Center

Home > Global Health Literature Digest > Economic analysis
Economic analysis of HIV prevention interventions in Andhra Pradesh state of India to inform resource allocation
Global Health Sciences Literature Digest
Published June 30, 2009
Journal Article

Dandona L, Kumar SG, Kumar G, Dandona R. Economic analysis of HIV prevention interventions in Andhra Pradesh state of India to inform resource allocation. AIDS 2009;23(2):233-42.


To conduct an economic analysis of publicly funded HIV prevention programs in Andhra Pradesh

Study Design

Economic analysis


Andhra Pradesh, India


Costs per HIV case averted


A representative sample was selected that consisted of 128 of the 551 publicly funded HIV prevention programs in the region. All of the 14 types of programs were included: voluntary counseling and testing centers; sexually transmitted infection (STI) clinics; prevention of mother-to-child transmission (PMTCT) clinics; blood banks; and programs aimed at female sex workers, men who have sex with men (MSM), truckers, migrant laborers, street children, and prisoners; and information education and communication (IEC), condom promotion, composite, and general public programs.

Program and fiscal data were abstracted from written records and interviews with program staff. Program costs were divided into five categories: personnel, recurrent goods (e.g., condoms, HIV test kits), recurrent services, capital goods, and office space rental (estimated based on local rates if the program was housed in public buildings). The costs included expenditures, resources used, and donated inputs. Costs in Indian rupees were converted into US dollars in the April 2005-March 2006 fiscal year.

The effect of the interventions was based upon the assumption that all HIV infections in India are sexually acquired. The probability of HIV infection was estimated using the Weinstein formula(1,2) and was calculated separately for the group that would receive the intervention and for their sex partners; the number of these new infections was added for the total number of new infections. For the interventions aimed at sex workers, the number of new HIV infections acquired by clients´┐Ż other female sex partners also was included. Similarly, for the MSM-targeted interventions, HIV infections acquired by their female partners were included. The probabilities for infection per sexual act and contact were derived from the literature(1,2,3) and were calculated as three time higher in the presence of an STI.1,3 For IEC interventions aimed at the general public it was assumed that there would be a 30% lower probability of HIV acquisition with each unprotected sex act, assuming less risky sex compared with higher risk groups. Condom efficacy was assumed to be 80% with vaginal sex, 70% with anal sex, and 90% with oral sex.(4,5) Population-based surveys, other surveys, and program data were used to derive the HIV prevalence, fraction of sex acts that were protected, average number of sex acts per partner, and the average number of sex acts.

The estimated effect of the interventions on increasing the use of condoms, reducing the number of partners, and increasing STI treatment was obtained from published reports(6,7) and other estimates when not available from published reports. The Weinstein formula was applied with various scenarios in which interventions would have had an impact on condom use, the number of partners, and per contact risk to estimate the number of HIV cases averted per 1,000 persons receiving the intervention.

For the effect of PMTCT programs, the estimate of transmission without intervention was 25% and 40% less with nevirapine.(8) The effect of screening blood was estimated at reducing infection from blood by 92%.(9)

The validity of the model was assessed using information on the magnitude of the HIV epidemic in Andhra Pradesh. Additional information from the literature, program data, and local trends were used to develop plausible values for each variable. Random values that fell within the plausible range were used in a sensitivity analysis. Using the intervention effect calculation, including the sensitivity analysis, the number of HIV infections averted from each intervention was calculated based upon all publicly funded prevention programs in the area for the fiscal year 2005-2006.

A unit cost for each targeted person or unit service was calculated based on the total target group served and the cost of the programs included in the analysis. The total cost of each intervention was calculated using the unit costs and the total publicly funded HIV prevention services provided in Andhra Pradesh for the fiscal year 2005-2006. The coverage of each intervention was compared to the estimated optimal coverage (i.e. the total needed coverage). The number of HIV infections that could be averted if there was optimal coverage was estimated by applying the intervention effect to that coverage with the assumption that the intervention had the maximum estimated effect. Sensitivity analyses using the 5th-95th percentile range for the effect also were conducted. The total cost of each intervention for the optimal coverage was calculated. The percentage of total investment for each intervention cost also was calculated. The total number of HIV infections that could be averted in Andhra Pradesh with optimal coverage, assuming the interventions was provided in the same distribution as they were for fiscal year 2005-2006, was found.


The prevention programs were estimated to have prevented 9,688 new HIV infections in fiscal year 2005-2006 in Andhra Pradesh. The greatest number of infections averted was among MSM and sex workers at 33.43 infections and 32.64 infections per 1,000 targeted population, respectively. The fewest was for IEC programs for the general public at 0.015 infections per 1000 targeted population. For the fiscal year 2005-2006, the 14 prevention programs evaluated are estimated to have averted 9,688 infections in Andhra Pradesh (range 6,624-16,898). The highest proportion of infections averted was from STI clinic programs (23.9%), followed by blood banks (18.1%), sex worker programs (13.7%), and counseling and testing clinics (13.5%). There were an estimated 55,396 new HIV infections acquired in this time period in Andhra Pradesh.

There was a large difference between the actual and optimal levels of coverage of the prevention programs. With optimal coverage, an estimated number of averted HIV infections would be 2.4 times the estimated number (22,916, range 7138-27968) but would require increases in the proportion of funds allocated to migrant laborer programs (20.5% of total costs), counseling and testing centers (18.4% of total costs), and MSM-focused programs (6.2% of total costs). Optimal coverage would cost US$38.8 million annually.


If funds for prevention programs were allocated to those with the greatest effect on averting HIV infections, the increased funding for HIV services in India should be adequate.

Quality Rating

Based upon the Quality of Health Economic Studies (QHES) instrument, this study is of high quality. Particular strengths included the demonstrated need for and value of the findings, use of a known and established model for estimating the risk of HIV infection, documentation of the source of information used as parameter estimates in the models, use of sensitivity analyses, appropriate outcomes, and conclusions that were based upon the findings and have practical implications.

In Context

India has one of the largest populations with HIV infection. The funding available for HIV programs in India has been increased from US$13.8 million in 2005-2006 to US$2.6 billion from 2007 through 2012. A large proportion of this funding will be allocated to prevention programs. Distribution of prevention funds should be based upon an understanding of where they will have the greatest effect on preventing new infections.

Programmatic Implications

With scarce resources, funds for HIV prevention programs should be targeted where they are most likely to have the greatest effect in averting new HIV infections. The methods used in this study can be applied in other areas to ensure that funds are used most efficiently.


  1. Joint United Nations Programme on HIV/AIDS. Epidemiological software and tools (2007): modes of transmission spreadsheets and manuals. Accessed June 2008.
  2. Royce RA, Sena A, Cates W Jr, Cohen MS. Sexual transmission of HIV. N Engl J Med 1997;336:1072-8.
  3. Gouws E, White PJ, Stover J, Brown T. Short term estimates of adult HIV incidence by mode of transmission: Kenya and Thailand as examples. Sex Transm Infect 2006;82 (Suppl 3):iii51-5.
  4. Weller S, Davis K. Condom effectiveness in reducing heterosexual HIV transmission. Cochrane Database Syst Rev 2002;1:CD003255.
  5. Dandona L, Dandona R, Kumar GA, et al. How much attention is needed towards men who sell sex to men for HIV prevention in India? BMC Public Health 2006;6:31.
  6. Stover J, Bertozzi S, Gutierrez JP, et al. The global impact of scaling up HIV/AIDS prevention programs in low- and middle-income countries. Science 2006;311:1474-6.
  7. Bollinger L, Cooper-Arnold C, Stover J. Where are the gaps? The effects of HIV-prevention interventions on behavioral change. Stud Fam Plann 2004;35:27-38.
  8. World Health Organization. Antiretroviral drugs for treating pregnant women and preventing HIV infection in infants. Geneva: WHO;2004.
  9. Baggaley RF, Boily MC, White RG, Alary M. Risk of HIV-1 transmission for parenteral exposure and blood transfusion: a systematic review and meta-analysis. AIDS 2006;20:805-12.