HIV is a major public health problem in the developing world. In South Africa, HIV disease is epidemic. Over the past 12 years, the HIV prevalence in the 15- to 49-year-old age group has grown at an alarming rate from less than 1% to 20%.1 The situation is even more dire among women in South Africa, where nearly 24% of all women between the ages of 20 and 24 years are HIV-positive. In the 15- to 24-year-old age group, girls and women comprise nearly two thirds of HIV-infected individuals. This burden is likely to worsen for women, because their risk of acquiring HIV infection is 7 times higher than the risk in men.2
Over the past decade, there have been significant advances in the interventions to prevent and treat HIV. They include patient education, condom promotion, voluntary counseling and testing, and, most notably, highly active antiretroviral therapy (HAART) in most industrialized countries. In many developing countries, however, including South Africa, the lack of adequate health care infrastructure and resources poses considerable obstacles to treating and limiting the spread of this disease.
The concept of a prophylactic vaccine against HIV infection has always interested researchers and international health organizations as an important strategy to control the HIV epidemic. Progress toward a truly effective vaccine has been limited by HIV's envelope protein genetic diversity, high replication rate, high mutation rate, and integration into the host cell genome, however. It is therefore likely that the first generation of HIV vaccines is going to be only partially effective in preventing HIV infection.3
In anticipation of the development of a prophylactic vaccine against HIV, it is useful to examine its clinical benefits and potential costs through mathematic modeling. Previous models have evaluated a vaccine's effect on high-risk behavior,4 protection of a community by a low-efficacy vaccine of short duration,5 and postponement of AIDS by a vaccine that does not prevent infection.6 Other models have examined the difference between preventive and therapeutic HIV vaccines7 and the economic benefits of a vaccine in developing and developed countries.8 All these models have targeted large populations in examining the cost-effectiveness of a potential HIV vaccine and, in some cases, have used herd immunity as an added protected feature for the population. None of these studies, however, targeted a specific population in a geographic area where the practical implementation of a new HIV vaccination program would originate.
This study examines the public health benefits and cost of a partially effective prophylactic HIV vaccine in a population restricted to 15-year-old South African adolescent girls and future offspring. We developed a Markov model to study the effects of this vaccine given to adolescent girls under a variety of HIV incidence rates and vaccine efficacy rates. The model follows this cohort over 10 years under a near-universal HAART utilization program and the status quo while using South African-specific demographic and HIV data.
In our modeling process, we include only female subjects, not to imply that excluding male subjects in a vaccination program is prudent but to demonstrate the profound effect of a limited program that focuses on the cooperation of girls and women. Vaccinating boys and men can serve to multiply a vaccine's beneficial effects. By agreeing to be vaccinated against HIV, however, a girl or woman is making a decision that can reduce her risk of acquiring HIV that does not need the cooperation of her partner. In addition, vaccinating female subjects has the profound effect of potentially preventing future mother-to-child transmission.
This cohort of 15-year-old adolescents was chosen because of accessibility through school-based programs, relatively low prevalence rates (not likely to be infected from vertical transmission and still not at peak sexual activity), and precedence of previous public health interventions through rubella vaccination or informed HIV education programs. Evidence of the accessibility of this age group is found in the popularity of LoveLife, South Africa's national HIV prevention program for youth, which combines a nationwide media campaign of education with sexual health services and community level outreach and support programs. A national survey among South African young people found that 85% of 15- to 24-year-olds have at least seen or heard of LoveLife and two thirds know of at least 4 of its program's initiatives.9 The penetration of LoveLife's message within this population shows that this group can be effectively reached by a coordinated national campaign with a clear message.
Vaccination occurs only once in our population, which consists of all South African 15-year-old adolescent girls as determined by the 2001 South African census. This one cohort is then followed for 10 years with no further vaccine booster as the beneficial effect of the vaccine gradually wanes. To model waning vaccine efficacy, we assume that 5% of all vaccinated individuals would lose all benefits of the vaccine yearly.
Putting a price on a theoretic HIV vaccine can be difficult. It is possible that an HIV vaccine may be produced in a similar fashion to the recombinant hepatitis B virus vaccine. In South Africa, the price for hepatitis B virus vaccine varies between US $3 to $5 per 10-dose vial down from $25 per dose a few years ago.10 In a 1997 study of mass measles vaccination in South Africa, the vaccine implementation costs were estimated to be approximately $1 and included vaccine cost, syringe cost, the nurse's time, and 20% operating costs.11 Specific studies have been done in which the cost of a theoretic HIV vaccine has varied from $5 to $100 per dose, including delivery, with $20 per dose as the favored compromise price.12 This study assumes the $20 dose in most of its models. In Figure 1, however, a sensitivity analysis varying the vaccine cost is performed.
The partially effective vaccine used in the model is considered to be “leaky” in that it provides only partial protection from HIV to everyone vaccinated.13 For example, a 50% effective vaccine decreases HIV incidence by 50% to everyone vaccinated per year. This vaccine with 50% efficacy is assumed to have a 100% “take” effect (fraction of vaccinated persons who show a response) and a 50% “degree” effect (degree of protection) where efficacy = take × degree.14 We analyze scenarios in which 50%, 40%, 30%, 20%, and 10% effective vaccines are compared with situations without vaccines. Other characteristics of the hypothetic vaccine include a 90% compliance rate and an initial cost of $20 for a course of vaccine per person. In the situation in which a vaccinated individual becomes infected with HIV, we assume the vaccine still affords certain advantages to the adolescent, such as decreased mother-to-child transmission of the virus for the next 6 years.
There are 8 possible states for an adolescent in our model, as shown in Figure 2. The first state is “Healthy Vaccinated” in which the individuals are HIV-negative and have generated the immune protection from the vaccine. The second state is “Healthy Unvaccinated” in which the individuals are HIV-negative but are vulnerable to the virus just as much as any other individual. These individuals consist of those who refuse vaccination or those for whom vaccine efficacy has completely worn off. The third state “HIV+ Asymptomatic Vaccinated” includes the adolescents who receive the vaccine but acquire HIV breakthrough infections anyway. These adolescents are assumed to have lower mother-to-child transmission rates than unvaccinated individuals for the following 6 years. In addition, these individuals are assumed to have a delay to progression to AIDS because of an assumed vaccine effect on viral set point, which wanes over time at a rate of 10% per year. The fourth state is “HIV+ Asymptomatic Unvaccinated” in which individuals acquire HIV infection and have a probability of advancing to AIDS depending on the number of years they has been HIV-positive. The next 2 states are “AIDS” and “AIDS on HAART.” The last 2 states include “Dead From AIDS” and “Dead From Other Causes,” which capture AIDS-related mortality and all-cause mortality, respectively.
At the beginning of our model, all 15-year-old adolescent girls are offered the vaccine. Based on prevalence data,15 8% of these women are already infected with HIV or have an AIDS diagnosis, and we assume that the vaccine has no effect. Therefore, the cohort is initially divided into “Healthy Vaccinated,” “Healthy Unvaccinated,” “HIV+ Asymptomatic,” and “AIDS” states based on the published literature.
As an individual progresses from year to year in the model, she experiences outcomes as would any other South African girl or woman of similar age, such as living uneventfully, becoming pregnant, becoming infected with HIV, or dying from other age-appropriate causes. HIV-positive individuals who are pregnant or nursing mothers can pass the virus on to their offspring. Progression from this asymptomatic HIV-positive state to AIDS depends on the number of years since seroconversion, using data from the CASCADE EU Concerted Action reports.16 Once an individual advances to AIDS, she has an additional chance of dying from AIDS-related illnesses. The number of adolescent HIV infections, infant HIV infections, and adolescent AIDS deaths over the 10-year period are recorded. Major events for the infants, in turn, include death or a chance of acquiring the AIDS virus from their mother through vaginal delivery or after 1 year of breast-feeding.
The determination of HIV incidence can be difficult, especially among subpopulations. In a 1999 study by Gouws et al, 15 a laboratory test was used to determine the incidence of HIV-1 subtype C infection among women attending public sector antenatal clinics in Hlabisa, a rural district in KwaZulu-Natal, South Africa. These results, which were confirmed by a mathematic model using prevalence data, estimated that incidence rates peaked at 22% for 22-year-old women. In the recent South African National HIV Prevalence, HIV Incidence, Behavior and Communication Survey, 2005, HIV incidence rates ranged between 0% and 11.6% depending on the age and gender of the individual.2 In light of the variety of possible incident rates possible within South Africa, we used incidence rates of 2.5%, 5%, 10%, and 15% to provide maximum applicability and adaptability as the incidence rate changes through time and geography. In addition, by using this range of incidence, we hope to model situations that might arise in other countries that might be progressing through different stages of an HIV epidemic.
The data used in the model are summarized in Table 1 The major sources of information include International Programs Center of the US Census Bureau; South African Demographic and Health Survey 1998; Actuarial Society of South Africa (ASSA) 2000 model; and United Nations Population Fund, State of World Populations for 2000.
In the scenario in which CD4 levels drop to less than 200 cells/μL and such patients would receive HAART, we assume that this policy is implemented in incremental steps, with 5% treated in the first year and 95% treated in the fourth and subsequent years. We used Treeage Pro Healthcare software (TreeAge Software, Inc., Williamstown, MA) to produce our model. Graphs were produced using SPSS 12.0 software (SPSS Inc., Chicago, IL).
Before the hypothetic vaccine is introduced into the population, data from the model were calibrated against the output from the Excel-based model used by the ASSA 2000 model in their simulations.18 The output of our model compared favorably in all 3 parameters (adolescent infections, HIV infections, and adolescent deaths) with the ASSA 2000 model, with a less than 1% difference in adolescent deaths and a 3% difference in total adolescent and HIV infections.
Costs and their sources are summarized in Table 2 Over the course of the model's 10-year evaluation, a discount rate of 3% is used. The ultimate HIV-related medical costs for the cohort of adolescents over the course of 10 years are compared between the vaccination and nonvaccination scenarios.
The cost of the vaccine, which includes the cost of delivery and implementation, can be variable, but we chose a cost of $20 per person for the group and investigated a range up to $100 per person. The difference in total medical costs between the 2 scenarios is determined to be cost-saving if the vaccination program costs less to the South African government as compared to not implementing it. If the cost of the vaccination program is less expensive than no program, a positive cost-utility ratio of dollars per infection averted is reported. Total cost of program implementation is calculated by multiplying vaccine cost and delivery by the number of 15-year-old female adolescents in South Africa.
With increasing vaccine efficacy, the number of adolescent HIV infections prevented increases. As shown in Figure 3, even under a variety of assumed HIV incidence rates, the drop in infections ranges from 31.4% in the 2.5% incidence group to 19.8% in the 15% incidence group over 10 years for a 50% effective vaccine. The greatest number of adolescent infections prevented occurs at the highest incidence rate, when nearly 77,000 infections are prevented over the period of 10 years.
Even with moderate use of antiretroviral medications in South Africa, where nearly two thirds of HIV-positive pregnant mothers receive antiretroviral prophylaxis, a partially effective vaccine can have dramatic effects in the reduction of HIV-positive infants at 1 year. As shown in Figure 4, mother-to-child transmission rates with a 50% effective vaccine can be decreased by 25.1% (4.7% to 3.5%) assuming a 2.5% HIV incidence rate and by 27.0% (14.8% to 10.7%) assuming a 15% incidence rate.
Without the use of universal HAART, the number of HIV-related deaths over 10 years in this group is strongly affected by vaccine efficacy rates. The greatest reduction in deaths occurs with the highest incidence rates, as seen in Figure 5. There is a minimal benefit in mortality reduction at the lowest HIV incidence rate of 2.5%.
Under our second scenario, we assume that within 3 years, the South African government has widely adopted the use of HAART for AIDS patients. Starting from an initial use of 5% in the first year, 25% in the second year, and 60% in the third year, there is near-universal (95%) use of HAART for the remainder of the 10-year period. Simply with the universal use of HAART, the number of HIV deaths drops by roughly three fourths, as seen in Figure 6. In this scenario, an HIV vaccine does not have an appreciable impact of preventing deaths when HAART is universally used. If HAART is used near its current low rate of 5%, however, an HIV vaccine has an appreciable effect on HIV deaths, especially with the higher HIV incidence rates (see Fig. 5).
We model this cost-savings scenario assuming near-universal use of HAART (95%) by AIDS patients. Assuming that most of the HIV-positive adolescents would seek medical care in the government health service, the total medical cost for the South African government relating to HIV medical care for the 510,000 cohort of adolescent girls depends on the HIV incidence rates. From our model, the total medical costs range between $1.2 and $1.9 billion depending on the HIV incidence rates (Table 3) Implementation of the $9.2 million vaccine program represents cost savings in all scenarios in which the vaccine cost is $20 per dose. In fact, even at the lowest vaccine efficacy and lowest incidence rate modeled, the program has a cost savings of $5.8 million. A more realistic example of 5% incidence and 30% vaccine efficacy has a cost savings of $68.2 million over 10 years.
The following cost-effectiveness analysis evaluates the vaccination program as the vaccine price is increased by increments of $20 to a maximum of $100. As shown in Figure 1, $20 per dose is cost-saving to the South African government under all HIV incidence rates and vaccine efficacies, as further demonstrated in Table 3. In terms of cost-effectiveness, even the worst-case scenario at $100 per dose, 10% vaccine efficacy, and a 2.5% HIV incidence rate demonstrates that it would cost $4500 per HIV case averted. The cost-neutral plane is defined at $0 per infection averted.
With South Africa's high prevalence rate, our model shows that a partially effective vaccine primarily prevents infections in our 15-year-old adolescent cohort even in the setting of widespread HAART use. Depending on the incidence rate, a partially effective HIV vaccine can lower adolescent infections between 20% and 31%. In the same way, the partial protection extends to their children during pregnancy and reduces mother-to-child transmission not only during the perinatal period but through the breast-feeding phase, resulting in a decrease of infant infections between 25% and 27%. A vaccine's primary advantage is prevention of infections in the first place, whereas the benefits of HAART occur later when a person is symptomatic.
Our model also shows that providing HAART to AIDS patients overshadows the partial effect of a future vaccine on progression to AIDS. In the presence of widespread HAART use, a partially effective vaccine (<50% effective) does not have a dramatic effect on adolescent mortality. A partially effective vaccine may lower HIV mortality, however, especially in settings in which HAART use has not yet become widespread.
An important result of our study is the extent to which a partially effective vaccination program can be cost-saving with respect to annual government medical expenditures. Our model suggests that even in the worst cost-effective scenario, where a 10% effective vaccine is used in a setting of a low HIV incidence rate of 2.5%, a $20 per dose vaccination program is not only cost-effective but cost-saving of approximately $6 million. In fact, in this setting, the cost of the vaccine could rise to approximately $40 per dose before it becomes cost-neutral. This provides more impetus for the South African government to consider using any partially effective HIV vaccine when one does become available. In addition, it is important to emphasize that the vaccination program is a single-time cost (excluding the need of a booster), whereas the use of HAART is a continual annual expenditure for a health care system with rising costs as the number of chronically HIV-infected individuals increases annually.
The characteristics of an eventual HIV vaccine are likely to dictate the circumstances of how, when, and where an HIV vaccination program is implemented throughout the world. This model, although based on South Africa, can be adapted for much wider applicability, and therefore could provide some assistance to the decision-making processes involved in using HIV vaccination programs. One advantage of the current model is its ability to be adapted to new circumstances. With changing HIV incidence rates, medical costs, or extent of HAART use, the model can be adapted for new scenarios as the situation changes. In addition, with a change of the basic model parameters, other countries may find this model useful in determining the benefits and cost-effectiveness of a partially effective HIV vaccination program.
It is important to realize the limitations of our study. Although the model is flexible enough to change HIV incidence rates each year, for simplicity and lack of evidence, we did not change HIV incidence rates over the 10-year simulation because of change in individual behavior. In addition, a model is as good as its data. Although we have extensively searched the literature for clear, accurate, and relevant data, many variables in our model are constantly changing as the situation changes in South Africa.
This study demonstrates that even a partially effective vaccine has an important role in the presence of widespread HAART use and should be implemented when such a vaccine becomes available.
2. Shisana O, Rehle T, Simbayi LC, et al. South African National HIV Prevalence, HIV Incidence, Behaviour and Communication Survey, 2005
. Cape Town, South Africa: Human Sciences Research Council; 2005.
3. McLean AR, Blower SM, Auvert B, et al. Modelling HIV vaccination. Trends Microbiol
4. Bogard E, Kuntz KM. The impact of a partially effective HIV vaccine on a population of intravenous drug users in Bangkok, Thailand: a dynamic model. J Acquir Immune Defic Syndr
5. Anderson RM, Swinton J, Garnett GP, et al. Potential impact of low efficacy HIV-1 vaccines in populations with high rates of infection. Proc R Soc Lond B Biol Sci
6. Van Ballegooijen M, Bogaards JA, Weverling GJ, et al. AIDS vaccines that allow HIV-1 to infect and escape immunologic control: a mathematic analysis of mass vaccination. J Acquir Immune Defic Syndr
7. Owens DK, Edwards DM, Shachter RD. Population effects of preventive and therapeutic HIV vaccines in early- and late-stage epidemics. AIDS
8. Bishai D, Lin MK, Kiyonga CW. Modeling the economic benefits of an AIDS vaccine. Vaccine
9. Pettifor AE, Rees HV, Steffenson A, et al. HIV and Sexual Behaviour Among Young South Africans: A National Survey of 15-24 Year Olds
. Johannesburg, South Africa: Reproductive Health Research Unit, University of the Witwatersrand; 2004.
10. Schoub BD, Matai U, Singh B, et al. Universal immunization of infants with low doses of a low-cost, plasma-derived hepatitis B vaccine in South Africa. Bull World Health Organ
11. Uzicanin A, Zhou F, Eggers R, et al. Economic analysis of the 1996-1997 mass measles immunization campaigns in South Africa. Vaccine
12. Stover J, Garnett GP, Seitz S, et al. The Epidemiological Impact of an HIV/AIDS Vaccine in Developing Countries: Working Paper Number 2811
. Washington, DC: European Commission and the Development Research Group of the World Bank; 2002.
13. Halloran ME, Haber M, Longini IM, Jr. Interpretation and estimation of vaccine efficacy under heterogeneity. Am J Epidemiol
14. McLean AR, Blower SM. Imperfect vaccines and herd immunity to HIV. Proc R Soc Lond B Biol Sci
15. Gouws E, Williams BG, Sheppard HW, et al. High incidence of HIV-1 in South Africa using a standardized algorithm for recent HIV seroconversion. J Acquir Immune Defic Syndr
16. Collaborative Group on AIDS Incubation and HIV Survival, Including the CASCADE EU Concerted Action. Time from HIV-1 seroconversion to AIDS and death before widespread use of highly-active antiretroviral therapy: a collaborative re-analysis. Lancet
17. Republic of South Africa Department of Health. National HIV and Syphilis Antenatal Sero-prevalence Survey in South Africa: 2002
. South Africa: Republic of South Africa Department of Health; 2003.
21. Sewankambo NK, Gray RH, Ahmad S, et al. Mortality associated with HIV infection in rural Rakai District, Uganda. AIDS
22. Fawzi W, Msamanga G, Renjifo B, et al. Predictors of intrauterine and intrapartum transmission of HIV-1 among Tanzanian women. AIDS
23. Guay LA, Musoke P, Fleming T, et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomized trial. Lancet
24. Morris AB, Cu-Uvin S, Harwell JI, et al. Multicenter review of protease inhibitors in 89 pregnancies. J Acquir Immune Defic Syndr
25. Le Coeur S, Kanshana S, Jourdain G. HIV-1 transmission from mother to child and its prevention. Med Trop
27. South African Department of Health. 1998 South Africa Demographic and Health Survey
. Republic of South Africa; 1998.
28. Brocklehurst P, French R. The association between maternal HIV infection and perinatal outcome: a systematic review of the literature and meta-analysis. BJOG
29. Nakiyingi JS, Bracher M, Whitworth JA, et al. Child survival in relation to mother's HIV infection and survival: evidence from a Ugandan cohort study. AIDS
30. Schim van der Loeff MF, Jaffar S, Aveika AA, et al. Mortality of HIV-1, HIV-2 and HIV-1/HIV-2 dually infected patients in a clinic-based cohort in The Gambia. AIDS
31. Orrell C, Bangsberg DR, Badri M, et al. Adherence is not a barrier to successful antiretroviral therapy in South Africa. AIDS
32. Mofenson LM, Lambert JS, Stiehm ER, et al. Risk factors for perinatal transmission of human immunodeficiency virus type 1 in women treated with zidovudine. Pediatric AIDS Clinical Trials Group Study 185 Team. N Engl J Med
33. Blecher M, Thomas S. Health care financing. In: Ijumba P, Day C, Ntuli A, eds. South African Health Review 2003/04
. Durban, South Africa: Health Systems Trust; 2004. Available at: http://www.hst.org.za/publications/423
34. Council for Medical Schemes. Annual Report of the Registrar of Medical Schemes 2003-4
. Pretoria, South Africa: Council for Medical Schemes; 2004. Available at: http://www.medicalschemes.com
35. Cleary S, Boulle A, McIntyre D, Coetzee D. Cost-Effectiveness of Antiretroviral Treatment for HIV-Positive Adults in a South African Township
. Durban, South Africa: Health Systems Trust; 2004.
36. Geffen S, Nattrass N, Raubenheimer C. The Cost of HIV Prevention and Treatment Interventions in South Africa: Working Paper No. 28
. Cape Town, South Africa: Centre for Social Science Research; 2003.
37. Skordis J, Nattrass N. Paying to waste lives: the affordability of reducing mother-to-child transmission of HIV in South Africa. J Health Econ
38. Mbewu A. Operational Plan for Comprehensive HIV and AIDS Care, Management and Treatment for South Africa
. South Africa: South African Department of Health; 2003.