Stover, John, Geoff P. Garnett, Steve Seitz and Steven Forsythe, "The epidemiological impact of an HIV/AIDS vaccine in developing countries", World Bank Policy Research Working Paper Number 2811, April 2002.
"The Potential Epidemiological Impact of Prophylactic Vaccines: Results of the iwgAIDS Model" by Steven Thomas Seitz, Computational Modeling Laboratory, University of Illinois
This is the second of two annexes describing the simulation models used to investigate the potential impact of an AIDS vaccine.
The iwgAIDS model
The iwgAIDS computer program provides a vehicle for modeling all major vectors of HIV transmission: sexual (heterosexual, homosexual, bisexual), intravenous drug (stimulant and/or opiate), mother-to-child, and blood within the context of a fully articulated demographic model that includes modules for fertility, mortality, and migration. Transmission cofactors include other sexually transmitted diseases (ulcerative and/or inflammatories) and sexual practices (e.g., dry sex, anal sex, condom use). The model is age-structured and stratified by region (e.g. rural and urban), sex, marital status, sexual preference, and infection status. A continuous risk formulation incorporates age-specific parameters, by stratification, for type and concurrency of partnerships (long-term, short-term, single encounter) and contact rates (entropy distributions for high risk and low risk). Infectivity may vary over a long incubation period, and transition from HIV-infected status to AIDS is modeled using separate probabilistic transition grids for children and adults. Another probabilistic transition grid models the transition fro AIDS to death. The user has control over all these parameters (e.g. the length of the transition grid, the structure of the probabilistic function over the long incubation period, etc.).
Prophylactic vaccine routines allow systematic modification of susceptibility and therapeutic routines modify infectivity. The vaccine modules allow separation of efficacy and decay by mode of infection. Similar distinctions may be entered for region, sex, and risk. The modeling reported here assumes a constant efficacy and decay across mode of infection and stratification.
Although the user may set the age of onset for sexual activity, these simulations assume that individuals begin entering the sexually active population at 15. For the adult vaccine series, the proportion vaccinated ramp up from 15 to 18.75 and then begin the ramp down process after 50. For the teen series, we begin the ramp up process at 12.75 and then ramp down starting at 18.75 years of age. The high-risk series focuses on specific strata (high risk single males, high risk single females, high risk paired males, and high risk paired females) in the simulation, although the distribution of risk in each environment has an associated age structure. For these simulations the antenatal mother series focuses on paired women during their childbearing years. Pairing rates in both environments are age-structured, so this is implicitly imported into the antenatal series as well.
The duration of a vaccine’s efficacy decays by a hazard function whose half-life is set alternately at 5 years, 10 years, or 50 years for the purposes of these simulations. Compared to a straight exponential decay, the half-life formulation results in slower extinction of coverage and hence this model will be less sensitive to changes in duration. While the straight exponential decay will demonstrate sensitivity to duration regardless of target series (all adults, teens, antenatal mothers, or high risk people), this formulation allows duration sensitivity to vary by target. The iwgAIDS model allows movement from one risk distribution to another with changes in age, pairing status, etc. The diffusion of high risk people into lower risk strata will in turn make this target less sensitive to changes in duration of efficacy, because risk status is more a function of life cycle than a fixed attribute on an individual that remains constant over the individual’s life cycle.
Publication date: January 27, 2003