Little K, Newell ML, Luo C, Ngongo N, Borja MC, McDermott P. Estimating the number of vertically HIV-infected children eligible for antiretroviral treatment in resource-limited settings. Int J Epidemiol 2007 Jun;36(3):679-87.
The complete text of this article and its very useful figures and tables are available free of charge at http://ije.oxfordjournals.org/cgi/content/full/36/3/679
To develop a model for estimating, by age, the number of vertically HIV-infected children eligible for antiretroviral therapy (ART) on the basis of clinical disease progression, allowing for antenatal HIV prevalence, use of interventions to prevent mother-to-child transmission (PMTCT), infant-feeding policies, availability of cotrimoxazole to prevent opportunistic infections, and ART.
The model's assumptions were informed by published evidence and expert opinion. It initially assumes a stable population with no changes in birth rates, population size, antenatal prevalence, coverage of PMTCT, or treatments over the previous ten years. The number of HIV-infected infants born depends on the probabilities of perinatal/in utero and breastfeeding transmissions, allowing for the extent of PMTCT and duration of breastfeeding; the proportion of HIV infected women attending antenatal clinics; and the crude live birth rate. Survival of infected children is independent of feeding method and PMTCT-exposure. Rates of progression to moderate to serious disease (MSD), as defined by World Health Organization guidelines, were inferred from the mortality of infected and uninfected children of HIV-infected mothers. It was assumed that 95% of children will develop MSD by age ten. Mortality rates for uninfected and infected children, as well as those receiving ART, were derived from African studies.(1,2,3) Treatment options were 1) co-trimoxazole only for prophylaxis of opportunistic infections; 2) ART only for children one year of age or older; 3) co-trimoxazole to all children and ART to children over one year of age; and 4) cotrimoxazole and ART to all children. The model allows for modifications in the number of infected and uninfected children over the ten-year period under different types of PMTCT treatment and coverage based on demographic information, antenatal prevalence, and treatment survival rates. To allow widespread use, the model was developed using the Microsoft Excel spreadsheet software.
The authors use South Africa as a hypothetical example for their model with published data on antenatal prevalence from 1993-2003,(4) as well as demographic information.(5)
PMTCT, as single-dose nevirapine (sdNVP), has been available in South Africa since 2000, and the authors assume a PMTCT program coverage of 11% in 2003,(6) as well as different infant-feeding approaches, in this example. Cotrimoxazole has been recommended for all HIV-exposed children since approximately 2000, and the authors assume that 100% coverage is reached at the end of 2003. Coverage of ART is assumed to be 0%. Antenatal prevalence information is available from 1993 to 2003, when the estimated population size was 47.432 million, and the birth rate 23.8 per thousand, giving 1,128,881 live births in 2003, which for an antenatal prevalence of 27.9% gives 314,958 children born to HIV-infected mothers. With sdNVP PMTCT, the in utero/perinatal transmission probability is assumed to be 0.1 and 0.2 without. With 11% coverage of sdNVP 18.9% of all exposed children would be infected at or before birth, in this case equaling 59,527 children infected by birth. The authors assumed that 5% of women refrain from breastfeeding with a postnatal transmission probability of zero, 50% to breastfeed for less than six months with a postnatal transmission probability of 0.04, and the remaining 45% to breastfeed for longer with a transmission probability of 0.20. Therefore, an additional 11% of exposed children, or 28,097, who were uninfected at birth are infected postnatally, for a total of 87,624 exposed children who acquire infection.
To calculate the number of children surviving to certain ages, the authors applied the cumulative mortality rate to the number of infected infants born in each of the past ten years, with mortality rates depending on available treatment in a given year. To calculate the number of children with MSD at each age, they apply the annual MSD rate to the corresponding birth cohort, allowing for the availability of cotrimoxazole and an MSD mortality rate. Finally, to calculate the number of children with MSD surviving into the next year, the mortality rates of the periods through which they have survived are applied, using the correct birth cohort to allow for changes in antenatal prevalence. The authors provide an illustrative example in which 670,970 infected births would take place over the ten-year period, leading to 53,602 new MSD cases, 21,605 surviving new MSD cases, and 45,479 total children eligible for ART.
|Without ART and cotrimoxazole prophylaxis, and with the rapid disease progression and high mortality rate early in life, as well as the difficulty of diagnosing vertically acquired infection in young infants, the number of infected infants surviving the first year who become eligible for treatment is comparatively small.|
|After the first year, the number becoming eligible for treatment remains fairly constant at ~50% higher than at the end of the first year.|
|When cotrimoxazole is available for all children under 18 months, and for all symptomatic children, the substantially reduced mortality results in an increased proportion of infants who develop MSD surviving long enough to be diagnosed and become eligible for treatment.|
|ART, when available only to children of one year or older (because of the difficulty of diagnosing infection in young infants and the rapid disease progression and high mortality in the first year), only impacts on numbers of surviving children known to be infected.|
|As children who become eligible for treatment remain on treatment for the rest of their lives and the mortality rate when on treatment is dramatically reduced, there is an accumulation of children in the treatment category.|
|With both ART and cotrimoxazole for all infants, there is a rise in the number of children eligible for ART and the number of children alive and not on treatment. By the end of the first year, approximately double the number of children would be eligible than without early treatment. Thus, the numbers on treatment across all age ranges then substantially increases.|
The authors conclude that this model is easy to implement, flexible, and can be used in ART programs at the national and local levels.
There is no standard for assessing the quality of this model. The authors do point out some limitations. First, they assume that children who develop MSD are put on antiretroviral treatment either the year or month following their disease progression, depending on the availability of rapid diagnostics. The authors note that treatment could be initiated immediately once services are functioning well, increasing the demand for ART to a limited extent. Alternatively, children may go undiagnosed for long periods of time due to incomplete access to appropriate care and treatment in some settings. Secondly, as with all modeling, the generated output relies on population data with a margin of error and assumptions regarding mortality and progression rates. These estimates are based on only a few studies and could either overestimate or underestimate mortality and thus lead to overestimating or underestimating the number of infected children in the model. The use of locally derived data would be more appropriate, if available. The model does allow for an understanding of the uncertainty of the estimates through sensitivity analysis of each parameter (transmission, mortality, MSD mortality, and MSD progression rates).
With the gradual rollout of ART delaying progression of HIV disease in children in programs across sub-Saharan Africa and resource-limited settings elsewhere, reliable information on the number of vertically infected children eligible for such treatment is urgently required. This study does not mention any other publicly available models that address this issue.
Policymakers and program managers can use this model in any region for which there are sufficient data to estimate the number of vertically HIV-infected infants born each year, as well as the estimated number and ages of HIV-infected children requiring ART. This model may be very useful in terms of projecting demand for treatment and determining optimal treatment strategies given limited resources. It could also be used for monitoring the success of intervention programs in PMTCT or ART. The lack of infrastructure and personnel familiar with PMTCT and pediatric HIV/AIDS care, however, may limit a region's ability to identify pregnant HIV-infected women and HIV-exposed and infected children, as well as implement successful treatment programs. The accuracy of the model may therefore be limited.
- Newell ML, Coovadia H, Cortina-Borja M, Rollins N, Gaillard P, Dabis F, et al. Mortality of infected and uninfected infants born to HIV-infected mothers in Africa: a pooled analysis. Lancet 2004 Oct 2-8;364(9441):1236-43.
- Chintu C, Bhat GJ, Walker AS, Mulenga V, Sinyinza F, Lishimpi K, et al. Co-trimoxazole as prophylaxis against opportunistic infections in HIV-infected Zambian children (CHAP): a double-blind randomised placebo-controlled trial. Lancet 2004 Nov 20-26;364(9448):1865-71.
- Fassinou P, Elenga N, Rouet F, Laguide R, Kouakoussui KA, Timite M, et al. Highly active antiretroviral therapies among HIV-1-infected children in Abidjan, Cote d'Ivoire. AIDS 2004 Sep 24;18(14):1905-13.
- Makubalo L, Netshidzivhani P, Mahlasela L, du Plessis R. National HIV and Syphilis antenatal sero-prevalence survey in South Africa 2003. Accessed November 30, 2005.
- UNPOP. World population prospects: The 2004 revision. Accessed November 19, 2005.
- USAIDS, UNAIDS, WHO, UNICEF, The POLICY Project. Coverage of selected services for HIV/AIDS prevention, care and support in low and middle income countries in 2003. Accessed March 12, 2004.