University of California, San Francisco Logo

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

Forecast of demand for antiretroviral drugs in low and middle-income countries: 2007-2008
Global Health Sciences Literature Digest
Published October 15, 2007
Journal Article

Galarraga O, O'Brien ME, Gutierrez JP, Renaud-Thery F, Nguimfack BD, Beusenberg M, et al. Forecast of demand for antiretroviral drugs in low and middle-income countries: 2007-2008. AIDS 2007 Jul;21 Suppl 4:S97-103.


To estimate the volume of current and future demand for first- and second-line antiretroviral drugs in low- and middle-income countries for the years 2007-2008.

Study Design

This forecast analysis was accomplished using regression analysis and documented assumptions. Forecasts were developed in two major steps: 1) estimating the number of individuals receiving antiretroviral drugs at the end of a given year, and 2) forecasting antiretroviral product demand. Projection of patient load was done using linear regression prediction based on the last three observations available (December 2005, June 2006, and December 2006) from 127 countries. This approach assumed that the rate of change in antiretroviral therapy (ART) scale-up and its associated constraining factors remain constant. It was also assumed that an individual becomes eligible for treatment two years before death in the absence of ART. This estimation included patients being treated on both first- and second-line drugs. The probability of migration from first- to second-line drugs was evaluated based on evidence from a WHO survey,(1) Clinton Foundation field reports, and consultation with experts. Survival assumptions were based on empirical data from Senegal, Brazil and the United States.(2,3,4) Two approaches were used to forecast the demand for antiretroviral drugs: empirical and normative. The empirical model used country data from a WHO survey on current regimen distribution and composition.(1) Since this exercise did not allow for a forecasting model that takes different determinants of change into consideration, the relative share of each drug regimen was assumed to remain constant over the two-year projection period. Data from countries not represented in the survey were extrapolated from the empirical model. The normative approach used simulated patient histories and current and future country guidelines to estimate how relative volumes may change over time. This approach assumed that countries implement their national treatment guidelines. Twenty-one high-volume base countries were used in the normative modeling and scale-up to December 2008. Product volumes used for regional and global estimates were inflated to reflect demand for countries where there were no data available for analysis.


Low and middle-income countries.


There was no intervention.

Primary Outcomes

Primary outcomes are the total number of patients on ART, total number of individual doses for the given years, and metric tons of active pharmaceutical ingredient needed to treat the estimated number of patients.


The reported number of individuals on treatment was 2.01 million in December 2006, and estimated to increase to 2.69 million by the end of 2007, and 3.38 million by the end of 2008. Detailed information from 30 countries was used, representing 89.5% of the total volume of antiretroviral-treated patients in low- and middle-income countries as of December 2006. The largest contributors in terms of patient load were Kenya (6.2%), Thailand (5.6%), and Uganda (4.8%). The normative model used data from 21 countries, representing 83% of the total global volume. The highest-volume countries for this approach were South Africa (16.1%), Brazil (8.9%) and Kenya (6.2%). Fourteen countries were represented in both samples. Globally, the number of individuals on second-line treatment was estimated to increase from 4.6% to 5.4% of the total ART users between 2007 and 2008, representing approximately 120, 000 individuals in 2007 to over 180, 000 in 2008. The largest volume of antiretroviral drugs needed based on both empirical and normative estimations was predicted to be for lamivudine, zidovudine, efavirenz, and nevirapine. The largest demand increase was seen in emtricitabine, tenofovir, abacavir, didanosine, and lopinavir. The largest single increase was for emtricitabine, which was recently added to the national guideline of a country.


The authors conclude that the two approaches used for antiretroviral drug demand forecasting should not be interpreted as a range of possibilities, but rather as two alternative scenarios, one representing 'what is happening' and the other 'what should be happening'. The results from this analysis suggest that countries will require much more tenofovir, approximately double the current quantities of zidovudine, and less stavudine. The results also suggest that less saquinavir and nelfinavir will be required than are currently being used. Active pharmaceutical ingredient (API) capacity, based on current reporting, is likely to be adequate for most APIs, except lopinavir. The authors expect that countries will be bringing practice more in line with their own guidelines, with many of the barriers to implementing guidelines lowered. They also note that any changes, dramatic or otherwise, in treatment scale-up or product preference will likely occur gradually, and thus will not affect their forecast for the upcoming 12 months. The authors state that this forecast will be regularly updated and that continual improvement can be made to the projection model, using clinical data and the latest WHO-published number on ART patients.

Quality Rating

In the absence of more accurate data, this analysis resulted in a clear forecast that can be easily updated. Some limitations to this analysis include 1) the assumption that there will not be any significant changes in the HIV treatment landscape (though authors did not have accurate future financing and drug delivery capacity information); (5) 2) both approaches are assumed to have a representative sample of all low- and middle-income countries; 3) the analysis relied on data from published studies and not from actual volumes procured by countries; and 4) the assumption that countries will continue to scale up at the same rate. The empirical model extrapolates current trends linearly, and for short-term prediction, may be more accurate if there is no change in relative volumes. The normative model takes into account planned changes in country protocols, but does not allow for deviations from protocol.

In Context

According to the authors, this analysis is the first attempt in the literature to explicitly quantify antiretroviral drug API demand for all low- and middle-income countries. Other API forecasting has only been done by manufacturers confidentially.

Programmatic Implications

Antiretroviral drug demand forecast information should be made available to industry, governments, and researchers to improve planning and implementation of HIV treatment programs. The information can reduce risk of supply shortages, ensure uninterrupted supply, and aid in negotiations on price reductions.


  1. Renaud-Thery F, Nguimfack BD, Vitoria M, Lee E, Graaff P, Samb B, et al. Use of antiretroviral therapy in resource-limited countries in 2006: distribution and uptake of first- and second-line regimens. AIDS 2007 Jul;21 Suppl 4:S89-95.
  2. Etard JF, Ndiaye I, Thierry-Mieg M, Gueye NF, Gueye PM, Laniece I, et al. Mortality and causes of death in adults receiving highly active antiretroviral therapy in Senegal: a 7-year cohort study. AIDS 2006 May 12;20(8):1181-9.
  3. Marins JR, Barros MB, Machado H, Chen S, Jamal LF, Hearst N. Characteristics and survival of AIDS patients with hepatitis C: the Brazilian National Cohort of 1995-1996. AIDS 2005 Oct;19 Suppl 4:S27-30.
  4. Walensky RP, Paltiel AD, Losina E, Mercincavage LM, Schackman BR, Sax PE, et al. The survival benefits of AIDS treatment in the United States. J Infect Dis 2006 Jul 1;194(1):11-9.
  5. UNAIDS. Resource Needs for an Expanded Response to AIDS in Low- and Middle-Income Countries. 2006; 1-39. [PDF, 976 KB]