The number of people receiving ART in low- and middle-income countries has expanded rapidly in recent years. Much work has focused appropriately on what is required to continue this scale-up to reach all those in need of treatment. This study examines a previously unexplored aspect of treatment, the financing required to continue supporting patients receiving ART.
Investments by the Global Fund together with other partners have enabled and catalyzed a rapid expansion of HIV programs in low- and middle-income countries that is reducing mortality and new infections 
. As of the end of 2009, Global Fund grants co-supported, fully or partially, around half of all patients receiving ART globally. For the 3.5 million patients receiving ART as of 2011, at current cost levels the annual resource need ranges from US$ 1.9 billion in 2011 to US$ 1.7 billion in 2020. The Global Fund does not support all of these costs; the USA President's Emergency Plan for AIDS Relief (PEPFAR), other donors, national governments and other domestic resources also provide partial support for the patients benefiting from Global Fund investments. Based on expenditure reporting from Global Fund recipient HIV/AIDS programs, across countries with Global Fund-supported ART the average annual expenditure from Global Fund grants per person-year of ART was estimated at about a quarter of the full program-level ART delivery costs estimated in this paper 
. This is roughly consistent with the Global Fund's share in overall international AIDS control funding, which in 2008 equalled 19% 
The costs of treatment in this analysis vary only by first and second line treatment. In reality costs may be different for different types of patients, pregnant women, children, TB patients. We have used an average per patient cost, which implicitly assumes that the distribution of patients by type will not change.
Sensitivity analyses showed that program costs are sensitive to ARV prices and regimens used, in particular for second-line ARV regimens. A balanced use of lower-cost generic ARVs and more expensive innovator drugs will impact costs. Efforts to improve treatment quality to retain patients on first-line regimens will be critically important to manage future costs. Equally, considerable variations in the protocols and frequencies of laboratory testing and associated cost point to opportunities for efficiency gains in patient monitoring. Recent studies in resource-constrained settings suggest that CD4-based patient monitoring is generally cost-effective compared to clinical monitoring alone, but that viral load (HIV RNA) monitoring is not always cost-effective 
. While lack of laboratory capacity should not impede countries to roll-out first-line ART, there is a need to develop affordable, point-of-care viral load and CD4 assays that can supplement clinical monitoring and ensure timely detection of treatment failure and regimen switches 
To further refine this type of assessment, we will need progressively better information on country-level variation and determinants of cost of ART delivery 
. Recently established routine tracking of national program budgets and expenditures can contribute to fill this current data gap 
, and inform the effects of ARV drug price declines, changing treatment eligibility criteria and regimen mixes, and the proportion of patients on first- and second-line ART 
. As of now, a paucity of country-level expenditure data on especially non-ARV components limits our understanding of current and future ART delivery cost in different settings. Routine expenditure data collection by national AIDS programs will also assist in optimizing value for money, an important determinant of future national and international (donor) commitments to ongoing ART scale-up 
Also, program achievements in terms of patient retention and survival, on first-line and second-line regimens, deserves improved monitoring and assessment. In-depth studies and evaluations in countries with relatively advanced patient tracking systems suggest that routine national program records may overestimate actual retention and survival in several high-HIV countries, given elevated mortality among patients lost-to-follow-up, and (passive) over-counting of retention time among patients who dropped out from recent clinic or pharmacy appointments. On the other hand, patients transferring between facilities may be over counted as new initiations and undercounted in retention 
. In high-HIV countries with substantive ART coverage, triangulation of program and clinical data with population-level mortality trends from vital registration, surveys and censuses would provide the ultimate reassurance on population-level health impact 
We estimate continued investment needs for a cohort of 3.5 million people receiving ARVs. The global need for treatment and financing of HIV is far larger than this number. The 2010 revised WHO HIV Treatment Guidelines 
raise the estimate of the number of people in need of treatment by about 50% 
. UNAIDS estimates suggest funding for ART would have to expand to US$ 7 billion annually by 2015 in order to achieve 80% coverage of those most acutely in need (with CD4 counts under 200 cells/µL), and would further increase by 50% to achieve 80% coverage under the expanded treatment eligibility criteria (covering all those with CD4 counts under 350 cells/µL) 
. Globally the number of patients on ART has been increasing at almost one million patients per year between 2006 and 2008. If that rate of scale up were to continue through 2020 and the Global Fund continued to support about 62% of ART patients as it did in 2009 
, then the number of patients needing Global Fund support would increase to 9 million by 2020. This would correspond to an annual co-investment need by the Global Fund of $5.2 billion—assuming constant grant expenditures per patient supported and continued proportionate co-investments from domestic sources, PEPFAR and other donors. However, there is no guarantee that scale-up in funding from domestic and other donor sources would keep pace with the ongoing rate of new ART patient initiations. Thus the Global Fund is likely to remain an important source of support for most countries. While some middle-income countries with strong economic growth could assume greater responsibility for funding their own programs, thus reducing the Global Fund investment needs, this is unlikely to be the case in low-income countries where the majority of Global Fund beneficiaries live 
While this study examined ART costs for existing patients through the year 2020, investment needs continue beyond this date, as by that year 2.3 million people (66% of the 2011 cohort) will still be alive and in need of continuing treatment Support for these patients will also include non-ART costs such as prevention and treatment of opportunistic infections including tuberculosis and malaria 
. Costs considered did not include the above-facility level costs such as human resources at drug distribution centers, district/provincial/national program management, monitoring and evaluation, health worker training in AIDS management and health system strengthening in general. A recent evaluation across 6 PEPFAR-supported African AIDS programs estimated above-facility-level costs at 20% of overall costs 
In conclusion, the rapid expansion of ART in low- and middle-income countries in recent years has produced dramatic declines in mortality and represents an important success in our efforts to improve global health. Sustaining and protecting gains will require sustained investment from international donors such as the Global Fund, as well as sustained contributions by the governments of supported country programs.