[procaare] Part 6 - Harvard Consensus Statement on ARV Treatment for AIDS in Poor Counties

[Richard Marlink <marlink@hsph.harvard.edu>]

*The Harvard 'Consensus Statement on Antiretroviral Treatment for AIDS in Poor Countries;
has been broken into 6 parts in text format to be easily received by forum members. To
view the entire document online, please go to: http://aids.harvard.edu - ProCAARE
Moderator*

PART 6

Consensus Statement on Antiretroviral Treatment for AIDS in Poor Countries

By Individual Members of the Faculty of Harvard University (*1)
[* To see the list of 148 signatories, please go to website as noted at the top *]

ANNEX A

Estimating the Cost of Expanded AIDS Treatment in Africa

As the main text of the Consensus Statement makes clear, low- income countries (i. e.,
those having an annual per capita GNP < $755 annually on World Bank criteria) lack
sufficient resources to finance AIDS treatment by themselves, even with discounts of 90%
or more on drug costs. (*49) A few somewhat wealthier developing countries (e. g.,
Botswana and South Africa) could finance limited AIDS treatment, but even then only a
fraction of their needs. With the current supply of domestic resources, no country in sub-
Saharan Africa can undertake widespread AIDS treatment; these countries are simply too
poor relative to the prevalence and costs of the disease. This argument is often lost in
popularized comparisons to Brazil, which has furnished free AIDS treatment to its
citizens. Brazil's ability to provide treatment stems from the following: first, Brazil's
average annual income is $4,400 (1999 estimates), and second, only 0.5% of adults there
are HIV- positive. This is in stark contrast to Sub- Saharan Africa, where the average
annual income is about $500 (1999 estimates) and the prevalence of adult infection about
9%, to say nothing of the most affected countries, where the infection rate can reach 40%.
(*50 *51)

The combination of low income and high HIV prevalence indicates that if AIDS treatment is
supplied in Africa, international aid will have to pay for nearly all of it. Additional
donor assistance also will be needed for countries where low income or high prevalence or
both put resource needs for AIDS treatment beyond the financial capacity of the national
government. Donated funds would finance both material requirements (e. g., medications,
including antiretroviral drugs and drugs for opportunistic infections) and operational
requirements (e. g., research and clinical operations) for AIDS treatment. We estimate
that as of today, Africa would represent approximately 80 percent of the global needs for
donor support and that remaining donor support would assist countries in South and
Southeast Asia (e. g., India, where nearly 5 million people are infected with HIV) and in
the Americas (e. g., Dominican Republic and Haiti). (*52) Accordingly, this Annex focuses
on the costs of AIDS treatment in Africa and recognizes that a global program would
require approximately 25 percent more in overall donor financing than the Africa- specific
program outlined here. We do not make cost estimates for the expansion of tuberculosis
treatment that is needed in any event and that should accompany an expanded AIDS treatment
effort, but endorse the additional funding of the global Stop TB campaign.

This costing model is based on a series of per- patient unit costs multiplied by the
number of patients treated. We perform the analysis as static, taking into account only
the need for treatment within the next 3 years. However, similar methods could be used to
project future costs by using epidemiological projections of HIV prevalence, incidence,
and future AIDS mortality to adjust the number of HIV- infected individuals needing
treatment.

1. HIV testing costs
Prior to receiving treatment, each patient must test obtain counseling and test positive
for HIV infection. Because the CDC and other agencies already have expended considerable
effort on widespread HIV testing in Africa, we have estimated additional testing costs
only for those most likely to benefit from immediate therapy. Determining HIV status is a
non- recurring cost on an annual basis. The cost of an episode of counseling and testing
has been estimated between $3 to $18, with the Harvard- Haiti project reporting a cost of
$7. This is consistent with the assumptions of other published studies. (*53 *54) Thus, we
assume conservatively that each episode of counseling and testing costs $10 for those who
test negative, and $20 for those whose test is repeated and who are confirmed positive. We
estimate an HIV prevalence of 30% among those tested, when targeted to patients in
hospitals and clinics. With an overall HIV prevalence of 10% in sub- Saharan Africa, it is
likely that targeted testing will yield 30% of patients infected. (*55) Of this 30%, we
estimate that 1 in 3 will have advanced HIV disease and therefore require treatment. Thus,
to achieve our goal of treatment for 1 million HIV- infected patients, approximately 10
million people will need to be tested. Of these 10 million individuals, three million will
test positive for HIV, with 1 million candidates for treatment. The breakdown is as
follows:
Initial screening tests
(10 million people) x ($ 10/ person) = $100 million (one time cost)
Confirmation of HIV- positive status
(3million people) x ($ 10/ person) = $30 million (one time cost)

It is important to note that counseling and testing expenses would be spread over several
years. That said, the above testing effort would cost $130 million total, or $43 million
annually if spread over three years. In addition to serving as a screening tool to select
candidates for treatment, counseling and testing has the added benefit of informing those
who are HIV- negative of their status, which has been shown to result in people changing
their behavior to avoid future HIV infection. (*56)

2. Drug costs:
For most patients (70%), we assume an annual drug cost of $500 per patient per year for
HAART (see main text). For the remaining 30% of patients, we assume a more expensive
regimen is necessary at increased costs of $1,000 per patient per year. This assumption is
based on data that show that patients who develop virologic resistance to an initial
regimen typically require more or different drugs in a "salvage" regimen as well as other
treatment strategies for late- stage AIDS. This yields a probability- weighted, per
patient drug cost of $650/ year across the board.

For symptomatic AIDS treatment, such as demonstrated by the Harvard group in Haiti (see
main text), we assume that only patients with advanced HIV disease satisfy the criteria to
begin treatment. Furthermore, because the time from AIDS onset to death is typically under
one year in Africa, (*57 *58) we estimate that the number of patients who would begin
therapy in Africa is roughly equal to the number of AIDS deaths reported by UNAIDS in
2000. Therefore, approximately 2.4 million people in Africa are anticipated to be
candidates for initial treatment. (*59) We calculate the drug costs for treating 1 million
patients as follows:
(1million people) x ($ 650/ patient- year) = $650 million/ year

It is important to note that this approach may underestimate the number of candidates for
treatment, because it is retrospective by one year in a growing epidemic and because the
number of AIDS deaths is an imperfect proxy for the number of people living with advanced
AIDS. In addition, three factors may further limit the number of patients who receive
initial treatment: (1) Not every AIDS patient will be interested in, willing, or able to
be treated; (2) Many AIDS patients are beyond the reach of the governmental or non-
governmental health systems, either as they exist now or as they might exist in the next
3- 5 years, and; (3) Not all countries presently have the top- level political commitment
to commence widespread AIDS testing and treatment.

Despite these limitations, we consider it ambitious but possible for 1 million people to
receive HIV/ AIDS treatment within 3 years. This would likely be less than one third of
late- stage AIDS patients in Africa, but over a one hundred- fold increase in the number
of such patients receiving HAART today.

We calculate the drug cost for treating 1 million patients as:
DRUG: (1million people) x ($ 650/ patient- year) = $650 million/ year

3. Directly observed therapy (DOT) costs:
If the drugs are administered through directly observed therapy, additional costs will
accrue. For DOT in Haiti, an accompagnateur (i. e., a treatment observer) is typically
paid $100/ month to supervise the medication of 6 patients. This would be an appropriate
wage level in most of Africa and would keep turnover of treatment observers low. Assuming
capital expenditures are negligible, the average cost per patient is therefore $200/ year.

Total annual costs for DOT are as follows:
DOT: (1million people) x ($ 200/ patient- year) = $200 million/ year

5. Clinical costs
For those who test HIV- positive and begin HAART, approximately 6 clinic visits annually
are likely to be needed to effectively monitor the therapeutic response to and toxicity
from antiretroviral drugs. Each clinic visit would require consultation with a physician,
nurse, or other health worker, and, if available, a panel of relatively inexpensive blood
tests. These tests would not include more expensive CD4 cellcounts and HIV viral load
testing, as these would be performed regularly only on those patients in clinical trials,
in order to determine the contribution of such tests to outcomes. Unit costs for an
outpatient consultation are very low in impoverished regions with poor health
infrastructure (sub- Saharan Africa, $3) and slightly higher in middle income countries
with a more established health infrastructure (Thailand, $14). (*60) Taking the latter
figure, plus an allowance for the blood tests and opportunistic infection prophylaxis, we
estimate that the total cost of each clinic visit would not exceed $25 per visit, or $150
annually. While the costs of laboratory tests, such as CD4 cell count and HIV viral load,
in the developing world are not well- defined, costs for a single CD4 cell count and HIV
viral load test are an estimated $80 per person per year to define treatment failure. We
estimate the clinical costs of ongoing treatment for 1 million patients as follows:

CLINICAL: (1million people) x ($ 230/ patient- year) = $230 million/ year

5. Clinical research
In keeping with the view that a scaling- up of AIDS treatment must be accompanied by
clinical research in order to determine optimal treatment strategies in poor countries,
additional costs will be associated with the enrolling and monitoring of patients in
different trials. These costs will vary greatly depending on the scientific question posed
by the trial and the laboratory or clinical work necessary for data collection. We
conservatively estimate that most trials can be supported for under $500 per patient per
year, an amount sufficient to enroll and follow each patient in the trial and to perform
periodic CD4 cell count or HIV viral load testing, at a remote facility if necessary. In
the United States, nearly 1 million people have been treated for AIDS, with about 100,000
of those (10%) enrolled in clinical trials. through the AIDS Clinical Trials Group, the
CPCRA, HIVNET, the VA system, and other research groups. Based on these numbers, we
estimate that in the first several years about 50,000 people in resource- poor countries
would participate in trials. Our calculations are as follows:

RESEARCH: (50,000 people) x ($ 500/ patient- year) = $25 million/ year

6. Total

Summing these costs, we estimate the following total:

TESTING: Annualized cost based on 3- year cycle (see above) = $43 million/ year
DRUG: (1million people) x ($ 650/ patient- year) = $650 million/ year
DOT: (1million people) x ($ 200/ patient- year) = $200 million/ year
CLINICAL: (1million people) x ($ 150/ patient- year) = $230 million/ year
RESEARCH: (50,000 people) x ($ 500/ patient- year) = $25 million/ year

TOTAL = $1.123 billion/ year

We conclude that that the total cost of treatment, comprising the above expenditures,
would be approximately $1,123/ patient- year, or slightly over $1.1 billion annually for
the 1 million patients that we believe can be treated in Africa within the next 3 years.
This number would increase in later years, as treatment could be expanded to a larger
number of patients. By year 5 the aim would be to increase coverage to 3 million
individuals or more.

This would require approximately $3.3 billion annually, a sum that is small in proportion
(0.01% of an aggregate GNP of nearly $23 trillion) to the wealth of the donor countries
called on to fund this effort. (*61)

Our estimate of $1,123 per patient per year is consistent with other studies which show
non- drug costs of delivering HAART in the range of several hundred dollars, or roughly on
par with the discounted price of antiretroviral drugs themselves. For example, researchers
in Brazil have reported the non- drug HAART costs of about $350/ patient- year for that
government's national treatment program. (*62)

World Bank estimates, at over $800/ patient- year, are somewhat higher. (*63) Both these
estimates include advanced diagnostics such as CD4 or viral load testing; however, they do
not make provision for directly observed therapy in order to maximize patient adherence
and forestall drug resistance, nor do they include the cost of clinical research in order
to collect data and therefore optimize AIDS treatment in poor countries.

We believe that an immediate effort to treat 1 million AIDS patients in poor countries, as
described in this document, can take place with a limited amount of investment in new
infrastructure, the cost of which is implicit in the figures we present. However, as
treatment is expanded to a larger number of patients in increasingly remote areas,
infrastructure will becom limiting unless there are additional outlays for training
medical personnel and capital expenditures for physical infrastructure. Such additional
outlays would have multiple benefits beyond HIV/ AIDS treatment, as they would support a
more general expansion of health services in Sub- Saharan Africa. We do not estimate those
additional outlays here.

Cost- Effectiveness Considerations

The above discussion focuses on the costs of AIDS treatment, without considering the
benefits or the "effectiveness" of treatment. Cost- effectiveness analysis considers both
factors, specifically the total cost of an intervention and its corresponding clinical
effectiveness in order to understand the value of treatment. These two outcomes are
compared as a ratio, or cost per unit of life expectancy. More advanced cost-
effectiveness analyses compare two or more interventions; the ratio is calculated as the
incremental change in total costs, divided by the incremental change in life expectancy,
compared to another intervention. In this scenario, the clinical benefit (or life
expectancy) is measured in years of life saved.

There is no question that HAART therapy is cost effective in rich countries, compared not
only to other HIV interventions but also to interventions for a variety of diseases and
conditions.(* 64) Because HAART keeps people alive and generally in good health, each year
of effective treatment for those with advanced HIV disease (those who would otherwise die)
generally leads to an additional year of life saved. In fact, the cost- effectiveness of
AIDS treatment roughly corresponds to its actual cost. In sub- Saharan Africa, then, where
HIV/ AIDS treatment is predicted to cost approximately $1,123/ patient- year, its cost-
effectiveness ratio, the cost per unit of clinical benefit, will be approximately the
same.

It is important to note that this number is a preliminary estimate, since it is not based
on a detailed African model of HIV disease progression both with and without HAART.
Moreover, it does not incorporate the savings that HAART will permit in regard to hospital
stays and treatment for opportunistic infections, as has been the experience in the United
States, other wealthy countries, and middle- tier developing countries such as Brazil.
(*65 *66) Nor does this cost estimate include HAART's epidemiological benefits, which have
been shown to reduce overall disease incidence both by reducing the HIV viral load and
transmissibility of HIV- positive individuals and by improving the efficacy of prevention
programs (see main text).

Finally, this estimate does not consider the enormous economic and social gains that will
be achieved by saving the lives of parents, and thereby reducing the number of children
that are orphaned by AIDS.

Given the societal- wide ramifications of AIDS discussed in the text, and the ethical and
practical considerations facing the donor world, we believe that expenditures of
approximately $1100 per year of life saved should be fully acceptable to the international
community. We note, in addition, that such expenditure in Africa would also be justified
according to conventional criteria used in the cost- effectiveness literature. According
to theoretical studies, and to the practice in the American public health literature, the
economic value of a life- year saved is commonly estimated to be 2 to 3 times the average
annual U. S. income, and sometimes higher. (*67) On this basis, medical interventions that
save a life- year at a cost of 2 to 3 times the average income (i. e., an intervention
cost of $70,000 to $105,000, given the average U. S. income of $35,000) are often deemed
to be acceptable investments in American public health. Recent studies show that HAART in
the United States has a cost- effectiveness ratio of about $15,000 per year of life saved,
and thus provides excellent value on the cost- effectiveness spectrum. (*68)

Given the lower treatment costs in Africa, HAART in Africa is likely to be about fifteen
times more cost- effective than HAART in the United States, and fifty or more times as
cost- effective as many other routinely accepted medical therapies in the United States.

In the African context, where average annual income is around $500 per year, and even
higher for AIDS patients at the prime of their working lives, a medical intervention of
$1,100 per lifeyear saved would also fall within the conventional bounds of 2 to 3 times
average annual income. This is even more clearly the case in countries with higher per
capita incomes. Finally, this type of intervention will be even more cost- effective when
one considers the decrease in the spread of HIV infection and other social savings that
could be achieved by treating large numbers of patients.

Conclusions

We have outlined the likely cost and cost- effectiveness implications of a major effort to
bring AIDS treatment to Sub- Saharan African countries. In order to provide treatment for
1 million HIV- infected individuals, we estimate costs of about $1.1 billion annually.
This cost may be trebled, to about $3.3 billion, within five years in order to treat 3
million people with AIDS. The cost of a global program that includes not only Africa but
also the low- income and/ or highprevalence countries in other parts of the world would
add approximately 25 percent to this cost, bringing the total donor needs to around $1.4
billion annually during the first three years, and around $4.2 billion annually by the
fifth year. While the cost of these therapies remains far beyond the reach of African and
other poor countries, the modest overall costs to high- income countries with large- scale
treatment and prevention programs, and their potential contribution to prevention of
future HIV transmission should be persuasive to the international community. It is
increasingly clear that immediate, widespread AIDS treatment will be an extremely sound
global investment in the economic, social, and political wellbeing of the world's
resource- poor countries, those that have been hardest hit by the scourge of AIDS.

Figure 1:

[**cannot be seen in this text document**]

Trends in age- adjusted AIDS death rates, 1985- 1999. Shown are annual AIDS deaths  for
sub- Saharan Africa (solid line) and the United States (dashed line). In the U. S., HAART
was introduced in 1995, accounting for the visible decline in deaths. Sub- Saharan Africa,
with apparently more virulent subtypes of HIV and ineffective health systems, has
experienced a constant increase without the diminution in deaths that HAART might allow.
Source: UNAIDS.

REFERENCES
***********
*49 United Nations Conference on Trade and Development. Criteria for Identifying LDCs.
Available at: http:// www. unctad. org/ en/ subsites/ ldcs/ document/ criteria. htm# B.
*50 T. Rosenberg. Look At Brazil. New York Times Magazine (January 28, 2001).
*51 UNAIDS. Report on the global HIV/ AIDS epidemic. Geneva: UNAIDS, 2000.
*52 Among the universe of countries that are either low income (< $755 in 1999) or high
prevalence ( 2% infection rate of adult population) or both, Sub- Saharan African
countries include an estimated 25 million HIV- positive individuals, and the rest of the
world another 5 million (including 3.2 million in India, 1.7 million in other parts of
South and Southeast Asia, and 0.5 million in the Americas).
*53 B. Varghese and T. Peterman. Test and Protect: HIV testing and counseling for HIV
prevention in Africa. Presented at International AIDS Economic Network Symposium, Durban,
South Africa, July 7- 8 2000. Available at http:// www. iaen. org/ conferences/ durbansym/
papers/ 85Varghese. pdf.
*54 E. Marseille et al. (1999). Cost effectiveness of single- dose nevirapine regimen for
mothers and babies to decrease vertical HIV- 1 transmission in sub- Saharan Africa. Lancet
354: 803- 809.
*55 Since individuals with advanced AIDS disease will present themselves for testing at
public clinics and other treatment sites, in order to join in treatment programs, the
proportion of those tested that are HIV- positive will likely be much higher than the
overall prevalence rate.
*56 Voluntary HIV- 1 Counseling and Testing Efficacy Study Group (2000). Efficacy of
voluntary HIV- 1 counseling and testing in individuals and couples in Kenya, Tanzania and
Trinidad: a randomized trial. Lancet 356: 103- 112.
*57 D. Morgan, et al. (2000). Survival by AIDS defining condition in rural Uganda. Sex
Transm Infect 76: 193- 7.
*58 D. Morgan et al. (1997). HIV- 1 disease progression and AIDS- defining disorders in
rural Uganda. Lancet 350: 245- 50.
*59 UNAIDS (2000). AIDS Epidemic Update: December 2000. Available at
http:// www. unaids. org/ wac/ 2000/ wad00/ files/ WAD_ epidemic_ report. PDF.
*60 See Table 4.1 in World Bank (1997). Confronting AIDS (Oxford University Press, New
York).
*61 The high- income countries had a combined GNP of $22.921 billion in 1999, according to
the World Bank Development Report 2000/ 2001.
*62 D. Cyrillo, L. Paulani, B. Aguirre. Direct Costs of AIDS Treatments in Brazil: A
methodological comparison. Presented at International AIDS Economic Network Symposium,
Durban, South Africa, July 7- 8 2000. Available at http:// www. iaen. org/ conferences/
durbansym/ papers/ 13cyrillo. pdf.
*63 World Bank AIDS Campaign Team for Africa (2000). Costs of Scaling HIV Program
Activities to a National Level in Sub- Saharan Africa: Methods and Estimates.
*64 K. Freedberg et al.( 2001). The Cost Effectiveness of Combination Antiretroviral
Therapy for HIV Disease. NEJM 344: 824- 831.
*65. S. Bozzette et. al. (2001). Expenditures for the Care of HIV- Infected Patients in
the Era of Highly Active Antiretroviral Therapy, New England Journal of Medicine 344: 817-
823
*66 "Antiretroviral Therapy: Brazil's Experience," mimeo, Ministry of Health, National
STD/ AIDS Programme, 2000.
*67 See C. Phelps and A. Garber (1997). "Economic Foundations of Cost Effectiveness
Analysis," Journal of Health Economics 16: 1- 31. Their own analysis comes up with a
figure of around two times annual median income as the threshold cut off point (p. 25), a
criterion that varies with the age of the patient. An intervention like HAART that applies
heavily to workers in the prime years of working life would tend to have higher threshold
levels for cost effectiveness. Moreover, these authors cite other works and conventional
criteria that put the threshold at much higher than two times annual income.
*68 K. Freedberg et al.( 2001). The Cost Effectiveness of Combination Antiretroviral
Therapy for HIV Disease. NEJM 344: 824- 831.

--
Send mail for the `ProCAARE' conference to `procaare@usa.healthnet.org'.
Mail administrative requests to `majordomo@usa.healthnet.org'.
For additional assistance, send mail to:  `owner-procaare@usa.healthnet.org'.