1. Scaling up antiretroviral therapy in resource poor countries: The impact of speed on survival
Rochelle P. Walensky, MD, MPH
Associate Professor of Medicine
Harvard Medical School
Divisions of Infectious Disease
Massachusetts General Hospital
Brigham and Women’s Hospital
Supported by NIAID, NIMH, and
Doris Duke Charitable Foundation

2. PEPfAR Reauthorized July 2008
$48 billion
Is this too much? Not enough?
How can we measure the value when dollars are spent?

3. Overview Methods of cost-effectiveness analysis What is the value of :
Antiretroviral therapy (Côte d’Ivoire)?
Second-line therapy (India)?
Earlier treatment initiation (South Africa)?
What is the projected impact of PEPfAR funding over the next 5 years?

4. Cost-effectiveness: Common Misconceptions
“Cost-Effective” = “Cheap”
“Cost-Effective” = “Saves Money”
“Cost-Effective” = Additional benefit worth the additional cost ($/YLS)

5. Cost-effectiveness is about Value for Money
Two different outcome measures
Cost in dollars
Effectiveness: years of life saved (YLS)
quality-adjusted life years (QALYs)
Cost-effectiveness ratio:
Additional Resource Use ($)
Additional Health Benefits (YLS)
To assert that a program is cost-effective, it must be defended by some sort of analysis whereby both costs and effectiveness are measured. Effectiveness is generally measured in terms of years of life saved or quality-adjusted years of life saved. Once you have both of these measures for all programs or treatments to be compared, you create what is called a CE ratio with incremental costs in the numerator and incremental health benefits in the denominator. The next questions are, are resources limited and what are we willing to pay? 9 20

6. Cost-effectiveness of Preventing AIDS Complications (CEPAC)-International
CEPAC is a simulation model of HIV disease and treatment that incorporates CD4 count, HIV RNA, ART, OIs
Clinical data are from Côte d’Ivoire, India, and South Africa
Model provides outcomes measured in:
Mean projected life expectancy
1-, 5-, 10-yr morbidity and mortality
Mean projected per person costs
Designed to address the most critical questions in resource-limited settings
To address these objectives, we used the CEPAC-International model. This is a first-order Monte Carlo state transition model of HIV disease and treatment. The CEPAC International model has been adapted from a US-based model for resource-limited settings where CD4 counts and HIV RNA may not be available, and where, clinical data have been derived from South Africa. Model outcomes are denominated for this analysis in undiscounted mean projected life expectancy
Funded by NIAID/NIMH/DDCF

7. Methods: Cost-effectiveness Thresholds
The Commission on Macroeconomics and Health of the WHO have suggested that interventions are:
Very cost-effective: the CE ratio is <1X Gross Domestic Product (GDP) per capita for that country
$803 for Côte d’Ivoire
$590 for India
$5,300 for South Africa
$43,000 for US
Cost-effective: the CE ratio is
<3 x GDP per capita for that country
Based on the Commission on Macroeconomics and Health, the World Health Organization suggests that interventions are “very cost-effective” if they have cost-effectiveness ratios less than the Gross Domestic Product (GDP) per capita of that country, or $803 for Cote d’Ivoire and “cost-effective” if they have ratios less than 3 times the per capita GDP of that country, or $2,409 for Cote d’Ivoire.
We used these definitions in this analysis.
Macroeconomics and Health: WHO 2001

8. Data from Côte d’Ivoire
Collaborators: Xavier Anglaret, MD, PhD, ANRS 059
Mean age 33 yr, 40% male
Mean CD4 count 331/µl
ART 52-week efficacy/annual cost:
First-line (NNRTI): 51% suppression, $292/yr

9. Model Validation: Data from Côte d’Ivoire
Cumulative Incidence
Goldie et al NEJM 2006

10. Cost-effectiveness of ART: Côte d’Ivoire
Strategy
CD4 Test
ART Start
Criteria
ART Stop
Life Expectancy (mo)
Costs ($)
C-E Ratio ($/yr life saved)
No Treatment --
31.4
780
T/S alone No
32.8
810
240
T/S + ART
2 ODs
1 OD
41.4
1,230
590
50.7
1,720
620
3 ODs
56.8
2,170
890
5 ODs
57.9
2,260
1,060
Yes
CD4<200, CD4<350 and 1 severe OD,
or 1 OD
90%
in CD4 count
69.6
3,420
1,180
GDP: $803
Goldie et al NEJM 2006

11. But what about expensive second-line therapy?

12. Data from India YRG Care, N. Kumarasamy, MD
Mean age: 32.6 yr, 66% male
Mean CD4 count: 318/µl
ART 48-week efficacy/annual cost:
First-line (NNRTI): 55% suppression, $222/yr
Second-line (PI): 65% suppression, $1,435/yr

13. The value of two lines of ART: India
Strategy
Lifetime
cost ($)*
Survival (months)
C-E Ratio ($/YLS)
No treatment
580
35.9 --
One line ART (NNRTI) (<200/µl)
1,540
62.4
450
Two lines ART (NNRTI/PI) (<200/µl)
4,980
84.8
1,850
When long term outcomes are projected, the two ART strategies result in mean life expectancy of years with deferred ART, and years with earlier ART. Per person lifetime costs of deferred ART are $10,820, compared to $11,660 with earlier ART. The incremental cost effectiveness of deferred ART compared to no treatment is $1,110/life year saved and of earlier ART compared to deferred ART is $1200/life years saved.
GDP: $590
*All costs in 2005 US$, converted using GDP deflators
& the mean exchange rate between the Rupee & US$
Freedberg et al., AIDS 2007, World Bank

14. When to Start?
The question of “when to start” ART has drawn interest globally
In resource-limited settings, this question is crucial:
Limited available ART regimens
Increased incidence of opportunistic infections (OIs) and tuberculosis (TB) at higher CD4 counts
A clinical trial may be well-suited to address the “when to start” question
Results from such a trial will not be available for at least 5 years; two trials are currently enrolling
Good morning. The question of “when to start” antiretroviral therapy, or ART, in HIV-infected patients has drawn interest globally. Especially in resource-limited setting, this question must be addressed in the context of limited available ART regimens, in addition to an increased incidence of opportunistic infections and tuberculosis at comparatively higher CD4 cell counts.

15. Data from South Africa CTAC/Gugulethu Cohorts Mean age 33 yr, 55% male
Robin Wood, FCP, MMed, DTM&H, U CapeTown
Mean age 33 yr, 55% male
Mean CD4 count 375/µl
ART 48-week efficacy/annual cost:
First-line (NNRTI): 84% suppression, $288/yr
Second-line (PI): 71% suppression, $564/yr

16. “When to Start” ART: South Africa
Strategy
Lifetimecost ($)*
Survival (years)
C-E Ratio ($/YLS)
No treatment
3,620
3.83 --
Deferred ART (<250/µl or OI)
10,820
10.33
1,110
Earlier ART (<350/µl or OI)
11,660
11.03
1,200
When long term outcomes are projected, the two ART strategies result in mean life expectancy of years with deferred ART, and years with earlier ART. Per person lifetime costs of deferred ART are $10,820, compared to $11,660 with earlier ART. The incremental cost effectiveness of deferred ART compared to no treatment is $1,110/life year saved and of earlier ART compared to deferred ART is $1200/life years saved.
GDP: $5,300
*All costs in 2006 US$, converted using GDP
deflators & the mean exchange rate between the
SA Rand & US$
Walensky et al CROI 2008 [abstract]

17. Results: Proportion Alive at 5 Years
Earlier ART
Deferred ART
No ART
Model-generated survival curves also diverge for the three ART strategies. No treatment, in blue, has 50% survival at 4.2 years. By year 3, earlier ART maintains a consistent 5% absolute increase in the proportion alive.
Walensky et al CROI 2008 [abstract]

18. ART Roll Out: The impact of speed on survival
Alternative ART rollout scenarios in South Africa OR what might more PEPFAR funds mean?
To forecast
Number of lives lost awaiting therapy
Number of patients in treatment
To project when total ART need would be met
To inform decisions regarding the life-saving value of alternative treatment expansion scenarios
Our objectives were to examine alternative ART rollout scenarios in South Africa to:
First, forecast the number of lives lost while awaiting therapy among treatment-eligible patients, and the number of patients in treatment over the next several years;
Second, to project when total ART need would be met;
And third, to inform decisions regarding the life-saving value of alternative treatment expansion scenarios.

19. 4 Growth Scenarios New ART Slots by 2012 Source Zero growth 100,000
New ART Slots by 2012
Source
Zero growth
100,000
---
Constant growth
600,000
ASSA
Moderate growth
2,100,000
SA Joint Task Team
Rapid growth
2,400,000
We examine four different growth scenarios for ART rollout. These scenarios are similar to proposals made by the Actuarial Society of South Africa, or ASSA, and the South African Joint Task Team on HIV/AIDS. They range from a zero-growth scenario, which provides 750,000 new treatment slots by 2010, to a rapid growth scenario, which provides 2.8 million treatment slots by 2010.
Walensky et al., JID 2008

20. Percent ART Need Met by Year
Rapid
Moderate
Constant
% ART Need Met
Zero
This results slide shows the percent ART need met by year. On the horizontal axis is the year, and on the vertical axis, the percent of ART need met. All scenarios begin with the same number of patients treated in 2007, and hence the same percent of coverage; they then diverge in subsequent years. Treatment coverage under the zero growth scenario, shown in white, decreases steadily through 2012 at which time 28% of people in need of therapy will receive it. In the constant growth scenario, shown in yellow, treatment rates stay relatively unchanged to achieve 52% of necessary coverage by In the moderate growth scenarios, shown in orange, treatment coverage increases steadily to reach 97% of those in need by The rapid growth scenario, shown in purple, achieves full treatment coverage by 2011.
Year
Walensky et al., JID 2008

21. Projected Deaths and Patients Alive on ART: 2007-2012
AIDS Deaths
Alive on ART
Zero growth
2,465,000
1,290,000
Constant growth
2,160,000
1,595,000
Moderate growth
1,449,000
2,306,000
Rapid growth
1,232,000
2,523,000
This table shows the number of AIDS deaths and patients alive on ART for 2005 to 2010, as we total the deaths for each year from the previous slide. The number of AIDS deaths ranges from 2,197,000 in the zero growth scenario, down to 1,161,000 in the rapid growth scenario. The number of patients alive and receiving therapy ranges from 693,000 in the zero growth scenario to 2,572,000 in the rapid growth scenario.
Walensky et al., JID 2008

22. Impact of No CD4 Monitoring on Deaths
Decisions on treatment initiation and switching based on clinical criteria
CD4
Monitoring
No CD4 Monitoring
Increased Deaths
Zero growth
2,465,000
2,568,000
103,000
Constant growth
2,160,000
2,436,000
276,000
Moderate growth
1,449,000
2,237,000
788,000
Rapid growth
1,232,000
2,218,000
986,000
This shows the impact of the availability of CD4 monitoring on the number of deaths. In the absence of CD4 monitoring, decisions on treatment initiation and switching are based solely on clinical criteria. The number of AIDS deaths with CD4 monitoring available from 2007 to 2012 is the same as on the previous slide, shown here in the left column. The number of deaths with no CD4 monitoring available is substantially higher in every scenario. In the zero growth scenario, deaths increase from 2,465,000 with CD4 monitoring to 2,568,000 with no CD4 monitoring. This effect is even more pronounced in the rapid growth scenario, where deaths increase from 1,232,000 with CD4 monitoring to 2,218,000 with no CD4 monitoring. Compared to corresponding scenarios where CD4 monitoring is available, the lack of CD4 monitoring results in 103,000 to 986,000 increased deaths.
Walensky et al JID 2008

23. Timeline of Major HIV Interventions
This figure depicts the timeline of major HIV interventions we considered and when they became standard of care in the US. We include prophylaxis for Pneumocystis jiroveci pneumonia (or PCP), starting in 1989 and add Mycobacterium avium complex, or MAC, prophylaxis starting in Prevention of mother to child transmission with AZT began in marks the beginning of highly active antiretroviral therapy. We characterize the improvements in ART efficacy over time by dividing ART into 4 eras, beginning in 1996, 1998, 2000 and 2003.
Walensky et al., JID 2006

24. AIDS Survival by Era Walensky et al., JID 2006
This figure shows model-based survival curves for patients diagnosed in the first year of each of the six treatment eras (1989, 1993, 1996, 1998, 2000, 2003) and untreated disease, shown in gray. The area between the curves illustrates the improvement in AIDS-associated survival in the US over time. Median survival was 1.7 years for the PCP era shown in yellow, 7.5 years for ART 1 shown in light blue, and 14.1 years for ART 4, the upper most curve shown in pink.
Walensky et al., JID 2006

25. Future Directions What are the next most important questions?
The clinical and economic value of CD4 and/or HIV RNA laboratory monitoring?
The cost-effectiveness of genotype testing?
The clinical and economic impact of routine HIV screening?
Which of these results are country-specific? Which are generalizable around the globe?
In conclusion, we find that earlier initiation of ART in South Africa improves both short-term and long-term survival, and is cost-effective by international standards. The addition of TLTBI to both deferred and earlier ART strategies leads to a small survival benefit, but is cost-saving. While awaiting the results of clinical trials on the “when to start” question, the results of this study indicate that earlier ART initiation with the addition of TLTBI should be the treatment goal in South Africa.

26. Conclusions
In resource-limited settings, quantitative assessments of value for money are critical, given multiple treatment strategies.
Cost-effectiveness analyses have demonstrated the value of 1st and 2nd-line therapy, of earlier ART initiation, and of the need for increasingly rapid ART roll-out in international settings.
HIV simulation modeling is a powerful tool to inform health policy and to understand survival, costs and cost-effectiveness of alternative clinical interventions.
In conclusion, we find that earlier initiation of ART in South Africa improves both short-term and long-term survival, and is cost-effective by international standards. The addition of TLTBI to both deferred and earlier ART strategies leads to a small survival benefit, but is cost-saving. While awaiting the results of clinical trials on the “when to start” question, the results of this study indicate that earlier ART initiation with the addition of TLTBI should be the treatment goal in South Africa.

27. CEPAC (and other) Investigators: US
Harvard Medical School
Ingrid Bassett, MD, MPH
Melissa Bender, MD, MPH
Sarah Chung
John Chiosi
Andrea Ciaranello, MD
Kenneth Freedberg, MD, MSc
Louise Ivers, MD
Elena Losina, PhD
Ben Linas, MD, MPH
Zhigang Lu, MD
Brandon Morris
Paul Sax, MD
Caroline Sloan
Heather Smith
Lauren Uhler
Rochelle Walensky, MD, MPH
Bingxia Wang, PhD
Harvard SPH
Sue Goldie, MD, MPH
Kara Cotich
Callie Scott
George Seage, III DSc, MPH
Milton Weinstein, PhD
April Kimmel, MSc
Cornell
Bruce Schackman, PhD, MBA
Yale
David Paltiel, PhD
Lille, France
Yazdan Yazdanpanah, MD, PhD
And also, but a terrific modeling group, we fondly call the cost-effectiveness of preventing AIDS complications or CEPAC investigators. This group is led by Kenneth Freedberg and has the generous support of the CDC, NIMH, and NIAID. Thank you.

28. CEPAC Investigators: International
Côte d’Ivoire
Xavier Anglaret, MD, PhD
Eugene Messou, MD
Catherine Seyler, MD, PhD
Siaka Touré, MD, MPH
OECS
Kathleen Allen-Ferdinand, MD
Paul Ricketts, MD
Hazel Williams-Roberts, MD
South Africa
Neil Martinson, MBBCh, MPH
James McIntyre, MBChB,MRCOG
Lerato Mohapi, MBBCH
Robin Wood, MD
India
N. Kumarasamy, MBBS, PhD
Tim Flanigan, MD
Kenneth Mayer, MD