Transcript Cost

HIV Monitoring Strategies
in Resource-Limited
Settings
Eran Bendavid, MD
Center for Health Policy
Division of Infectious Diseases
Stanford University, Stanford CA
November 6, 2007
Context
Although the number of infected people
receiving highly active anti-retroviral
therapy (HAART) for HIV in low- and
middle- income countries increased
dramatically, disease monitoring for the
infected population is not well defined.
Purpose
Analyze the relative costs and
effectiveness of monitoring patients with
symptoms alone, with CD4 counts, or with
CD4 counts and viral load measurements
using data from South Africa.
The Epidemic in SSA
HIV epidemic in sub-Saharan Africa, 1985-2005
Number of people
living with HIV (millions)
% HIV prevalence,
adult (15-49)
30
15.0
25
12.5
20
10.0
15
7.5
10
5.0
5
2.5
0
0.0
1985
1990
Number of people living with HIV
% HIV prevalence, adult (15-49)
1995
2000
2005
Funding for Global HIV
9000
Signing of
Declaration of
Commitment on
HIV/AIDS
8000
7000
8297
US$ million
6000
5000
4000
3000
2000
1623
1000
292
0
1996 1997 1998 1999 2000
2001 2002
2003
2004 2005
HIV Treatment Globally
Effect of HAART
Reduce viral load
Raise CD4 count
Prevent opportunistic
diseases
HIV Management in Africa
• Monitoring used to determine when to
start HAART, when to switch from first
line regimen to second line regimen, and
when to stop HAART.
• Patients can be monitored with
symptoms alone, with symptoms and
CD4, or with symptoms, CD4, and viral
load measurement, similar to the
standard of care in the US.
Examples
No CD4
Test
Presentation
Time (No Scale)
Develop an OD,
then start HAART
Die of HIV or
other causes
Start HAART with development of ODs
W/ CD4
Test
Presentation
Time (No Scale)
CD4 below 350
Start HAART
Develop an OD
Die of HIV or
other causes
350 cells/μl
CD4
Start HAART when CD4 falls below 350
or with development of an OD
Model Inputs
• Health status determined by…
–
–
–
–
–
–
Age, gender
Current, lowest, and highest measured CD4
Current and steady state viral load
Medication toxicity
Treatment failure
History of opportunistic diseases
• Clinical data obtained from Cape Town
AIDS Cohort and Khayelitsha MSF Clinic
Toxicity, adt’l ODs
Failure (CD4 or VL)
Develop an OD
CD4 criteria
Y
Criteria to
start HAART
First line
HAART
Criteria to
Y
change HAART
Second line
HAART
N
Y
N
HIV +
Clinic visit
No HAART
Y
N
One month
Progression
N
Criteria to
stop HAART
Cost-Effectiveness
Effectiveness
E
B
D
A
C
Costs
Value measured in incremental cost-effectiveness ratio:
ICER =
CostX - CostA
EffectivenessX - EffectivenessA
Cost-Effectiveness
F
5
E
4
Effectiveness
D
ICERD-A =
ICERE-A =
2
A
200
200-0
2-0
200-0
4-0
350
Costs
= 100
= 50
ICERF-E =
350-200
5-4
= 150
Cost-Effectiveness
Effectiveness
E
Cost-saving
F
D
A
Bad idea
Costs
Cost-effectiveness threshold: the amount
of money that we are willing to spend to
gain one unit of effectiveness
Results
Life Expectancy (mo)
80
76
72
68
64
60
56
3400
Symptombased
3800
4200
4600
Cost (2007 USD)
5000
5400
5800
Results
Life Expectancy (mo)
80
76
CD4 only,
HAART @ 350/μl
CD4 & viral load,
HAART @ 350/μl
CD4 & viral load
every 3 mo
72
68
64
CD4 only,
HAART @ 200/μl
60
56
3400
Symptombased
3800
4200
4600
Cost (2007 USD)
5000
5400
5800
Results
Life Expectancy (mo)
80
76
CD4 & VL: 6 mo
HAART: 350
CD4: 6 mo
HAART: 350
CD4: 3 mo
HAART: 350
72
68
64
CD4 & VL: 3 mo
HAART: 350
CD4 & VL: 3 mo
HAART: 200
CD4 & VL: 6 mo
HAART: 200
CD4: 3 mo
HAART: 200
CD4 only,
HAART @ 200/μl
60
56
3400
Symptombased
3800
4200
4600
Cost (2007 USD)
5000
5400
5800
Results
Life Expectancy (mo)
80
76
CD4 only,
$134/life-year
gained
HAART @ 350/μl
CD4 & viral load,
$5,188/life-year
HAART @ 350/μlgained
CD4 & viral load
$125,000/life-year
every 3 mo
gained
72
68
64
CD4
only,
Cost-Saving
HAART @ 200/μl
60
56
3400
Symptombased
3800
4200
4600
Cost (2007 USD)
5000
5400
5800
Sensitivity Analysis
Incremental Cost of CD4
Monitoring (2007 USD)
1000
800
600
400
ICER = $1,250/life-year gained
Malawi
Lesotho
200
Zim
0
-200
-400
-600
50
100
150
Swaziland
200
Botswana
S. Africa
Cost-saving nearly $600 per patient
-800
Cost of Inpatient Day
Conclusions
• Compared to symptom-based
management, CD4 monitoring could
substantially increase length of life in
many parts of southern Africa.
• This increase in life expectancy may save
health care costs by reducing number of
opportunistic diseases and inpatient
days.
Conclusions
• Cost savings may be realized in countries
where hospital costs are relatively high
(South Africa, Botswana, Namibia,
Swaziland).
• The World Health Organization considers
an intervention cost-effective if the
incremental cost-effectiveness ratio is
around twice the per-capita GDP.
Conclusions
• By that standard, monitoring CD4 even in
places where hospital costs are very low
(e.g. Malawi) is a cost-effective
intervention.
• Starting HAART at CD4 350 cells/μl was
substantially more effective than starting
at CD4 200 cells/μl at a small additional
cost.
Conclusions
• As access to treatment improves in
resource-limited settings, the large
investments in HAART could be greatly
leveraged by monitoring CD4 and
initiating treatment before development
of symptoms.
• Technological challenges for monitoring
CD4 are fewer today with low-cost
diagnostic equipment.
Conclusions
• Providing access to CD4 monitoring for
one million people and starting HAART at
a CD4 of 200 cells/μl could provide
683,000 years of life more than providing
HAART without CD4 monitoring.
• Starting HAART at 350 cells/μl could
provide an additional 450,000 years of
life.
Limitations
• Most data was taken from Cape Town
region
– Burden of disease
– Availability of health care
• Did not account for indirect costs such as
lost wages and travel time.
Acknowledgments
• Stanford
– Center for Health Policy: Doug Owens, Alan Garber
– Infectious Diseases: David Katzenstein, Robert Shafer,
Dennis Israelski
– Also: Elisa Long, Margaret Brandeau, Sean Young,
Gillian Sanders, Ahmed Bayoumi
• Funding Sources
– AHRQ
– NIDA (Peter Hartsock)
• University of Cape Town: Motasim Badri, Robin
Wood, Linda-Gail Bekker, Gary Maartens