Slides from the HIV Prevention Resource Allocation Model session at the 2009 National HIV Prevention Conference in Atlanta. Primary presenter: Arielle Lasry, Division of HIV/AIDS Prevention, CDC
Andheri East ) Call Girls in Mumbai Phone No 9004268417 Elite Escort Service ...
Res Allocation Model Nhpc09 Lasry
1. A model for allocating
HIV prevention resources in the
United States
Arielle Lasry1, Stephanie Sansom1, Katherine Hicks2,
Vladislav Uzunangelov2
1 Division of HIV/AIDS Prevention, National Center for HIV, Viral Hepatitis,
STD, and TB Prevention, Centers for Disease Control and Prevention, Atlanta, Georgia (GA)
2 RTI International, Research Triangle Park, North Carolina (NC)
National HIV Prevention Conference
Atlanta, August 26, 2009
Disclaimer: The findings and conclusions in this study are those of the authors and do not
necessarily represent the views of the Centers for Disease Control and Prevention.
Presentation outline
Background
How the model works
Model output
Summary, limitations & next steps
2. Presentation outline
Background
How the model works
Model output
Summary, limitations & next steps
Background
Generally, healthcare resource allocation is a
process used to determine how to distribute
resources among programs, populations or regions
from a limited budget.
The way health funds are allocated has an important
influence on health outcomes.
3. Background
CDC’s Division of HIV/AIDS Prevention (DHAP) has total
budget of approximately $650 Million.
approximately $325 Million funds health departments
and community based organizations for core HIV testing
and prevention programs domestically.
We continue to face considerable challenges.
The overall number of new HIV infections per year has
not declined for more than a decade.
HIV resources are not unlimited.
The resource allocation model evaluates how to allocate
HIV prevention funds to further reduce new HIV infections
given a budget of $325 Million.
Modeling vs. the real world
Models are a convenient representation of the real world.
Models can help us project epidemic outcomes, better
understand causal relationships and identify areas where
prevention programs can have the most impact.
Translation of model outcomes into the real word is difficult
because models are a simplified representation of a complex
reality.
Some simplifying assumptions of the resource allocation model:
Population subgroups are reachable and can be perfectly
targeted.
All other funding, including that from state and local
government and the private sector, remains constant.
Administrative costs of disbursing funds at multiple levels
not considered.
4. Presentation outline
Background
How the model works
Model output
Summary, limitations & next steps
Resource allocation model
Uses the best data and estimates available on HIV
incidence, prevalence, prevention program costs and
benefits, current spending, etc.
Projects HIV infections for the United States as a whole
given different allocation strategies.
Based on the best currently available data, suggests
hypothetical allocation to minimize incidence.
Provides information that could be considered in future
decision-making processes for resource allocation.
One of many inputs and information sources – none
should be used alone.
Not intended to replace local decision-making.
5. Populations considered
Three transmission related risk groups
High-risk Men who have Injection Drug
heterosexuals sex w/men Users
(HRH) (MSM) (IDU)
Three race/ethnicity categories
Black Hispanics Other races*
* Mainly whites, + A/PI, AI/AN
Two gender categories (M/F)
We end up with 15 risk populations (2X3X3 - 3)
Infection transmission
Each of the 15 risk populations is modeled as 3
compartments.
ti x e ti x e ti x e
HIV+ HIV+
Susceptibles undiagnosed diagnosed
yrt n e
Infection from contact with HIV+
diagnosed or undiagnosed
Screening & diagnosis
The 15 risk populations interact (mix) with one another
thereby generating new infections.
6. Intervention types
Behavioral
Testing
interventions
Intervention to reduce Targeted testing to
Risk risk among susceptibles identify positives
populations and the infected unaware of their status
Testing in general
healthcare settings to
General e.g. Social marketing identify positives
population unaware
of their status
Compares the outcome of these interventions in terms of estimated
HIV infections prevented when targeted to the general population and
to risk populations defined by race/ethnicity, gender, and risk group.
How the model works
1. Epidemic model:
Simulates the epidemic outcome
given a defined allocation.
Defined as dynamic
compartmental model and written
S− U+ D+
out as a system of difference
equations
noitacollA weN
oiranecs snoitcefni
2. Optimization engine:
Generates different allocation
scenarios, which feed into the
epidemic model and stops when
best outcome is reached.
Yes No
Aims to minimize the total number Improve? Stop
of new infections over 5 years, by
deciding how much to allocate to
the interventions considered.
7. Summary of data used
1. Population data 4. Intervention costs and
Total size of risk population outcomes
Number of positives Cost of testing by target group
% unaware Level of background testing
Cost of behavioral interventions
2. Rates of movement in and
by target group
out of each risk population Effect and duration of behavioral
Entry into susceptible intervention by target group
Exit rate from susceptible and 5. Constraints
undiagnosed+
Maximum reachability (%) by
Exit rate from diagnosed + intervention category by risk
(death and disease)
population
3. Transmission (optional) Minimum or Maximum
Mixing % investment by intervention,
Incidence by subpopulation target group and/or risk
Effective contact rate for population
diagnosed and undiagnosed Budget
Validation and quality assurance
Validation of input data
Internal vetting and sign-off by subject matter experts
within DHAP.
External review committees provided written reviews
and participated in a series of conference calls. Their
feedback was incorporated into our data estimates.
Validation of model structure
Modeling experts (outside CDC) provided written
review of model and participated in conference call.
Comments were used to update the model.
Quality assurance
Several measures taken including sensitivity analysis.
Model demonstrated stability and robustness.
8. Presentation outline
Background
How the model works
Model output
Summary, limitations & next steps
Allocations by intervention type
$350
$300
Testing Testing
(Risk pop) (Risk pop)
$250
Testing
$200
(Gen pop)
$150
Behavioral
Behavioral Intervention
$100
intervention (Risk pop)
(Risk pop)
$50
Behavioral intvn. (Gen pop)
$-
Baseline Model
9. Allocations to behavioral
interventions by serostatus
$250
$200
Untargeted
$150
HIV+ Diagnosed
HIV+
$100 Diagnosed
Susceptibles
& HIV+
$50
Undiagnosed
Susceptibles & HIV+ Undiagnosed
$-
Baseline Model
Allocations to behavioral
interventions by race/ethnicity
$250
$200
Others
Untargeted
$150
Others
$100 Hispanic
Hispanic
$50
Black Black
$-
Baseline Model
*Others: Whites, APIs, American Indians and Alaska Natives
10. Allocations to behavioral
interventions by risk group
$250
$200 HRH
Untargeted IDU
$150
HRH
$100
IDU MSM
$50
MSM
$-
Baseline Model
Allocations to testing
by race/ethnicity
$180
$160
$140
$120 Untargeted
$100
$80 Others
Others
$60
Hispanic
$40 Hispanic
$20 Black Black
$-
Baseline Model
*Others: Whites, APIs, American Indians and Alaska Natives
11. Allocations to testing
by risk group
$180
$160
$140
$120 Untargeted
$100
HRH
$80 IDU
$60 HRH
$40
MSM
$20
IDU
MSM
$-
Baseline Model
Presentation outline
Background
How the model works
Model output
Summary, limitations & next steps
12. Select model output
Directs resources for testing and behavioral
interventions to those at greatest risk (not general
population).
Increases allocation to behavioral interventions for
diagnosed positives.
Increases allocation to testing for MSMs and IDUs.
More than doubles total allocation to MSMs.
More than doubles total allocation to IDUs.
Increases allocation to behavioral interventions for
Blacks.
Limitations
Budget only includes DHAP extramural funds for testing
and behavioral programs, not all HIV prevention funds.
Accounts for current levels of non-CDC funded
screening and behavioral intervention efforts.
Assumes non-CDC funding levels are constant.
Data
Data are often uncertain.
Data updates required as new evidence emerges.
Assumes that resources can be “perfectly” targeted.
Considers prevention strategies that are currently
federally funded (i.e. no needle exchange or biomedical
strategies).
Does not account for regional/geographical differences.
13. Next steps
Continuous model refinements
Data updates
Broaden scope of interventions
Explore how model could be adapted for regional/local
planning uses.
Consider how the model might be integrated into
DHAP’s priority setting process.
Resource allocation model - Technical briefing
September 14th, 2009 from 1:00-2:00PM ET
Resource allocation model - Program briefing
September 15th, 2009 from 1:30-2:30PM ET
Thank you
Questions?