Generative AI in Health Care a scoping review and a persoanl experience.
World AIDS Day 2016: Economic evaluation for HIV in South Africa
1. Using Economic Evaluation to Strengthen
South Africa’s National HIV Program:
Boston University’s Experience
Sydney Rosen
Department of Global Health
Boston University School of Public Health
Health Economics and Epidemiology Research Office,
University of the Witwatersrand
December 1, 2016
2. What To Expect Today
• Ten minutes about economic evaluation
• The setting: a safari to South Africa
• How much does it cost?
• Is it cost-effective?
• What are the benefits?
• Can we afford to do what we want?
• Final thoughts (if any are left)
With thanks to all the colleagues whose work I’ve borrowed…
4. Economics is about the allocation of
scarce resources to meet competing ends
• What resources do we have?
• What are our goals (ends)?
• Who decides?
• Who wins?
• Who loses?
• Economics is about value (and values), not money. Money is
merely a unit of measurement.
5. Economic evaluation is about how to
allocate resources under constraint
• Allocating resources: Deploying our constrained resources to
achieve our goals over space and time
• Minimizing costs: Choosing strategies to spend the least we can to
generate the outcomes or impacts we want
• Maximizing benefits: Getting the most health and other desirable
outcomes from the resources we have
• Various other things that we won’t talk about today
6. Within the world of evaluation…
Can it work?
(Efficacy)
Does it work?
(Effectiveness)
Is it feasible to
implement in the
setting we care
about? (Feasibility)
How much does it
cost, and can we
afford it?
Is it cost-effective?
Do the benefits
exceed the costs?
Source: Drummond et al, Methods for the Economic Evaluation of Health Care Programmes, 2005
7. Some types of economic evaluation
• Value of changes in resources used
• Can be incremental or total costCost
• Cost per successful health outcome achieved
• Outcomes are natural (meaningful) units of healthCost/outcome
• Difference in cost per successful outcome achieved
• Comparison between options with same outcomeCost/effectiveness
• Cost per unit of utility gained or disutility avoided
• DALYs or QALYs can combine multiple or disparate outcomesCost/utility
• Ratio of value of total costs to value of total benefits
• Costs and benefits are valued in monetary units and presented
as net benefits (= benefits-costs)
Cost/benefit
8. More about types of evaluation
Approach Question Definition and uses
Incremental
cost
How much more does it cost
than what I’m already
spending?
Additional cost of adding an additional service, per patient
served; budgeting to scale up additional service
Total cost How much should I budget? Total cost per patient to achieve the outcomes being
evaluated; budgeting to provide service to a specified
number of patients
Cost/outcome How much does it cost to get
the outcomes I want?
Average cost per successful or unsuccessful outcome
achieved; monitoring changes and identifying
opportunities for better resource allocation
Cost-
effectiveness
How does my intervention
compare to standard care or
other interventions with the
same outcome?
Difference in costs divided by difference in specific health
outcomes; comparing different service delivery strategies
for achieving a single outcome
Cost-utility How does my intervention
compare to other health
interventions with different or
multiple outcomes?
Difference in costs divided by difference in utility
outcomes (QALYs, DALYs); determining whether an
intervention is a “good buy” compared to dissimilar
interventions or a threshold
Cost-benefit Do the benefits of my
intervention exceed the costs?
Ratio of costs to benefits, in monetary terms; deciding
whether the service should be provided at all
9. 9
Cost-effectiveness is not all it’s cracked up to be
Cost-effectiveness analysis is a comparison between options
• A new intervention compared to standard of care
• One intervention compared to another intervention
Each intervention has a cost, expressed in monetary terms
Each intervention has an effect, expressed in health or utility terms
• For example, proportion of patients achieving viral suppression, or disability
adjusted life years (DALYs)
Cost-effectiveness does not tell us if something is “worth it”. Let’s see what it
does tell us.
10. When is cost-effectiveness analysis useful?
Effectiveness
Cost
Cost of status quo Effectivenessofstatusquo
Higher or equal cost, lower
effectiveness—almost never do it
Lower cost, lower effectiveness—
only do it to reduce budget
Higher cost, higher effectiveness—
do it if it’s affordable and cost-
effective compared to alternatives
Lower cost, higher or equal
effectiveness—almost always do it
✔
?
?
• Minimize cost to achieve a predetermined target—for example, 90-90-90
• Maximize benefits for a predetermined cost—for example, a national budget
• Compare two approaches to achieving the same health outcomes—for
example, a new drug or device compared to an existing one
--BUSPH Professor Bruce Larson
11. Last and least: What exactly is an ICER?
An ICER is an Incremental Cost Effectiveness Ratio
ICER =
Difference in costs
Difference in effects
Standard care
Cost = $8,000
Effects = 10
units of health
New
intervention
Cost = $12,000
Effects = 14
units of health
Difference in
costs = $4000
Difference in
effects = 4 units
ICER = $1000
Is an ICER of $1000 worth it?
It depends on two main things:
What is the value of a unit of health?
What is the alternative use of the resources?
Upshot: Context matters. Whether something is cost-
effective or not depends on the comparison in question and
can vary widely (if not wildly) by place, time, and situation. A
thing cannot be “cost-effective” in the abstract.
12. What To Expect Today
• Ten minutes about economic evaluation
• The setting: a safari to South Africa
• How much does it cost?
• Is it cost-effective?
• What are the benefits?
• Can we afford to do what we want?
• Final thoughts (if any are left)
14. HIV prevalence 2014
The problem
Sub-Saharan Africa has the most severe HIV/AIDS epidemic in the world
and southern Africa the most severe in Africa.
HIV prevalence adults 15-49 in
2013 (UNAIDS)
In South Africa:
• Adult HIV prevalence is estimated at 19%:
- 7 million HIV-positive people
- 19% of global burden (0.7% of global
population)
- Nearly 300,000 new infections/year
• HIV and tuberculosis are the two leading causes
of adult deaths; 70% of TB patients have HIV co-
infection
• Life expectancy at birth fell from 62 in 1995 to 52
in 2010
Sources: Stats SA, National Department of Health, UN Population Division
15. The solution
• An estimated 3.4 million
patients are on ART (just
under half the HIV-positive
population)
• Among those on ART, 78%
are virally suppressed
• Mortality has declined and
life expectancy is rising
• Adult incidence is declining
and vertical transmission is
< 2%
• Large-scale, public sector provision of ART began in 2004
16. The need for efficient resource allocation
• In September 2016, South Africa adopted “treat all” guidelines; all HIV-positive
individuals are eligible for ART
• For the country to reach its first two “90-90-90” targets (90% diagnosed, 90% on
treatment), > 2.4 million more people must be placed on ART
• The estimated additional cost of this is about $1 billion/year, or about a 60%
increase over current expenditure on HIV (SA funds 85% of its own HIV
program; poorer countries fund very little)
• The South African economy is weak, with low growth and unemployment at
36%, and donor funding is unlikely to increase
• Economic evaluation (combined with some clever modeling) is helping South
Africa decide how to allocate its constrained resources as efficiently as possible
to maximize health benefits
Source: Stats SA; personal communication, Gesine Meyer-Rath
17. • The Department of Global Health at BUSPH began collaborating with the
University of the Witwatersrand (Wits University) in Johannesburg in 2004.
• We helped to found HE2RO, the Health Economics and Epidemiology
Research Office, in partnership with Wits.
• In collaboration with HE2RO, we’ve been conducting applied research on the
economics and epidemiology of the HIV epidemic for > 10 years.
• We work closely with the South African Government, particularly the National
Department of Health, to inform its policies.
• HE2RO now has ≈ 60 investigators and staff and dozens of studies.
• Funders include USAID, the NIH, the Gates Foundation, the CDC, the World
Bank, and others.
BUSPH’s partnership in South Africa
18. What To Expect Today
• Ten minutes about economic evaluation
• The setting: a safari to South Africa
• How much does it cost?
• Is it cost-effective?
• What are the benefits?
• Can we afford to do what we want?
• Final thoughts (if any are left)
20. • Cost and price are not the same: I produce
something at its cost; I sell it at its price
• Costs are independent of source of funding
• We typically discount and inflate future costs
(and benefits)
Cost = monetary value
of resources utilized to
provide a service or
achieve an outcome
• NO: “HIV treatment costs $10,000.”
• YES: “In 2016, first-line HIV treatment per
patient per 12-month period in the United
States, delivered on an outpatient basis and
including medications, laboratory tests, clinic
visits, patient management, and overhead
for these services cost, Medicaid average of
$10,000 (USD 2016).”
Cost to whom, when,
for what, for how long?
How much does it cost is a loaded question
21. How much does HIV treatment cost in South Africa?
$-
$200
$400
$600
$800
$1,000
$1,200
$1,400
All outcomes In care and
responding
In care but not
responding
No longer in care
Costperpatient(2006USD)
Site 1 (hospital based HIV clinic) Site 2 (NGO clinic)
Site 3 (NGO clinic) Site 4 (private GPs)
• Study conducted in 2005-
2006
• Retrospective cohort with
chart review and unit cost
data collection
• Primary outcome = adult
patient “in care and
responding” 12 months
after ART initiation
• Perspective = provider
(Department of Health)
• Included all resources used
to provide outpatient ART
• First empirical data on
actual cost of providing ART
in Africa Today, first-line ART in South Africa costs about
$200/patient/year. In the U.S. it is $10,000-$20,000.
22. What To Expect Today
• Ten minutes about economic evaluation
• The setting: a safari to South Africa
• How much does it cost?
• Is it cost-effective?
• What are the benefits?
• Can we afford to do what we want?
• Final thoughts (if any are left)
24. Cost-effectiveness of task-shifting and decentralization
• Two strategies for maintaining
stable patients on ART:
- Centralized, full-service HIV
clinics
- Primary health clinics
• Why use cost-effectiveness
analysis?
- Minimize cost to achieve a pre-
determined outcome (viral
suppression) for maximum
number of patients
• Evaluation of a pilot
implementation project
following an RCT
- RCT controlled for everything
- Pilot was somewhere between
RCT and routine care
Centralized PHC
Effects (% of cohort in
care and responding at
12 months)
90% 96%
Costs (average
cost/patient for 12
months)
$539 $486
Most improvements in health are not cost-saving. In general, we don’t get
more for less. We get more for more, and less for less.
25. An example of getting more for more—the RapIT study
• Randomized trial of same-day
ART initiation for adult patients
- Used point-of-care tests and
accelerated procedures
- Patients could be dispensed ARVs
on the day of their first clinic visit,
compared to 4-6 visits required
under standard care
- Outcomes = ART initiation < 90
days and viral suppression < 10
months
• Intervention was effective,
increasing
- ART initiation by 36%
- Viral suppression by 26%
• Intervention cost more. Cost per
patient achieving primary
outcome=
- $524 in rapid arm
- $483 in standard arm
Is $780 (base case) or $220 (best case) per additional patient virally
suppressed cost-effective? It depends on what other ways there are to
achieve it…and on the value of the outcome.
26. A useful application of cost-effectiveness analysis: screening
costs versus treatment savings
• South Africa endorsed two
strategies for identifying
cryptococcal meningitis in HIV
patients
- Reflexive CrAg screening for all
patients with CD4 < 100
- Provider-initiated CrAg screening
based on clinician’s decision
- Which is a better strategy?
• Cost-effectiveness model
developed using programmatic
data from pilot screening
programs in South Africa
- Reflexive strategy was more
effective (saved more lives) and
- Costs for treatment were lower
under reflexive approach, but
- Costs for screening were higher
under reflexive strategy
Per 100,000 CD4 counts Reflexive
screening
Provider
initiated
screening
Difference
(reflexive -
provider
initiated)
Number screened 9,500 1,536 7,964
Cost of screening $47,044 $9,508 $37,536
Cost of treatment $209,399 $264,564 -$55,165
Total cost $256,443 $274,072 -$17,629
Number of additional surviving patients 148 118 30
Total years of life saved 3,189 2,542 647
Cost per life year saved $80 $108 -$27
Reflexive screening is clearly cost-effective compared to provider-initiated
screening, because savings on treatment more than offset costs of
screening. But Government has to pay for screening up front, in a lump
sum; treatment costs are incurred less conspicuously.
27. Using cost-effectiveness analysis to inform policy
Intervention ICER ($/Life-year saved)
Condom availability Cost saving
Male medical circumcision Cost saving
ART at current guidelines (CD4 < 500) 106
PMTCT 138
Universal treatment (“treat all”) 243
Infant testing at 6 weeks 274
SBCC campaign (HCT, reduction in MSP) 761
SBCC campaign (condoms) 1,216
General population HCT 1,233
SBCC campaign 3 (condoms, HCT, MMC) 1,819
HCT for sex workers 2,644
Infant testing at birth 2,937
PrEP for sex workers 9,894
HCT for adolescents 19,546
PrEP for young women 26,216
Early infant male circumcision 53,785,494
Source: Meyer-Rath et al, Optimising South Africa’s HIV response: Results of the HIV and TB
Investment Case, CROI 2016
28. What To Expect Today
• Ten minutes about economic evaluation
• The setting: a safari to South Africa
• How much does it cost?
• Is it cost-effective?
• What are the benefits?
• Can we afford to do what we want?
• Final thoughts (if any are left)
30. Benefits can be estimated, but it’s a lot of work
• Prospective, adult cohort
followed for six years after
ART initiation
• 879 patients interviewed at
routine clinic visits
(average 8 interviews per
patient)
• Study focused on general
wellness, activities,
employment
• Probability of not being
able to do normal activities
in previous week fell from
47% before starting
treatment to 5% at 5 years
• Employment increased
from 32% to 44%
31. A major benefit: patients on ART can go back to
work and contribute to the economy
Comparison of productivity between HIV-negative
workers and HIV-positive workers on ART
We could estimate the overall benefits of ART using data like these…but we
rarely do. The decision to provide ART to all was political, not economic.
32. What To Expect Today
• Ten minutes about economic evaluation
• The setting: a safari to South Africa
• How much does it cost?
• Is it cost-effective?
• What are the benefits?
• Can we afford to do what we want?
• Final thoughts (if any are left)
34. Cost-effectiveness doesn’t matter if you can’t afford it
Country GDP/capita Health
expenditure/capita
(USD 2014)
Estimated cost
of
ART/patient/year
South Africa $5,692 $570 $263 (2015)
Kenya $1,377 $78 $249 (2013)
Malawi $381 $29 $135 (2016)
India $1,582 $75 No data
China $7,925 $420 No data
U.S. $55,837 $9,403 >$10,000
Sources: World Bank; unpublished reports
HIV treatment is extremely effective. Its cost has
plummeted (except in the U.S.). But can we afford it?
35. The South African HIV Investment Case
1
2
3
4
2015 2017 2019 2021 2023 2025 2027 2029 2031 2033
USD(billions)
Total cost of HIV programme
Baseline Government targets
Unconstrained optimisation Constrained optimisation
Budget Constraint 90/90/90
National Department of Health
and South African National
AIDS Council
36. What To Expect Today
• Ten minutes about economic evaluation
• The setting: a safari to South Africa
• How much does it cost?
• Is it cost-effective?
• What are the benefits?
• Can we afford to do what we want?
• Final thoughts (if any are left)
38. Why should we care about economics?
“Cost is never the first, but almost always the second
consideration in rolling out an intervention. As such it
stands between evidence and implementation.”
--BUSPH Assistant Professor Gesine Meyer-Rath
Resources are always constrained. There are always
options. Economics helps us choose.
Economics is about benefits, not just costs. It forces
us to confront our values. How much is a specific
health benefit to someone (else) worth?
Economic evaluation can incorporate considerations of
equity and fairness, but only if we are willing to make
explicit the values or weights we place on these
things.
39. Last and least: When to call an economist
Wrong answer: After we’ve finished our randomized,
controlled clinical trial and suddenly realize that it would
be nice to know if our intervention is cost-effective.
Right answer: At the start of our study, to design it
in a way that will make the results generalizable and
capture all the resources used.
Wrong answer: To prove that we can afford what we’ve
already decided to do, regardless of our resources.
Right answer: To decide what we can afford to do,
for whom, when, and how, given our resource
constraints.
40. Acknowledgements
• BUSPH South Africa team: Gesine Meyer-Rath, Matthew
Fox, Jacob Bor, Bruce Larson, Alana Brennan, and others
• HE2RO: Lawrence Long, Mhairi Maskew, Ian Sanne, and
others
• Funders: USAID, NIH, Gates Foundation, CDC, and others
• Collaborators: South African National Department of Health,
Right to Care, City of Johannesburg, and others