Determining the financial and human resource implications of HIV
treatment scale-up: A multi-country analysis in Malawi, Rwanda,
Swaziland and Zambia
Clinton Health Access Initiative and the Harvard School of Public Health in Collaboration with
Ministries
Health of Swaziland, Malawi, Zambia and Rwanda
CHAI
slideofwarehouse
International AIDS Economic Network
July 2014
This work has been funded by aid from the UK Government. The views
expressed do not necessarily reflect the UK Government’s official policies.
Countries are weighing multiple policy options with significant
implications for ART targets
ART Eligibility by Policy Option
2013 WHO Guidelines
40
35 M
Patients (Mill)
35
30
25.9 M
25
20
20.6 M
23.2 M
16.7 M
15
10
5
Policy
Options
0
Eligible
Population
On ART
2010
Option
B+
2010 Guidelines
Pregnant
women
Guidelines
andSDC
SDC
and
+ All pregnant
All under 2;
women;
Other CD4<350 Serodiscordant
couples
Children U5
Children
U5
2013
Guidelines
2013 Guidelines
(Full)
(Full)
+ All under 5
+ All CD4<500
Universal
Universal
Treatment
Treatment
All
The funding outlook for HIV is uncertain. There is a need for evidence to inform policy and
resource allocation decisions.
Source: WHO, 2013, Global Update on HIV Treatment; UNAIDS, 2014.
2
We assessed the affordability and feasibility of scale-up under
different policy options in 4 contexts
Objective
• Estimate the cost and HRH implications of reaching universal access by 2020 under
different ART eligibility policy options in Swaziland, Rwanda, Malawi and Zambia.
Methodology
• Costs: 2010/11 MATCH study and 2012 study in Swaziland updated to reflect recent
pricing and costs; Non-treatment costs from local sources and global literature.3
• Epidemiology: The Barnighausen, Bloom and Humair model (BBH), an analytically
derived HIV “combination intervention” model.3
• HRH: CHAI’s workload-based demand model.
• Available resources: Assumed available donor resources flat-line and domestic
resources increase with GDP.
• Scenario analysis: Decision making tool used by government representatives to examine
the impact of different eligibility options, models of care and testing strategies.
1- Under 2010 Guidelines; 2-Based on public reporting (expenditure tracking data unavailable); 3-Bärnighausen, T., D. E. Bloom and S. Humair
(2012). "Economics of antiretroviral treatment vs. circumcision for HIV prevention." Proceedings of the National Academy of Sciences 109(52): 3
21271-21276.; 3-CDC and the Government of the Kingdom of Swaziland, unpublished
Modeling began with the results of previous facility costing studies
Multi-Country Analysis of Treatment Costs for HIV/AIDS (MATCH) Study 2010/2011
Cost per ART Patient-Year by Country, USD
Legend
$1,200
Max
3rd Q
$1,000
Median
1st Q
Min
Avg
$800
$682
$600
$400
$200
$186
$136
$278
$232
$Malawi
Ethiopia
Rwanda
Zambia
RSA *
*RSA cost include updated ARV prices, which were renegotiated by the RSA government in early 2010 and are 53% lower than those
observed during the costing period
5
Costs were adjusted based on current pricing and expected
changes with treatment scale-up
Expected changes:
1
2
3
5
Changes in the patient mix are expected to drive
changes in cost and influence total resource needs
• Treatment intensity and cost
were adjusted by patient type:
- Commodities (e.g., 1L vs. 2L)
- Service delivery (e.g., level
and frequency of visits)
- Models of care (e.g., task
shifting)
• Patient mix changed by policy
option and over time, with a
significant proportion of patients
expected to be stable, less
complex adults.
• The weighted average cost PPPY
under the 2013 Guidelines is 510% less than the 2010
Guidelines by 2020.
Note: Epidemiological modeling from BBH
Patient Mix
Illustrative Cost PPPY
$450
$300
$150
$New
Adults >350
New Est Adults - Est Adults - PMTCT
Adults <350
>350
<350
ARV
Lab
Personnel
Pediatric
Patients
Other Costs
Illustrative Projected ART Patient Mix in 2020
New <200
New 200-350
2010 Guidelines
New 350-500
Est <200
2013 Guidelines
Est 200-350
Est 350-500
0%
20%
40%
60%
80%
100%
6
Changes in price and mix of commodities by patient
type are expected to affect drug and lab costs…
• Within patient types, the unit
costs for commodities was
adjusted for normative
consumption and price changes
• ARVs: The 1L/2L mix and cost
determined unit costs over
time; Costs are unlikely to
increase substantially in the
short-term given price
reductions and slow uptake of
newer 3L drugs
• Labs: The uptake of new
technologies and the mix of
conventional and POC tests was
included; Lab costs will increase,
but remain a small proportion of
overall Tx costs
Commodity Mix
ARVs – Illustrative Changes in Price and Mix
ESTABLISHED PATIENTS – GENERIC ACCESSIBLE COUNTRIES
2013
2014
2015
% 1L regimens
96.0%
95.5%
95.0%
% 2l regimens
4.0%
4.5%
5.0%
Average 1L cost
$120
$132
$129
Average 2L cost
$393
$351
$342
Labs – Illustrative Changes in Price and Mix
NEW PATIENTS - SELECT COUNTRIES
2013
2014
2015
Avg cost / Conventional CD4 test
$9.62
$9.14
$8.68
Avg cost / Device-based POC CD4
$11.62 $11.04 $10.49
Average cost / Device-free POC CD4 test $11.17 $10.61 $10.08
% Conventional CD4
85%
75%
65%
% Device-based POC CD4
15%
20%
25%
% Device-free POC CD4
0%
5%
10%
1
1
1
Average # CD4 tests
Source: CHAI, 2014
7
…While how and where care is delivered will have
Models of Care
a significant impact on personnel costs
- Where patients seek care
- How often
- With which cadre
- For how long
• Task-shifting and multi-month
scripts have been adopted for
certain patients in Swaziland
and Malawi
• More intensive home visit
programs for complex patients
have been adopted in Rwanda
• Countries are piloting other
models of care (e.g., ART clubs
and SMS messaging)
Illustrative Example for Personnel: Non-complex
patients can be managed with fewer touch-points
Visits by Cadre Per Year for Different Patient
Types
12
10
Visits
• Amongst patient types, different
models of service delivery are
being applied. This affects:
8
6
4
2
0
Complex
Non-complex
8
Our methodology is as robust as current evidence allows, but
contains important limitations
Key Limitations
• Treatment and care, testing, condoms and VMMC are included. The following are
excluded:
- Other HIV-related and prevention interventions that are difficult to reliably
cost, such as BCC and OVC care;
- Program management costs that will increase, but not proportionately with
patients on ART; and
- Systems costs, such as expansion of supply chain and lab systems.
• The implication is not that these costs are not important, but that they are not well
understood. Interpretation of the data must take into account the need to set aside
funding for non-costed programs.
• The implications of scale-up on costs require further refinement. Economies of scale will
reduce costs, but decentralization may lead to services provided in smaller facilities with
lower utilization rates in some areas.
• The costs of identifying, initiating and retaining asymptomatic patients requires further
refinement. Analysis is underway.
9
At universal access, costed programs account for < 60% of projected
available resources
Universal Access to Treatment in 2020
Costs increase by 10-20% in moving to the 2013 Guidelines
$100
Swaziland
$150
Millions (USD)
Millions (USD)
$150
5%
9%
$50
$-
$100
17%
8%
$50
$0
2010 Guidelines 2013 Guidelines
(Full)
Universal
Treatment
Zambia
$750
Millions (USD)
Rwanda
$600
$450
19%
1%
$300
$150
$0
2010 Guidelines
2013 Guidelines (Full) Universal Treatment
2010 Guidelines 2013 Guidelines
(Full)
Universal
Treatment
MC
Condoms
Tests
Palliative Care
Pre-ART
Pediatric
PMTCT
ART
Projected Resources
Note: Testing strategy mix varied across policy options; Resources are projected from national resource mapping exercises in 20122013 with the exception of Zambia where publicly available data was used.
10
In Malawi, universal access may not be affordable There is an urgent need for additional funding
Universal Access to Treatment in 2020
Condoms
Malawi
$350
MC
$300
Millions
$250
$200
$150
Resource Envelope
Tests
Gov Health Expenditure/
Total Gov Expenditure1
6.7%
Palliative
Care
Pre-ART
Health Expenditure as %
of GDP3
9.2%
Total Health Expenditure
(% External) 2
$642 M
(81%)
Total HIV expenditure
(% External) 2
$215 M
(99%)
Pediatric
$100
PMTCT
$50
ART
$2010 Guidelines 2013 Guidelines
(Full)
•
•
Universal
Treatment
Projected
Resources
Malawi is one of the poorest countries in the world with little ability to contribute additional
funding towards HIV.
Universal access under the 2013 Guidelines would account for almost half of the current health
budget.
1- Malawi NHA 2011/2012; 2-National Resource Mapping, 2013; 3-World Dev.Indicators, 2012
11
Innovative models of care are mitigating costs in the short and longterm
Models of care can reduce the costs of scale-up in the short-term…
Malawi Cost PPPY
300
•
Task shifting,
MMS
200
•
100
•
0
Before Intervention
ARVs
Labs
Personnel
Optimized
Other Costs
In Malawi Multi-month scripts (MMS) and task
shifting have reduced personnel costs by ~30%.
Home visits for complex patients would only
slightly increase costs (~ 5%) and could
improve retention
Additional evidence is needed on the effects of
these models on retention.
…and in the long-term by improving patient retention.
% Change in Cumulative Cost
Costs of achieving UA vs.
Retention
% Change in
Retention
0%
0%
3%
-1%
5%
10%
Across 4 countries, a 5% increase in retention
results in the following by 2020:
• 4-6% Reduction in new infections
• 4-6% Reduction in AIDS-related deaths
• Up to 4% reduction in treatment/testing costs
-2%
-3%
-4%
Malawi
Rwanda
Zambia
Note: Figures do not include spending required for
finding patients who have been lost to follow-up.
Additional analysis is underway.
12
Countries will face HRH challenges, but changes in patient mix
make the incremental impact of the 2013 Guidelines negligible
Swaziland: Facility-Level HRH Required for HIV Treatment and Care (Without Testing)
- 1.2% Health
Workers Required
+ 7.4% Patients
800
145,835
156,654
Patients
160,000
120,000
80,000
40,000
0
Health Workers (FTE)
200,000
754
745
600
400
200
0
2010 WHO Guidelines
New Adult, CD4 < 350
PMTCT Patients
2013 WHO Guidelines
New Adult, CD4 > 350
Pediatric & Infant HIV Patients
2010 WHO Guidelines
Est Adult, CD4 < 350
Pre-ART Patients
2013 WHO Guidelines
Est Adult, CD4 > 350
• Existing HRH shortages (Current workforce accounts for 53% of the projected sector-wide
need in 2020)
• Negligible incremental change in HRH needs for HIV under the 2013 Guidelines due to
epidemiological changes and lower intensity of care for asymptomatic patients
• Finding, testing and linking patients is not included and will require significant staff time
depending on the strategy
• Similar change in health workers required seen in Zambia (-0.2%) and Malawi (-0.7%)
13
Conclusion: Debate should shift from whether to scale-up ART to
how to do so efficiently
Key Takeaways
Affordability: In Swaziland, Rwanda and Zambia, the cost of scale-up is manageable
within the existing funding envelope, if programs run more efficiently. Malawi will face
significant financial constraints without additional aid.
Feasibility: There are sector-wide HRH shortages. However, the incremental HRH to reach
universal access to ART under more aggressive scenarios is less than expected.
Key Considerations:
- Excluded costs such as BCC and OVC and program management are important, but
additional evidence is needed on cost and impact.
- Upfront investment may be required (e.g., reaching hard-to-reach populations,
building up systems and covering remote areas).
- Country specific challenges exist given different levels of operational capacity,
resources and efficiency of spending.
- HRH requirements will depend on the strategies used to find, test and link patients.
Additional analysis is needed.
14
Annex
15
Human Resource Analysis Annex
Human Resource Analysis
• In 2009, CHAI developed a workforce optimization model that was then rolled out
across 5 countries between 2009 and 2013. Model is in STATA.
• For this project, the model has been updated in Zambia, Swaziland and Malawi.
Rwanda was omitted as data was unavailable and scale-up is unlikely to result in HR
constraints
• Data sources included:
- Facility visits and time-motion observations
- Interviews with clinical experts
- National treatment protocols
- Facility-level service delivery data from national health management
information systems (HMIS)
- Facility staffing data
Methodology:
Activities and
times based
on facility
observations
and interviews
Incidence
Data:
HMIS adjusted
for pop.
growth; BBH
Total time to
meet
demand
Health worker
productivity
Optimal
workforce for
each health
worker cadre
16