Post on 28-Mar-2015
transcript
1
Facility-based Unit Costing for Antiretroviral Treatment in Ethiopia, Malawi, Rwanda, South Africa, and Zambia
Preliminary Findings
Agenda
2
I. Overview of Methodology
II. Key findings and areas of opportunities
III. Annex slides
Objective Through this exercise, we set out to answer the following research questions:• What are typical costs of ARV treatment per patient year in each country?• How do these costs vary across countries, funding streams, administrators, facility
types, rural/urban location, and other factors? How do patient outcomes vary?• What are the drivers of this variation in treatment cost across sites?• What lessons can be learnt from this variation to improve the efficiency and
effectiveness of treatment at the global, national, and facility level?
Project plan • CHAI supported the governments of Ethiopia, Malawi, Rwanda, South Africa, and Zambia, to conduct facility-level costing of HIV treatment (ART). The study was conducted at 161 health facilities across the 5 countries.
• Study methodology was designed by the study team, refined through field testing and reviewed by a Technical Advisory Group in consultation with an expert group
• Data collection was executed during 2011• The study team has completed a preliminary descriptive analysis. An analysis of the
determinants of cost and retention will follow in the coming months
The overall objective of the study was to identify opportunities to improve efficiency and effectiveness of HIV treatment spending
The study provides a top-down assessment of ART costs per patient-year in a random sample of sites
4
Study periodThe study captured actual spend data for the most recently available full year on record (either fiscal or calendar year), which in most countries began in 2010 and began no earlier than mid-2009
Level The study was conducted at facility level, not at a program, patient or cohort level. It includes all ART related spend at the facility level, but no spend incurred above the facility (e.g., OH, system strengthening) or outside the facility (e.g., community programs), or on inpatient costs
Costing approach
Top down: starting with total cost at a given facility, allocated costs to ART, and divided by total patient years to arrive at an average cost per patient-year
Patient outcome indicators
Basic complexity and outcome indicators derived from patient chart reviews. The indicators are used to adjust costs and arrive at cost per patient alive and on ART at 12 months
Agenda
5
I. Overview of Methodology
II. Key findings and areas of opportunities
III. Annex slides
1. Facility-level treatment costs in LICs and LMICs are lower than previously understood, ranging from USD 136 - 278 per patient-year. Not unexpectedly, treatment costs are higher in RSA due to higher salaries and more laboratory testing.
2. Facilities have demonstrated the ability to keep established patients alive and on ART, with some variation across the sample. Average annual attrition rates for established patients range from 2% in Rwanda to 8% in RSA. Retention rates for new patients are lower and appear to be strongly correlated with CD4 at initiation.
3. The findings from our study, together with data from other sources, may have some interesting implications on the E2 agenda over the next several years:a. Aggressive scale up in LIC/LMICs should be affordable.b. Given the low cost of treatment, there are not significant savings opportunities at the
facility level in LIC/LMICs. c. To generate savings, we will need to focus optimization efforts on non-facility treatment
and other program costs. A continued focus on drug optimization is also required to prevent costs from increasing as patients move from D4T to TDF
d. Unlike in LIC/LMICs, optimizing service delivery design in RSA (and potentially other MICs) will be necessary to achieve universal access
e. There are opportunities to improve outcomes without substantially increasing cost. A focus on earlier initiation is likely needed to make significant improvements in patient outcomes.
Key findings
6
Malawi Ethiopia Rwanda Zambia RSA -
100
200
300
400
500
600
700
800
900
Cost of treatment per ART patient year by countryUS Dollars
Facility level costs are lower than expected in LIC/LMIC countries, at an average of $136 in Malawi, $186 in Ethiopia, $232 in Rwanda, and $278 in Zambia. Average cost in RSA is $682
Max
Min
Median
1st Q
3rd Q
Avg
Legend
Country GDP per capita
Malawi $876 Ethiopia $1,033 Rwanda $1,155 Zambia $1,550 South Africa $10,486
*
*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
**
Cost of treatment per ART patient-year by countryUS Dollars
ARVs constitute ~50% of total cost in all LIC/LMICs; ARVs and personnel together constitute over 70% of total cost in all countries
*Lab category includes reagents and consumables only in all countries except SA** Simple average costs are typically higher than median costs. Averages will be weighted by sampling frame in the next phase of analysis.
Simple average and median cost of treatment per ART patient-year by countryUS DollarCountry ARVs Personnel Labs* Other Total Cost
Mean Median Mean Median Mean Median Mean Median Mean MedianMalawi $66 $63 $29 $22 $5 $1 $36 $32 $136 $117Ethiopia $103 $101 $28 $22 $16 $15 $39 $35 $186 $184Rwanda $114 $112 $67 $56 $15 $16 $37 $30 $232 $210Zambia $155 $155 $73 $46 $13 $13 $37 $30 $278 $250RSA $181 $179 $334 $284 $102 $102 $65 $56 $682 $615
Malawi Ethiopia Rwanda Zambia RSA$0
$100
$200
$300
$400
$500
$600
$700
$800
$900Personnel ARVs Labs Other
1. Facility-level treatment costs in LICs and LMICs are lower than previously understood, ranging from USD 136 - 278 per patient-year. Not unexpectedly, treatment costs are higher in RSA due to higher salaries and more laboratory testing.
2. Facilities have demonstrated the ability to keep established patients alive and on ART, with some variation across the sample. Average annual attrition rates for established patients range from 2% in Rwanda to 8% in RSA. Retention rates for new patients are lower and appear to be strongly correlated with CD4 at initiation.
3. The findings from our study, together with data from other sources, may have some interesting implications on the E2 agenda over the next several years:a. Aggressive scale up in LIC/LMICs should be affordable.b. Given the low cost of treatment, there are not significant savings opportunities at the
facility level in LIC/LMICs. c. To generate savings, we will need to focus optimization efforts on non-facility treatment
and other program costs. A continued focus on drug optimization is also required to prevent costs from increasing as patients move from D4T to TDF
d. Unlike in LIC/LMICs, optimizing service delivery design in RSA (and potentially other MICs) will be necessary to achieve universal access
e. There are opportunities to improve outcomes without substantially increasing cost. A focus on earlier initiation is likely needed to make significant improvements in patient outcomes.
Key findings
9
Average annual attrition rates for established patients range from 2 – 8%, while retention in the first 12 months of treatment is significantly lower
10
Average attrition (at 12 months) for established patientsPercent
• Sites in the sample have demonstrated an ability to keep patients alive and on treatment. However, variation is significant for both new and established patients.
• Rwanda stands out for their ability to manage both newly initiated patients as well as those receiving long-term care.
• Further analysis is required to better understand the observed differences across the sample.
Pending Further Analysis
Average retention (at 12 months) for new patientsPercent
Rwanda Malawi RSA Zambia Ethiopia0%
10%20%30%40%50%60%70%80%90%
100% 95%
77% 82%89% 84%
Rwanda Malawi RSA Zambia Ethiopia0%1%2%3%4%5%6%7%8%9%
2%
6%
8%
6%7%
* Transferred patients were excluded from analysis, which may reduce retention rates for sites that transfer stable patients, and increase rates for sites that transfer sicker patients.
Preliminary analysis shows a strong correlation between CD4 at initiation and new patient retention at 12 months
11
Retention rates at 12 months for new patients by CD4 level at initiation Percent, CD4 count
Preliminary pending further analysis
* Only 20% of patients in the sample had an available CD4 value at initiation
• Preliminary analysis of 7,611 patient records suggests a strong correlation between stage of initiation and ability to maintain patients alive and on treatment
• This trend appears to support evidence in the literature linking early initiation to better outcomes
• However only 27% of patients across the sample initiate at CD4 >200, while 20% initiated with no CD4 test
• ~$90ppy is spent on manage pre-ART patients, yet outcomes (e.g, baseline CD4) are sub-optimal. Better understanding of pre-ART program spending and drivers of late initiation should be prioritized0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
82%87% 90% 93%
76%
0-100 100-200 200-300 300-400 No CD4
24% 29% 18% 9% 20%% of patients in category
1. Facility level treatment costs in LIC and LMICs are lower than previously understood, ranging from USD 136 - 278 per patient per year. Not unexpectedly, treatment costs are higher in RSA due to higher salaries and more laboratory testing.
2. Facilities have demonstrated the ability to keep established patients alive and on ART, with some variation across countries and facilities. Average annual attrition rates for established patients range from 2% in Rwanda to 8% in RSA. Retention rates for new patients are lower and appear to be strongly correlated with CD4 at initiation.
3. The findings from our study, together with data from other sources, may have some interesting implications on the E2 agenda over the next several years:a. Aggressive scale up in LIC/LMICs should be affordable.b. Given the low cost of treatment, there are not significant savings opportunities at the
facility level in LIC/LMICs. c. To generate savings, we will likely need to focus optimization efforts on non-facility
treatment and other program costs. A continued focus on drug optimization is also required to prevent costs from increasing as patients move from D4T to TDF
d. Unlike in LIC/LMICs, optimizing service delivery design in RSA (and potentially other MICs) will be necessary to achieve universal access
e. There are opportunities to improve outcomes without substantially increasing cost. A focus on earlier initiation is likely needed to make significant improvements in patient outcomes.
Key findings
12
Universal access to treatment in LIC/LMIC would only cost an estimated $2B in facility-level costs. Aggressive scale up in LIC/LMICs should be affordable.
Global HIV Spending Current At Universal Access$0
$2,000
$4,000
$6,000
$8,000
$10,000
$12,000
$14,000
$16,000
$18,000
ART Pre-ART
Rough estimates of funding required for treatment in LIC/LMICs, USD millions
2
* Assuming that cost pppy stays constant as countries scale up to achieve universal access
Total Cost
Current volume = ~3.7M patients (2010 UNAIDS Report)Universal Access= ~10M patients (LIC /LMIC share of 15M HLM target)
$2B
$0.8B
1. The facility-level cost of universal access to treatment in LIC/LMICs will be around ~$2B. Aggressive scale up should be affordable.
2. This cost estimate does not include a number of important cost elements that will be critical to achieving universal access (community outreach, etc) but we do not expect that those will drive cost.
3. Work to better understand the non-facility based cost of treatment should be prioritized.
$16B
1. Facility level treatment costs in LIC and LMICs are lower than previously understood, ranging from USD 136 - 278 per patient per year. Not unexpectedly, treatment costs are higher in RSA due to higher salaries and more laboratory testing.
2. Facilities have demonstrated the ability to keep established patients alive and on ART, with some variation across countries and facilities. Average annual attrition rates for established patients range from 2% in Rwanda to 8% in RSA. Retention rates for new patients are lower and appear to be strongly correlated with CD4 at initiation.
3. The findings from our study, together with data from other sources, may have some interesting implications on the E2 agenda over the next several years:a. Aggressive scale up in LIC/LMICs should be affordable.b. Given the low cost of treatment, there are not significant savings opportunities at the
facility level in LIC/LMICs. c. To generate savings, we will likely need to focus optimization efforts on non-facility
treatment and other program costs. A continued focus on drug optimization is also required to prevent costs from increasing as patients move from D4T to TDF
d. Unlike in LIC/LMICs, optimizing service delivery design in RSA (and potentially other MICs) will be necessary to achieve universal access
e. There are opportunities to improve outcomes without substantially increasing cost. A focus on earlier initiation is likely needed to make significant improvements in patient outcomes.
Key findings
14
-
Given the low cost of treatment, there do not appear to be significant savings to be gained by optimizing service delivery design in LIC/LMICs.
1. Current non-ARV costs 2. Shift all to 1st quartile 3. Calculate savings
Weighted average
1st quartile of facilities by country
Average savings per patient
Total patients in visited sites
Weighted* Average
1st quartile
Savings ppy % savings
Rwanda 11,848 91 69 22 24%
Malawi 38,625 65 43 22 34%
Zambia 28,424 98 75 23 23%
Ethiopia 33,444 72 62 10 14%
Weighted average 78 59 19 24%
=
$19 pppy
X
10 M patients
=
$190 M
4. Savings potential in study countries, excl. SA 5. Savings potential in LIC/LMC portion of 15M target
* Weighted by patient load across facilities in the sample. Since smaller facilities have higher costs, weighted is lower than simple average
Rwanda Malawi Ethiopia Zambia -
20
40
60
80
100
120
140
1. Facility level treatment costs in LIC and LMICs are lower than previously understood, ranging from USD 136 - 278 per patient per year. Not unexpectedly, treatment costs are higher in RSA due to higher salaries and more laboratory testing.
2. Facilities have demonstrated the ability to keep established patients alive and on ART, with some variation across countries and facilities. Average annual attrition rates for established patients range from 2% in Rwanda to 8% in RSA. Retention rates for new patients are lower and appear to be strongly correlated with CD4 at initiation.
3. The findings from our study, together with data from other sources, may have some interesting implications on the E2 agenda over the next several years:a. Aggressive scale up in LIC/LMICs should be affordable.b. Given the low cost of treatment, there are not significant savings opportunities at the
facility level in LIC/LMICs. c. To generate savings, we will likely need to focus optimization efforts on non-facility
treatment and other program costs. A continued focus on drug optimization is also required to prevent costs from increasing as patients move from D4T to TDF
d. Unlike in LIC/LMICs, optimizing service delivery design in RSA (and potentially other MICs) will be necessary to achieve universal access
e. There are opportunities to improve outcomes without substantially increasing cost. A focus on earlier initiation is likely needed to make significant improvements in patient outcomes.
Key findings
16
To generate savings, we will need to focus optimization efforts on non-facility treatment and other program costs.
* See, for example, Effectiveness of HIV prevention strategies in resource-poor countries: tailoring the intervention to the context, Padian et al, AIDS 2006Sources: JAWP data and CHAI analysis, UNAIDS for prevention spending, GFATM Enhanced Financial Reporting Mechanism
BCC - community
BCC - Mass
media
Blood safety
Condom distrib
ution
PMTCTPEP
STI d
iagnosis and Tx
Testing and Counselin
g0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
New cases Funding New cases FundingMozambique Kenya
0%
50%
100%
30.0%
0.3%
30.0%
0.4%
High risk Other groups
Across population groupsIn two countries, high risk groups account for a large % of new infections but receive <1% of prevention spending
% of new cases and prevention fundingUS Dollars
Prevention spending at the GFATM % of total prevention
Across interventionsBCC accounted for >45% of prevention, and 14% of total GFATM HIV spending to 2009; Far less was spent on PMTCT, and other interventions.
Country 1 Country 2
Examples of allocative inefficiencies in HIV spending
1. Facility level treatment costs in LIC and LMICs are lower than previously understood, ranging from USD 136 - 278 per patient per year. Not unexpectedly, treatment costs are higher in RSA due to higher salaries and more laboratory testing.
2. Facilities have demonstrated the ability to keep established patients alive and on ART, with some variation across countries and facilities. Average annual attrition rates for established patients range from 2% in Rwanda to 8% in RSA. Retention rates for new patients are lower and appear to be strongly correlated with CD4 at initiation.
3. The findings from our study, together with data from other sources, may have some interesting implications on the E2 agenda over the next several years:a. Aggressive scale up in LIC/LMICs should be affordable.b. Given the low cost of treatment, there are not significant savings opportunities at the
facility level in LIC/LMICs. c. To generate savings, we will likely need to focus optimization efforts on non-facility
treatment and other program costs. A continued focus on drug optimization is also required to prevent costs from increasing as patients move from D4T to TDF
d. Unlike in LIC/LMICs, optimizing service delivery design in RSA (and potentially other MICs) will be necessary to achieve universal access
e. There are opportunities to improve outcomes without substantially increasing cost. A focus on earlier initiation is likely needed to make significant improvements in patient outcomes.
Key findings
18
19
In RSA, given higher patient numbers and costs, the opportunity for efficiency gains is substantial
Potential opportunities include:
Weighted average: $402
Potential size of savings based on non-ARV cost distribution in sample of facilities
1st quartile of facilities: $376
Average savings: $26
x2015 patient numbers: 2.93M
=Potential savings: $76M
=
-
• Optimizing staffing models across facilities
• Rationalizing frequency of visits to balance demand on HRH, travel burden on patients and retention/adherence benefits of clinical visits
• Optimizing testing level, especially non-CD4/VL
• Increasing the proportion of CD4 tests that that are returned to patients
Weighted average brought down by low cost, large facilities. Given aggressive scale-up plans in RSA,
saving potential might be greater
1. Facility level treatment costs in LIC and LMICs are lower than previously understood, ranging from USD 136 - 278 per patient per year. Not unexpectedly, treatment costs are higher in RSA due to higher salaries and more laboratory testing.
2. Facilities have demonstrated the ability to keep established patients alive and on ART, with some variation across countries and facilities. Average annual attrition rates for established patients range from 2% in Rwanda to 8% in RSA. Retention rates for new patients are lower and appear to be strongly correlated with CD4 at initiation.
3. The findings from our study, together with data from other sources, may have some interesting implications on the E2 agenda over the next several years:a. Aggressive scale up in LIC/LMICs should be affordable.b. Given the low cost of treatment, there are not significant savings opportunities at the
facility level in LIC/LMICs. c. To generate savings, we will likely need to focus optimization efforts on non-facility
treatment and other program costs. A continued focus on drug optimization is also required to prevent costs from increasing as patients move from D4T to TDF
d. Unlike in LIC/LMICs, optimizing service delivery design in RSA (and potentially other MICs) will be necessary to achieve universal access
e. There are opportunities to improve outcomes without substantially increasing cost. A focus on earlier initiation is likely needed to make significant improvements in patient outcomes.
Key findings
20
There are clear opportunities to improve patient outcomes without substantially increasing cost and these should prioritized
RSA Zambia Rwanda Ethiopia Malawi$0
$2
$4
$6
$8
$10
$12
$4.5
$3.4
$2.0
$4.5
$3.6
$10.9 $11.2
$5.3
$6.9 $6.6
Average cost pppy of providing cotrimoxazoleUS Dollar
Current spend on cotrim pppy
Spend on cotrim pppy at 100% uptake
Increasing uptake of cotrim to 100% will increase costs by $2-$10 pppy
Driving earlier initiation through increased testing and improved pre-ART program will help increase retention
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
82%87% 90%
93%
76%
0-100 100-200 200-300 300-400 No CD4
New patient retention at 12 months with baseline CD4Percent, CD4 count
24% 29% 18% 9% 20%% of patients in category
Universal access to treatment in LIC/LMIC would only cost an estimated $2B in facility-level costs. Aggressive ART scale-up should be affordable.
Global HIV Spending Current At Universal Access$0
$2,000
$4,000
$6,000
$8,000
$10,000
$12,000
$14,000
$16,000
$18,000
ART Pre-ART
Rough estimates of funding required for treatment in LIC/LMICs, USD millions
1
2
1 Assuming that cost pppy stays constant as countries scale up to achieve universal access2 Assuming rates found across the sample: pre-ART costs pppy are 50% of ART cost pppy, and number of pre-ART patient years are 25% of ART patients
Total Cost
Current volume = ~3.7M patients (2010 UNAIDS Report)Universal Access= ~10M patients (LIC /LMIC share of 15M HLM target)
$2B$0.8B
$16B