AMDS Partners and Stakeholders Meeting CHAI HIV Diagnostics Forecasting Overview 29-30 th September, 2014
Agenda Introduction Overview of Global Diagnostics Forecasting Overview of Country Forecasting Methodologies Forecasting Example: EID Commodities FORLABS Introduction Recommendations for Future Diagnostics Forecasting Appendix 2
CHAI operates programs in more than twenty-five countries around the world Historically, CHAI has supported forecasting and procurement in these countries to promote access to high-quality medicines and diagnostics for HIV, TB, Malaria, and Essential Child Medicines.
CHAI works extensively with its program countries on HIV diagnostics forecasting, particularly for EID CHAI country teams have historically procured commodities for EID under the UNITAID Pediatric Project EID quantifications were developed in coordination with Ministries of Health, country teams, and CHAI global team staff using: CHAI Global EID Forecasting Tool Country team-originated tools Many of these tools have subsequently been transitioned to Ministries of Health CHAI s work also supports forecasting for other HIV diagnostics, including rapid test kits, viral load monitoring, and CD4 Point of Care staging and monitoring.
Agenda Introduction Overview of Global Diagnostics Forecasting Overview of Country Forecasting Methodologies Forecasting Example: EID Commodities FORLABS Introduction Recommendations for Future Diagnostics Forecasting Appendix 5
CHAI s global diagnostics forecasts provide market intelligence to suppliers - Approach - CHAI s global diagnostics forecasts estimate the market size for CD4, viral load, and EID testing in 21 high-burden countries The global forecasts are generated by a number of sources, including: Country and WHO-reported patient numbers Country testing guidelines and scale-up plans Assumptions regarding market dynamics (i.e., timing of introduction of POC, actual scale-up rates) These global forecasts are primarily shared with suppliers to inform them of existing demand and expected future market trends Forecasts are updated on an annual basis
A bottom-up market forecast of viral load using country current capacity and scale-up plans demonstrates significant growth Millions - Global Viral Load Forecast - 25 20 15 10 5 - Total Viral Load Demand 2013 2014 2015 2016 2017 2018 2019 Plasma volumes DBS volumes POC volumes Need To generate the forecast, a number of countries were consulted in detail regarding their scale-up plans. Publically-available patient numbers, test 1 volumes, and guidelines were used to estimate growth for the remaining countries.
The CD4 market is expected to continue to grow modestly alongside the scale-up of viral load Millions of Tests 30 - Global CD4 Forecast - 25 20 15 10 5 0 2013 2014 2015 2016 CD4 conventional CD4 POC As more POC products become available on the market, CHAI expects that some POC scale-up will begin to cannibalize some conventional testing volumes.
Though the EID DBS market is well-established, coverage remains low and there is a large loss to follow up Millions of Tests - Global EID Forecast - 3.0 2.5 2.0 1.5 1.0 0.5 0.0 EID DBS volumes EID POC volumes EID unmet need 2013 2014 2015 2016 Coverage of EID remains low in most countries, in part due to poor linkage between PMTCT programs and infant care.
Agenda Introduction Overview of Global Diagnostics Forecasting Overview of Country Forecasting Methodologies Forecasting Example: EID Commodities FORLABS Introduction Recommendations for Future Diagnostics Forecasting Appendix 10
Management of laboratory systems in most countries is highly fragmented, complicating diagnostics forecasting - Key Challenges- Laboratory management within countries silo-ed, with divisions across disease areas and among partners Complex commodity and equipment needs require highly specialized knowledge to make accurate forecasts Poor understanding of demand and wastage due to extensive system decentralization and lack of device connectivity Lack of standardized forecasting methodology List of specific testing commodities and anticipated consumption rate not always available Lack of inventory visibility, particularly at facility level Countries lack supply chain consolidation and rigorous oversight of laboratory management, further challenging diagnostics forecasting and procurement for HIV and TB.
Consumption, morbidity, and country testing targets can be used to generate demand forecasts for various diagnostic tests - Key Forecasting Methodologies- 1 Consumption data-based: Quantity of each product dispensed or consumed by facilities and labs over a given time period (i.e., 12 months) 2 Service data-based: Number of tests run for particular diagnostic test (this is useful in understanding wastage) 3 Morbidity and target-based: Prevalence related to specific target disease (i.e., HIV prevalence in infants under 18 months for EID testing), modified by country program performance plans CHAI relies on service data, consumption data, and country targets (performance plans) in its diagnostics forecasts to minimize wastage and more accurately model growth.
There are advantages and limitations to each forecasting methodology Based on historical consumption of commodities in time period (e.g., average monthly consumption) Advantages Consumption-Based Morbidity-Based Service Statistics-Based Forecast if patient numbers/ demographics are not known Realistic predictor of growth Best reflects repeat testing, quality control, and lab training requirements Limitations Fails to meet true testing demand Often cannot account for scaleup or stock outs Consumption data is difficult to collect and manage Based on number of patients, demographic information, testing guidelines, and usage rates per test Advantages Model scale up based on changes in guidelines Accounts for full demand Limitations Targets uncertain or aspirational Patient #, demographics difficult to collect Guidelines not always followed Does not capture repeat testing, quality control, training at the lab Use with caution Historical data captured at the program or facility level that details the number of tests performed Advantages Forecast if patient numbers/ demographics are not known Use for all testing (non-hiv) Limitations Fails to meet true testing demand Often cannot account for scaleup or stock outs Does not capture repeat testing, quality control, training at the lab Data may be difficult to capture Most models rely on a combination of different forecasting methods and do not rely on morbidity or service statistics-based forecasts alone. Use with caution 13
Stock Considerations Instrument and Product List Patient demographics Testing requirements Liaising with the the laboratory to to make assumptions and and collect inputs for is critical is critical to to generating generate accurate, and efficient efficient forecasts forecasts - Sample Forecast Inputs - Current ART patient numbers by site Future ART patient targets by site Adult vs. pediatric patient numbers HIV prevalence rate Loss-to-follow-up at diagnosis Attrition over time Migration from Pre-ART to ART Number of baseline and monitoring tests per patient per year for each type of test Different guidelines for different populations of patients, e.g. on different drug regimens Testing algorithms, e.g. serial vs. parallel How many controls are performed per test Product wastage as a result of spillage, incorrect measurement, expiries, or damage (typically ~3-10%) Assumptions on % of testing used for EQA and training Lead time stock, or stock on hand in between order and receipt of new stock (minimum stock) Buffer stock, or stock on hand kept to guard against delayed deliveries, increased consumption, or other unexpected events Full stock level calculation Which instruments are used in the country and key characteristics (e.g., throughput) Which instruments are placed at each site Which sites refer samples to different sites for testing Which products are used for each instrument Amount of each product used for 1 test Pack size, price, and expiry for each product If commodities are not accurately forecast, their procurement will not be consistent with program needs, jeopardizing patient care. 14
Agenda Introduction Overview of Global Diagnostics Forecasting Overview of Country Forecasting Methodologies Forecasting Example: EID Commodities FORLABS Introduction Recommendations for Future Diagnostics Forecasting Appendix 15
Example: Quantification of EID Commodities - Need - - Procurement - Calculation of total demand in terms of # Samples to be collected and # Tests to be performed Target # HEIs National targets EID Algorithm MTCT rate Consumption Data Scale Up plan (growth expected) Machine downtime Stock Outs 1 # Samples (excluding wastage) # Tests (excluding controls) Calculation of total # of product packs to be procured to fulfill the demand Sample Rejection Rate Loss from Labs & Central Stores # Controls Loss from Labs & Central Stores 2 # Samples (including wastage) # Tests (including controls) Data Input Product Type # EID Sites Stock Level Buffer Stock # Labs Sites Stock Level Buffer Stock Loss from Labs & Central Stores Product Type Platform 3 Starting point Intermediate Output # DBS kits # Reagents # Consumables bundles Output
Products Example: Quantification of EID Commodities - Need - - Procurement - 2013 3,860 Target # HEI HIV+ Infants (37%) are tested again (with new sample) to Confirm HIV- Positivity HIV- infants (63%) are tested again at the end of breast feeding 50% HIV- infants lost between 1 st and 2 nd test 5,790 # Samples (excluding wastage) # Tests (excluding wastage) 5,790 HIV+ Infants (37%) are tested again (with new sample) to Confirm HIV- Positivity HIV- infants (63%) are tested again at the end of breast feeding 50% HIV- infants lost between 1 st and 2 nd test 1-3% samples rejected 5,850 # Samples (including wastage) # Tests (including wastage) 6,510 12 Controls Used for Every 96 Tests DBS Bundles - 20 tests Reagent Kits - 48 tests or 96 tests Lab Consumables Bundles 960 tests 332 # DBS 93 # Reagents 9 # Consumables 10% Buffer Stock 3-5% Qtrly Loss from Labs & Central Stores (%)
Products Example: Quantification of EID Commodities The target number of exposed infants translates in a much higher number of tests needed in consideration of the national testing algorithm for EID 2013 Exposed babies - Need - - Procurement - 3,860 Exposed Infants Tests 3,860 Target # HEI HIV+ Infants (37%) are tested again (with new sample) To Confirm HIV-Posi vity HIV- infants (63%) are tested again at the end of breast feeding 50% HIV- infants lost between 1 st and 2 nd test 5,790 # Samples (excluding wastage) # Tests (excluding wastage) 5,790 HIV+ Infants (37%) are tested again (with new sample) To Confirm HIV-Posi vity HIV- infants (63%) are tested again at the end of breast feeding 50% HIV- infants lost between 1 st and 2 nd test 1-3% samples rejected 5,850 # Samples (including wastage) # Tests (including wastage) 6,510 12 Controls Used for Every 96 Tests (37%) DBS Bundles - 20 tests Reagent Kits - 48 tests or 96 tests Lab Consumables Bundles 960 tests 332 # DBS First Test Confirmatory Test 93 # Reagents 9 # Consumables 10% Buffer Stock 3-5% Qtrly Loss from Labs & Central Stores (%) (63%) End of Breastfeeding 3 rd Test Total Tests 3,860 710 1,220 5,790 HIV- or Not tested 36.8%
Agenda Introduction Overview of Global Diagnostics Forecasting Overview of Country Forecasting Methodologies Forecasting Example: EID Commodities FORLABS Introduction Recommendations for Future Diagnostics Forecasting Appendix 19
FORLABS the new diagnostic forecasting tool Software developed in collaboration with USAID, JSI, SCMS and CHAI Uses consumption, service statistics and morbidity data to forecast product need for lab services Capabilities Reports can be generated for individual diagnostic areas (CD4, Chem, Heme, VL, ect.), as well as aggregated tests Provide a summary of instrument utilization rates by platform/sites (test numbers or estimated test numbers vs. instrument capacity) Provide a summary of instrument diagnostic contribution (number and % of tests performed on each instrument platform) Report on comparison of forecast accuracy among methodologies against observed consumption Offer in a dashboard a graphical representation of various reports
Agenda Introduction Overview of Global Diagnostics Forecasting Overview of Country Forecasting Methodologies Forecasting Example: EID Commodities FORLABS Introduction Recommendations for Future Diagnostics Forecasting Appendix 21
A more standardized and automated forecasting could facilitate access to lower prices of diagnostics Tes ng & Commodity Forecas ng Online Market Dashboard Produc on Planning and Market Strategy FORLABS Data Collected New Products/ Pricing Outcome Laboratory network Types of labs Loca ons of labs Instruments per site Lower Manufacturing Costs (Lower Prices) Earlier Entry of New Diagnos cs Tests and Prices Market Intelligence Demand and actual tes ng volumes Commercializa on & Regulatory Tests run per site Price per test/ commodity by health facility level by test type by target pa ents by sector (pub/priv) Distributors Supply chain Procurement prac ce Regulatory pathways Outcome Manufacturers/Donors Market Value Market Needs Market Opportuni es Capacity and Produc on Inform expected revenues and market value for exis ng and new products Inform product design with key market needs and necessary tes ng capabili es Inform the device s relevance to the context Inform produc on and capacity planning through transparent demand pa erns Higher Transparency and Predictability Intelligence
Other Key Recommendations for Future of HIV/TB Diagnostics Forecasting and Procurement 1 Increase use of test-based commodity calculators: Molecular diagnostic tests require dozens of commodities. Countries should be able to leverage tools that translate the number of tests desired into the appropriate number of specific commodities to order. 2 Integrate Forecasting into LIMS: Lab Information Management Systems (LIMS) can integrate service delivery and consumption data to produce more accurate and efficient forecasts. 3 Encourage use of consumption data in forecasting: Where possible, consumption data should be used to calibrate morbidity and target-based forecasts against proven demand. 4 Promote use and development of standardized, web-based forecasting tools: Webbased forecasting tools can capture consumption data in real-time from separate labs, generating a more accurate understanding of demand. Where possible, countries should adopt forecasting tools that aggregate demand for multiple diagnostic tests. 5 Improve integration of forecasting across test types and disease areas: To take advantage of economies of scale as well as the integration of testing platforms, countries should move towards a more holistic approach towards forecasting for HIV/TB diagnostics.
Conclusion Questions?