Create Model Forecasts for Data- Based Decision Making Patrick J. Park, PharmD, MBA Director, Forecasting & Financial Planning Daiichi Sankyo, Inc.
Disclaimer The views and opinions expressed in this presentation are those of the presenters and should not be attributed to Daiichi Sankyo, Inc.
Agenda What is Forecasting? Data Sources for Forecasting Rx Based Forecasting Model Demand Data Ex-Factory Data Inventory Data Patient Based Forecasting Model Market Research Data Probabilistic Forecasting Monte Carlo Simulation Communication
What is Forecasting? to calculate or predict (some future event or condition) usually as a result of study and analysis of available pertinent data - Marriam-Webster s Learner s Dictionary to indicate as likely to occur - Marriam-Webster s Learner s Dictionary Process of gathering information and getting alignment with stakeholders Support decision making process As soon as a forecast is finalized, it is wrong!
Forecasting Method Rx Based Forecasting Often used for in-line products Based on historical data reported by the audits Baseline projection - Trx/Nrx/retail/non-retail Ex-Trend event product life cycle event, promotion impact, market events, market access Patient Based Forecasting Often used for pipeline products with limited history or market Model the entire market for a specific disease state Determine the target population based on product profile
Rx Based Forecasting Using historical data (market/product Trx/Nrx/etc) to create baseline forecasting Audit data (IMS/Symphony) Create Ex-trend events Based on input from marketing/sales/market research Reconciliation/Conversion Inventory data Ex-factory data
Audit Data IMS/Symphony provides Trx/Nrx data based on their pharmacy audit Retail and Mail order pharmacy prescription activity updated weekly/monthly By product/therapeutic class/ payment type/ physician specialty Non-retail data that includes activity from hospital, clinic, and long term care Ex-Factory Data Products shipped from Wholesaler to retail/ nonretail outlets Manufacturer Wholesaler Hospital/ Pharmacy Demand Data captured by IMS/Symphony Patient
Baseline Projection Actual Forecast Market Events Historical trend (both market and product) Historical market events/timing and any inflection point Historical seasonality Future market/product events Additional competitors LOE events
Baseline Projection Many statistical method to use to provide baseline projection Baseline projection can differ based on historical data points inclusion Forecasters insights into the product/market will determine what to include to create the baseline projection Work closely with the stakeholders to pick the most appropriate data points
Ex-Trend Events Consider any market events and product events Competitor entry/ product LOE, etc
Inventory Data and Conversion Bridging Ex-factory and Demand data 852/867 Inventory data Provide data set for manufacturer can use to determine the amount of inventory is in the retail/wholesale channel Baseline Projection Market Events Retail Demand Non Retail Demand Total Demand Gross Sales Ex-Factory Inventory
Patient Based Forecasting Market Research can help provide a lot of data for forecasting input Quan/Qual Market Research Various Database/Reports/ Literature Key Opinion Leaders
Patient Based Forecasting US Population in 2017: (325M) >18yo: 75% (246M) Source: US Census Quick Facts Prevalence: 2.4% (5.9M) Source: Decision Resources / Market Research
Patient Based Forecasting Prevalence: 2.4% (5.9M) Source: MR Diagnosed Moderate: 42.5% (2.5M) Diagnosed Severe: 12.5% (743K) Treated 70% (1.7M) Treated 90% (669K) Treatment 1: 32% Treatment 2: 5% Treatment 1: 39% Treatment 2: 5% Treatment 3: 13%
Patient Based Forecasting Source: US Census Quick Facts Source: Decision Resources/ MR Source: MR Source: Decision Resources/ MR Market Access: 90% Patient Share: 14% Adherence/Compliance: 64% Days of Therapy: 26 days
Forecast Summary
Risk Assessment Assumptions are best guesses Consensus on assumptions are agreed on by stakeholders, but there are other possibilities Leads to many scenarios based on different assumptions what-if analysis Adjust a few assumptions to see what the impact would be Leads to many different variations Communication can be difficult Base case/ Upside/ Downside Pick top few scenarios from what-if
Forecast Assumptions Assumptions Base Case Range Source 1 Launch Date 2017 Assumption 2 Expiry Date 2029 Assumption 3 Prevalence Rate 2.4% Decision Resources 4 Diagnose Rate 55% 50% - 75% Decision Resources 5 Drug Treated Rate 80% 80% - 85% Decision Resources 6 Market Access 90% 7 Moderate Patients Treatment 1 32% MR Treatment 2 5% MR 8 Severe Patients Treatment 1 39% MR Treatment 2 5% 5%-30% MR Treatment 3 13% 10%-40% MR 9 Patient Share % Increase 50% 20% - 50% MR Peak Share 14% Derived - Forecast Pro 10 Market Expansion % 10% 5% - 15% 11 Persistency 83% Decision Resources 12 Compliance 91% Decision Resources 13 Price/Yr $10K 14 Annual Price Increase 5% 3% - 8% 15 GTN 80% Peak Sales $559M
Forecast Assumptions Assumptions Base Case Range Source 1 Launch Date 2017 Assumption 2 Expiry Date 2029 Assumption 3 Prevalence Rate 2.4% Decision Resources 4 Diagnose Rate 55% 50% - 75% Decision Resources 5 Drug Treated Rate 80% 80% - 85% Decision Resources 6 Market Access 90% 7 Moderate Patients Treatment 1 32% MR Treatment 2 5% MR Monte Carlo Simulation can be performed for the range of assumptions 8 Severe Patients Treatment 1 39% MR Treatment 2 5% 5%-30% MR Treatment 3 13% 10%-40% MR 9 Patient Share % Increase 50% 20% - 50% MR Peak Share 14% Derived - Forecast Pro 10 Market Expansion % 10% 5% - 15% 11 Persistency 83% Decision Resources 12 Compliance 91% Decision Resources 13 Price/Yr $10K 14 Annual Price Increase 5% 3% - 8% 15 GTN 80% Peak Sales $559M
Probabilistic Forecast Using value ranges for multiple assumptions and generate output Technique used to understand the impact of risk and uncertainly in assumptions Probabilistic Forecasting helps decision making and any analysis In Monte Carlo simulations, it can run through over 1,000 scenarios and create a range of outputs
Probabilistic Forecast Monte Carlo Simulation on net Sales based on range of inputs on multiple assumptions It will provide probability of net sales range based on the range of assumptions 80% chance that the Net sales will be between $2.5B to $3.9B
Probabilistic Forecast Tornado chart and Sensitivity Analysis Informs us which assumption has the most impact on the forecast Annual price increase has the highest impact on net sales Market expansion has the least impact on net sales
Probabilistic Forecast Creates discussion around the assumptions that has the most impact Be able to identify if any of the assumptions require additional research Adjust range of inputs Eliminate assumptions New assumptions Drives more productive discussion on forecasting assumptions and lead to better decision making Be able to develop more informed forecast
Probabilistic Forecast Opportunity Creates discussion around the assumptions that has the most impact Be able to identify if any of the assumptions require additional research Adjust range of inputs Eliminate assumptions New assumptions Drives more productive discussion on forecasting assumptions and lead to better decision making Be able to develop more informed forecast Challenges Forecast needs to be a number/ target not a range
Forecasting Assumptions 2013 2014 2015 2016 2017 Actual Actual Forecast Forecast Forecast Gross Sales $100M $120M $130M $140M $150M Net Sales $75M $84M $84M $89M $92M Market Growth +5% +7% +8% +8% +8% Market Share 15% 16% 17% 18% 18% TRx Growth +5% +8% +9% +10% +10% Price / Unit $5.12 $5.63 $6.20 $6.81 $7.50 WAC Price Increase Apr 5%; Sep 5% Apr 5%; Sep 5% Apr 5%; Sep 5% Apr 5%; Sep 5% Apr 5%; Sep 5% GTN% 75% 70% 65% 165% 265% A&P $20M $22M $23M $24M $25M Sales Cost $30M $33M $35M $37M $38M
Forecasting Assumptions 2013 2014 2015 Actual Actual Forecast Product A TRx 1.2M 1.26M 1.36M Product Growth (YoY) 2% 5% 8% CVS Health T3 T2 1 of 3 T2 1 of 3 ESI T3 T3 T2 1 of 2 UHC / OptumRx T3 T2 1 of 3 T2 1 of 2 Prime T3 T3 T3 Aetna T3 T2 1 of 3 T2 1 of 2 WAC Price Increase Apr 5%; Sep 5% Apr 5%; Sep 5% Apr 5%; Sep 5% GTN% 75% 70% 65% Gross Sales $100M $120M $130M Net Sales $75M $84M $84M Validate assumptions and get an alignment from the team Build a story to explain changes in assumptions
Net Sales Bridge
Summary As soon as a forecast is finalized, it is wrong! Forecasting is science and art Communicate on changes and assumptions Get an alignment with the stakeholders Be a story teller
Questions?