Clinical In-Market Leading Indicators for Brand Performance

Clinical In-Market Leading Indicators for Brand Performance
Nitin Choudhary, Principal, Symphony Health Solutions; Ewa Kleczyk, PhD, Vice President of Client Analytics, Symphony Health Solutions; Rajkumar Rajabathar, Engagement Manager, Symphony Health Solutions
Abstract:
In today’s world, there are multiple treatment options available for overall care of a patient in a disease state. For a pharmaceutical company, the need for effective product based strategy has gained importance. Traditionally, strategies have been derived from knowledge on disease states and focused on key stakeholders in a patient’s journey. However, in many cases, there are key in-market factors that significantly influence a particular treatment choice. The patient’s continuum of care, condition, nature of disease, therapy administration method, safety, efficacy, side effects, associated comorbidities, and concomitant medications play a role in selecting a particular treatment choice or switch. Additionally, evolution of therapies in the market, product adoption, and physician prescribing behavior, insurance coverage, and promotions are key considerations informing brand strategy.

This paper outlines a two-step approach to develop a cost-effective brand strategy, illustrated by a case study. First, using patient, physician, and payer information we explore significant drivers and barriers for brand initiation and continuation. Second, using a finite list of drivers and promotions data we identify and track leading indicators that can explain significant changes in brand performance. This can provide actionable levers for sales, marketing, market access, clinical and brand teams. Such levers can help interpret implications to forecast, enhance messaging, and optimize allocation of resources to derive more value. As a result, pharmaceutical companies can develop and ultimately execute a more product and patient centric, data-driven brand strategy.


Keywords: leading indicators, brand strategy, market diagnostics, statistical modelling, healthcare claims

Introduction
A leading indicator is a measurable economic factor that changes before the economy starts to follow a particular pattern or trend. Historically, leading indicators are used to gain some sense of which way the economy is headed. Investors use such indicators to adjust their strategy to benefit from future market conditions.1 Unemployment rates and inflation2 are key economic indicators that affect markets.3 This technique has also been successfully applied in retail scenarios. Pinterest is an example of a leading indicator, the number of pins steadily increase as users are discussing within the social network and become engaged with an event or holiday.4 Retailers who have periods of peak demand during holidays can predict performance using such an approach. Economic leading indicators such as state of the economy can help explain consumer buying power within a target segment.
 
The situation is no different in the pharmaceutical world, where early warning signs can be used to identify brand performance changes prior to actual change occurrence.5 Pharmaceutical companies can use such early warning signs or leading indicators to develop and course correct brand strategy. However, predicting the future requires understanding of the past. In an industry with rich physician and patient level longitudinal data, leading indicators can be identified using a data-driven approach. This paper discusses a quantitative approach leveraging statistical modelling techniques to identify and track leading indicators.
 
Key Considerations for Brand Strategy
In a competitive marketplace, a cost effective, product-centric brand strategy6 requires tackling strategic questions, such as:
  1. How to develop a more effective brand plan?
  2. How to better utilize or understand disease state, patient behavior, and physician preferences?
  3. How to improve on commercial assessment methodologies?
  4. How to differentiate the brand in the market?
  5. How to improve market access for brand?
  6. How to optimally allocate promotional spend across channels, target audience and geographies?
In order to address each strategic question, it boils down to gaining a better understanding of the brand in the market place. Interactions between physicians, patients, payers and treatment choices drive the evolution of the market for a particular disease state.7 Understanding market dynamics can help identify leading indicators to quantify key dynamic questions such as:
  1. How many patients are new to market, switch treatments and discontinue?
  2. What is the treatment duration for each brand?
  3. What are the key patient characteristics that are drivers and barriers for new adoptions and continuations of a brand in the market?
  4. What are the significant, differentiating factors driving a particular brand’s starts and stops as compared to competitors in the market?
  5. What are the key differentiating leading indicators to predict brand volume?
Such market insights can have implications across a pharmaceutical company, providing levers for sales, marketing and market access teams to act upon.8,9,10 Figure 1 summarizes market insights from the case study and provides a conceptual framework of strategic and tactical implications.
Figure 1: Conceptual Framework of Implications Across a Pharmaceutical Company
SALES MARKETING MARKET ACCESS
STRATEGIC IMPLICATIONS Design or inform high value targeting strategy

Optimize allocation of resources
Design and deploy marketing campaigns specific to target audience Inform market access strategy with MCOs
TACTICAL IMPLICATIONS Identify patient mix for physicians to evaluate value of physician-rep interaction

Communicate insights and train reps on messaging to specific type of physicians

Cues for reps based on key moments of a patient journey
Design marketing tactics for targeting specific sub populations

Identify messaging and channels for optimal reach

Design scenarios for continuous improvement and return on investment
Identify plans with high rejections and reversals

Deploy educational programs around cost and coverage

Identify reasons for rejections and reversals
INSIGHTS New Initiations Continuations
Patients on commercial plans are more likely to initiate on product A

Patients with Medicare plans are less likely to initiate on product A

Patients between the age of 55 and 84 are likely to initiate on product A

Patient taking medication X are more likely to initiate on product A

Patients in middle income groups are more likely to initiate on product A

Patients with Y medication are less likely to initiate on product A
Affordability is a key factor in patient continuations

Patients on product A for more than 5 months are more likely to continue for 12 months

Side-effects are not a key factor in patient discontinuations in this disease state

Lower income groups are likely to switch away from product A

Patients with product A reversed claims are more likely to switch to product C

Patients less than the age of 65 are more likely to switch to product B

Patients with higher CV risk are more likely to switch away from product A
Case Study
Introduction
A retrospective study consisting of approximately 200,000 patients initiating a treatment in a market were studied over a one year period ending October 2016. The market was defined based on three products – product A, product B and product C. Product A is of interest and product C owns the majority of the patient share in the market. As illustrated by Figure 2, the majority of patients are new to market.  Switches away from product A to product C was a concern. In this study, physician preferences and promotions were considered to have minimal impact to the product losing share and thus, were not included in the analysis.
Figure 2: Product Initiation and Discontinuation Cohorts for Observation


Discontinuation cohorts were further differentiated based on duration of therapy before discontinuation. Figure 3 illustrates durations of therapies on product A prior to discontinuation or switch. Most patients on product A switch or discontinue within one month of product initiation. Comparing cohorts based on duration of treatment can help understand key barriers and drivers for continuations. Some patients continue for more than 12 months whereas others discontinue after a few months on the product.
Figure 3: Product Continuation Cohorts for Observation


Figure 4: Cohort Analysis Methodology
Analysis
A retrospective analysis was conducted to compare new, continuing and discontinued patient cohorts. From longitudinal historical claims data, dimensions such as patient demographics, plan, diagnosis, other medications, and procedures11,12,13 were studied 90 days prior to an initiation or discontinuation of product A. Analyzing patient cohorts helps uncover market dynamics. Six cohorts were analyzed:
  1. New To Market (NTM) initiation of product A vs. product B
  2. NTM initiation of product A vs. product C
  3. Continuation of product A for less than a month vs. 12 months
  4. Continuation of product A for 1-5 months vs. 12 months
  5. Continuation of product A for 6-11 months vs. 12 months
  6. Discontinuation of product A vs. switch to product C
Over 850 variables were developed to understand patient and payer dynamics between six cohorts of patients treated with product A. Univariate and bivariate analysis was conducted to assess validity of the data variables by ensuring at least 10% of the data is populated and to remove correlated variables. A shortlist of 50 features were fed into a logistic regression model. The end output was approximately 25 significant features (based on p-values) with higher odds ratios that are drivers or barriers for new initiations and continuations. Figure 5 illustrates odds ratios for significant patient demographics features (p value < 0.01) and high odds ratios. This helps identify that if a patient is new to market and over 85 years of age, that patient is more likely to initiate on product A as compared to product B.
 
Figure 5: Significant Patient Demographics Features Differentiating Between New to Market Patients Initiating on Product A (Odds Ratios >0) and Product B (Odds Ratios <0)


Analyzing patients continuing on product A for one month of initiation are significantly different from patients continuing for 12 months. As illustrated in Figure 6, rejected and reversal claims, mid-income groups, and Medicare plans characterize patients discontinuing product within one month of initiation.
Figure 6: Significant Patient Features Differentiating Between Patients Discontinuing Within 1 Month of Initiation (Odds Ratios >0) and 12 Months of Initiation (Odds Ratios <0)


Analysis of six patient cohorts provided a comprehensive list of significant patient features affecting market dynamics.
  1. Driving new initiations:
    1. Patients on commercial plans are more likely to initiate on product A
    2. Patients with Medicare plans are less likely to initiate on product A
    3. Patients between the age of 55 and 84 are likely to initiate on product A
    4. Patients on X medication are more likely to initiate on product A
    5. Patients in middle income groups are more likely to initiate on product A
    6. Patients with Y medication are less likely to initiate on product A
  2. Promote continuance
    1. Affordability is a key factor in patient continuations
    2. Patients on product A for more than 5 months are more likely to continue for 12+ months
    3. Side-effects are not a key factor in patient continuations in this disease state
  3. Arrest switches:
    1. Lower income groups are likely to switch away from product A
    2. Patients with product A reversed claims are more likely to switch to product C
    3. Patients less than the age of 65 are more likely to switch to product B
    4. Patients with higher CV risk are more likely to switch away from product A
Leading Indicators
Significant patient features were used to predict new and continuing patient volume depicted in Figure 7. About 25 significant features were derived from the driver analysis. Multivariate time series regression models can be used to predict time dependent variables such as prescriptions. A prediction model for new prescriptions and continuing prescriptions was built.
Figure 7: New and Continuing Patient Trend for Product A


Figure 8: Leading Indicators Methodology


Figure 8 illustrates the overall analysis methodology for leading indicators.14 A 5-step approach was used to select a final model for identification of leading indicators. Using patient features as inputs, a multivariate regression model was built on 70% of the data and validated on 30% of the data. Two models were built—one to predict projected new prescriptions and another to predict continuing prescriptions. The end output of the model provides coefficients of each lead & lag variables based on the output equation. The coefficients of the equation quantify the impact of each variable to prescriptions. The models were more accurate in predicting continuing prescriptions; the out-of-sample error for new and continuing models were 4% and 1.5% respectively. Adjusted R-squared values were 0.63 for new initiations and 0.83 for continuing models. The equation of the new prescriptions prediction model:
 
Product A NRx = Intercept + 0.14*(DX_X_12) – 0.112*(DX_Y_12) – 0.04*(TX_Z_16) + 0.7 (seasonality component)
 
DX_X_12 = Number of patients diagnosed with X, 12 weeks prior
DX_Y_12 = Number of patients diagnosed with Y, 12 weeks prior
TX_Z_16 = Number of patients initiating treatment Z, 16 weeks prior
 
For new patients, some examples of leading indicators for product A are:
  1. Population dynamics such as age, gender, geography and income
  2. Diagnosis of certain comorbid conditions
  3. Initiation of concomitant treatments
For continuing patients, some examples of leading indicators for product A are:
  1. Medicare and Medicaid plans
  2. Rejections and reversals
  3. Patient severity, certain procedures administered
Such a data driven modelling approach validated key business hypotheses pursued as part of brand strategy. Leading indicators also provide a conceptual framework to develop, execute and incorporate feedback in brand strategy. Brand strategy can be collaboratively executed through multi-channel tactics.
Results
Outputs from the statistical model were used as levers to inform brand strategy. Figure 9 illustrates examples of clinical leading indicators and their impact on brand performance. Diagnosis and treatments unrelated to the disease or product of interest were found to have a positive and negative impact on performance. Such indicators facilitated sales & marketing activities by enhancing product specific messaging.
Figure 9: Clinical Leading Indicators for Initiation of a Brand Based on Patient Volume and Impact on Brand Performance
Next Steps
Apart from patient features, there are several other factors that can be leading indicators for the brand. Physician preferences based on prescribing behavior and promotional spends affect brand performance. Figure 10 illustrates a more comprehensive set of variables that can be leveraged as leading indicators. Machine learning can also be used to train models for better accuracy.
Figure 10: Comprehensive Set of Features to Consider for Leading Indicators/
About the Authors
Rajkumar Rajabathar (RB), an engagement manager with Symphony Health Solutions, uses his 7 years of experience in providing high impact consulting services for pharmaceutical companies. He is passionate about using his knowledge of business, math and data & technology to solve complex business problems through a client-centric approach. With expertise from leading analytics for innovative growth strategy projects in the areas of: closed loop marketing strategy; improving R&D productivity through disparate patient data such as claims, EMR, prospective & retrospective clinical trials data; interactive analytics capability for commercial assessments; sales, marketing effectiveness & incentive compensation. With his experience, he is well positioned to guide his teams through strategy & analytics projects executed within the commercial effectiveness team. His deep knowledge of industry & analytics also enables him to develop new offerings for the team. In his prior role at Mu Sigma, he was recognized through impact & spot awards. RB holds a bachelor’s degree in Aerospace Engineering from Indian Institute of Technology, Madras.
 
Nitin Choudhary, a Principal with Symphony Health Solutions, uses his 11 years of experience in providing analytical consulting services to create effective and innovative solutions to problems faced by the Commercial Effectiveness team. With expertise in leveraging patient level integrated data to solve problems in fields like Advanced analytics, Management Science, Field Analytics, Brand & Data strategy, Managed markets & Contracting, Reporting, and KPI & Dashboards, Nitin is well prepared to guide teams through the complex pharmaceutical market. His knowledge of the industry is demonstrated through accomplishments such as Symphony Consultant of the year, industry presenter at PMSA conference, and Innovator of the Year from Cognizant. Nitin holds a Masters degree in Mathematics and Computing from the Indian Institute of Technology Delhi.
 
Ewa J. Kleczyk, PhD, is a Vice President of Client Analytics with Symphony Health Solutions in Conshohocken, PA and an Affiliated Graduate Faculty for the School of Economics at the University of Maine in Orono, ME. She also holds a doctorate degree in Economics from Virginia Tech. Ewa leverages over 12 years of industry experience and a strong passion for healthcare to develop key insights from healthcare claims data. Going forward, she is excited about the new and expanded role of machine learning in providing more exact and refined optimization of sales and targeting strategies and resources, as well as their deployment timing. In the future, Ewa believes that past and future legislation will ultimately result in patients having greater involvement in their healthcare decisions, leading marketers to change the way they approach patients and the information they make available to them.
 
 
References
1 Investopedia (IAC). Definition of a leading indicator. https://www.investopedia.com/terms/l/leadingindicator.asp
2 Philip. A. Klein. Leading indicators of inflation in market economies. International Journal of Forecasting. Vol. 2, Issue 4, 1986.
3 Investopedia (IAC). 4 key economic indicators that move the market by Brian Perry (Nov. 2017). https://www.investopedia.com/articles/fundamental-analysis/10/indicators-that-move-the-market.asp
4 American Marketing Association (AMA). How to use leading indicators to predict success and course correct for your brand by Jon Webb (May 2017). https://www.ama.org/publications/eNewsletters/Marketing-News-Weekly/Pages/predictive-insights-lead-indicators.aspx
5 Symphony Health Solutions. Leading indicators of brand performance (May 2013). https://symphonyhealth.com/wp content/uploads/2013/05/leading_indicators_sales_sheet.pdf
6 Boston Consulting Group (BCG), 5 ways to win commercially in pharmaceutical markets (Dec 2011).  http://img-stg.bcg.com/BCG_Brand_Renaissance_Dec_2012_tcm9-108418.pdf
7 Mark Paich, Corey Peck, Jason Valant. Pharmaceutical market dynamics and strategic planning: a system dynamics perspective. System Dynamics Review Vol 27, Issue 1, Jan 2011.
8 Cognizant. Key Indicators: An Early Warning System for Multichannel Campaign Management (2013). https://www.cognizant.com/whitepapers/Key-Indicators-An-Early-Warning-System-for-Multichannel-Campaign-Management.pdf
9 PMLive. Pharma needs to evolve from customer centricity to stakeholder centricity (June 2017). http://www.pmlive.com/pmhub/healthcare_creative_and_design/owen_health/white_papers_and_resources/pharma_brand_strategy_2.0
10 PWC Strategy&. A critical makeover for pharmaceutical companies: Overcoming industry obstacles with a cross-functional strategy (Jan 2017). https://www.strategyand.pwc.com/reports/critical-makeover-pharmaceutical-companies
11 McKinsey and Company. How big data can revolutionize pharmaceutical R&D (Apr 2013). https://www.mckinsey.com/industries/pharmaceuticals-and-medical-products/our-insights/how-big-data-can-revolutionize-pharmaceutical-r-and-d
12 Forbes. How big data is changing healthcare (Apr 2015). https://www.forbes.com/sites/bernardmarr/2015/04/21/how-big-data-is-changing-healthcare/
13 Forbes. Data Silos: Healthcare’ silent shame (Mar 2015). https://www.forbes.com/sites/davidshaywitz/2015/03/24/data-silos-healthcares-silent-tragedy/#244b9c4c2907
14 Fifth Joint European Comission (EC). Those unpredictable recessions by Sergey. V. Smirnov. https://www.oecd.org/std/leading-indicators/48985179.pdf