2016 Annual Conference

Forecasting in the pharmaceutical markets always has been challenging, and the environment in the next 5-10 years will present even greater hurdles to accurate forecasts. The advent of biosimilars, the consolidation of payers and providers, the ongoing effects of the Affordable Health Care Act, and the focus on personalized medicine are but a few of the challenges facing forecasters. This tutorial will present techniques that forecasters can use to model out the effects of these dynamics. We will discuss the overlap of best practices in forecasting with the tools of decision analysis to create more informed, robust and defensible forecasts.

This tutorial will build on the PMSA Virtual University series on Sales Planning and Operations, but will go into more depth, providing more detail and examples, and hands-on exercises. The goal will be to provide a solid foundation in basic Sales Planning analysis, particularly along the continuum that runs from raw data (usually Rx) through segmentation, promotion response, targeting, and sizing of sales forces. Call Planning and Alignments will be touched on lightly, but are not the primary focus of this tutorial. Attendees should be able to fully understand and participate in such analyses by the end of the tutorial, and understand the strengths and weaknesses of various analytical approaches.

It is the best of times, it is the worst of times…it is the summer of analytics, it is the winter of challenges to the success of analytics. This presentation frames a number of the challenges facing pharma commercial analytics teams, and takes a position as to how to maximize the value and staying power of commercial analytics teams.

Daniel Kinney leads Global Business Intelligence at Sanofi, driving data strategy, leading relationships with key Sanofi data partners, and driving business results through high-impact business intelligence offerings. Prior to Sanofi, Danny worked at IntegriChain, where he led Information Services and Advanced Analytics, and Bristol-Myers Squibb, where he led Worldwide Business Intelligence. Over the course of his 15 years in Pharma, Danny has led teams responsible for a broad range of activities, including secondary data acquisition, management, and reporting, advanced analytics, salesforce planning and analysis, managed markets, and competitive intelligence. In his previous life, Danny started his career as an Assistant Economist and Analyst at the Federal Reserve Bank of New York. Danny earned an MBA in Finance and International Business from Columbia Business School, and an AB in Economics cum laude from Princeton University.

This presentation will outline two novel approaches for forecasting uptake of a new product or therapeutic class. These proposed approaches challenge industry status quo methodology, and may improve forecast accuracy and generate additional insights into product uptake.

The first approach advances traditional methods by utilizing more robust models than the traditional analogue-based forecast model. In addition, cross-country product uptakes are analyzed and country-level uptakes of new pharmaceutical therapeutic classes are contrasted with in-country uptakes of new technologies over the past 200 years.

The second approach uses “big data” analytical methods to estimate the effect of prescriber influence networks on product uptake. A system of partial differential equations is used to model the effect of prescriber influence networks and prescriber similarity on adoption. This model utilizes anonymized patient-level data (APLD), Medicare referral data, and prescriber profile data to identify adoption “tipping points” and generate incremental improvements to new product forecasts.

The evolving treatment paradigm in Oncology has changed understanding brand and regimen performance relative to mass market therapies. Momentum is shifting to genotype and biomarker-based identification. Treatment decisions can be monotherapy or combination, using infusion or oral administration. Understanding sequencing and its rationale is also unique and challenging. And while Oncology data has progressed, it is increasingly challenging to build robust metrics as therapies are built for smaller targeted patient populations.

In this session, we aim to move from the thinking oncology data in in its common silos of “Research Strategies” and “Data Purchasing Strategies” to brand-centric Information Strategies that enable a much richer view of performance. Across silos of knowledge and information that exists today is an opportunity to better see and integrate across the full range of data available, and serve it back to the organization to enable better decisions for competitive advantage.

This session will offer:

• Perspective on the development of information strategies for better decision-making
• Discussion around formulating KPIs that offer impactful performance measurement
• A framework for considering the pros and cons of various data sources
• Examples of how to bring it all together and the change management required to achieve it

The wealth of data and information about disease state awareness available to patients has increased the pressure to take control of ones’ self-health. Conversations and follow ups from the exam rooms aren’t always translating to persistency to medication and we continue to see patients ending up with unexpected outcomes. Patients who are persistent typically achieve better outcomes. Identifying persistent new-to-brand patients helps pharma and providers appropriately target patients to increase the probability of achieving successful outcomes as well as brand loyalty. We present a new process that enhances the formulation and solution approach in identifying the lead/lag indicators by exploring the clinical presentation in the exam room (utilizing EHR data) to predict patient-level persistency (utilizing linked claims data) and to continuously monitor the patient response in achieving persistency. The essence of our approach utilizes the dependent variable present in claims data and the independent variables being drawn from clinical (EHR) data. The variation in persistency is captured through examining ongoing clinical evidence not available in claims data. This allows for a dynamic clinically-based analysis of multiple factors influencing persistency on an ongoing basis, replacing the more common static evaluation.

When performing physician segmentation, there are two primary ways of assigning physicians into segments in order to inform marketing and sales strategies: 1) Primary Market Research approach that leverages survey attitudinal information related to the treatment pathways and product perceptions and 2) Secondary Data Market Research that leverages physician prescribing behavior and / or patient level data information.

Symphony Health Solution implemented a hybrid approach for segmentation that allows for combining the attitudinal based segmentation developed via Primary Market Research with secondary physicians and patient level data in order to project segment assignments for all physicians of client interest without the need for a typing tool.

The objective of the PMSA presentation is to introduce the ‘Hybrid segmentation’ approach that allows project physician segments to a larger physician list via leveraging the attitudinal information, robustness of secondary data, and proprietary physician level prescribing propensities. The proposed presentation will review methodology steps needed for ensuring the “Hybrid’ approach yields high accuracy prediction and consistency in assigning physicians into indented segments. In addition, a case study that leveraged the methodology in the Oncology area will be reviewed to validate and predict attitudinal segment to universe of Oncologists.

We know many factors explain sales performance in a particular geography. Some factors are in control of sales & marketing teams, while others are underlying conditions that cannot be changed. To enable improved local market-based sales planning, we integrated various data sources into a model that identified key drivers of performance, and, given particular conditions, what tactics might be best suited for sales teams to drive performance.

Many variables were considered – personal and non-personal promotion, local presence of KOLs, payer and provider mix and local concentration, rep access restrictions – and those for which data were available were analyzed via CART analyses. The results were presented to local sales leaders with advice on local tactical planning.

For example, for one product, we observed significantly higher share in geographies with a disproportionate amount of business flowing through commercially self-insured healthcare plans (relative to fully-insured plans). It was determined that risk aversion and cost sensitivity of self-insured plans are likely driving preference to the relatively cost effective brand. Tactically, local sales managers can adjust planning to ensure reps understand the managed care landscape and ask the right questions with office staff pertaining to patient populations.

Deep learning has become a breakthrough resource across financial, communication and online consumer industries – for example, it underpins new tools that allow Spanish and English speakers to communicate in real time via automated translation, and is defining the future of driving through driverless cars. This talk will describe how deep learning, often referred to as a form of 'artificial intelligence', is increasingly also being applied across healthcare industries. For example, deep learning technologies are being used to help medical professionals distill insights from across medical imaging, genomics, pathology and pharmacokinetic data.

Lina Nilsson is vice president of strategy and operations at Enlitic, the medical company founded to dramatically improve diagnostic outcomes for billions of people around the world using deep learning technologies. Most recently, Lina was at the University of California, Berkeley, where she helped design and implement a $25M initiative to more efficiently spin out new social-impact technologies from the University. In her role as the Innovation Director at one of UC Berkeley's largest interdisciplinary centers, Lina supported a portfolio of over 90 technology projects in medicine, infrastructure, and communications, across 40 countries. Before this, Lina worked on CellScope, a novel cellphone-based technology for automated disease diagnosis. Lina has been a member of the Global Health Advisory Board for Investors' Circle and she is the founder of Tekla Labs, an open-science hardware collaborative. Lina has been recognized on MIT Technology Review's "TR35" annual list of the world's top 35 innovators under the age of 35, and her writings have been published in the New York Times, Washington Post, Science, and Make Magazine. Lina is a biomedical engineer by training and holds a doctorate from ETH Zurich.

In oncology, targeted therapies work by targeting the tumor’s specific genes, proteins, or tissue environment that may have been responsible for the cancer’s growth. As these are targeted narrowly, they can limit the damage to normal cells. Targeted therapies have generated great interest amongst pharmaceutical manufacturers and several products are currently being commercialized. The existing cancer therapies are targeted to patients whose tumors have a particular mutation such as EGFR, HER2, BCR-ABL, ALK or BRAF inhibitors, which has proven to take a front seat in many research and development activities.For pharmaceutical companies, it is important to understand the potential value of purchasing this data and building capabilities around it. This presentation evaluates the possibilities of leveraging diagnostic test data, based on an engagement that we explored by building capabilities around these data-sets and using it to enhance understanding of diagnosis, testing, and treatment behavior. In this presentation, we will demonstrate how to effectively use diagnostic testing data, as well as share our findings on its capabilities in forecasting and other commercial and clinical applications.

Our industry is awash with data but that does not mean we can find a database for every business question that comes our way. Many a times, we are stumped. We wonder if the database for the job ever exists. In many cases, the database does not exist, which is not that surprising after all. In several other cases, though, we admit defeat when we should not: the database does exist. Why did we fail to identify the database? The answer is that we limit our search to the familiar, just like the person who looks for the key under the proverbial lamp, not where he lost the key.

The aim of this talk is to redress this unfortunate situation. We'll start off by taking a close look at frequently asked questions around hospital and group practice affiliation, competitive activity, and Medicare reimbursement, and explain how to address these questions. We'll review a slew of databases that are becoming fixtures of our industry, thanks to the Freedom of Information Act. We'll not only identify which databases to deploy, but also discuss how they should be combined. We'll talk about caveats, as there are always caveats with any database. The best part about these databases is that they are free and contain powerful information. We did look the gift horse in the mouth and are delighted to report it is a pretty sight indeed!

The Holy Grail of promotion optimization is the timely delivery of the profit maximizing level of promotion through the right mix of channels with the most responsive content. While every marketer, sales manager and management scientist has thought through (i.e. dreamed of) such a systematic approach, pharmaceuticals lag behind other industries in implementing this ideal. This presentation focuses on the element of this system that has received the least attention, but has the greatest potential to improve marketing /sales relevance and performance—message optimization across multiple channels.

Though messages are shared through multiple channels, generally, learnings about HCP message preference are lost because responses are not collected, analyzed and then used to increase content relevance in the next campaign.

This presentation describes what field-collected message information can teach us about doctor-level message preference and how this information can be combined with content response from other channels for the purpose of:

• Learning which messages drive Rxs most.
• Beginning to understand how to construct a message preference history for each physician.
• How to develop a coordinated differential messaging strategy across channels for each doctor.

We will also note the significant impact that this information can have upon analysis done outside the U.S. The presentation will also discuss the value of collecting rep reported HCP message satisfaction data and how this can be the new “doctor level data,” supplementing HCP Rx data. Moreover, the impact that this information can have upon analysis done in other countries at the doctor level will be discussed.

Despite payer complaints, co-pay discount cards have become ubiquitous in the pharmaceutical industry with the vast majority of branded therapies operating some sort of coupon program. Partly in response to this, payers have instituted increasingly aggressive controls on drugs in competitive categories. At the same time, patient enrollment in plans with either high out-of-pocket co-insurance tiers or high deductibles has grown dramatically. This combination of trends is leading many brands to spend significantly more money on both their co-pay card programs and their managed care rebates to maintain access to patients. With co-pay card program design and individual payer contracting decisions evaluated separately and by different parts of an organization, the interaction between these two types of discounts is often not considered carefully. A modeling framework and approach that integrates the different discounts from a brand’s gross price to its net price is required to ensure profitable decision-making.

In this presentation, we will describe the multiple discounts issue and highlight situations where pharmaceutical manufacturers are at a heightened risk of making unprofitable decisions. We will also describe an enhanced modeling framework that can enable better aligned decision-making around discounting strategies for payers and patients. Finally, we will describe how brand teams, managed markets organizations and analytics leaders can work together effectively to ensure that discounting decisions are well coordinated to maximize profitability.

For digital marketing analytics, a tipping point has been reached in industry – ownership of these analytics is moving from agencies to internal marketing sciences groups. Using the TGaS benchmark data from across more than 40 top pharma companies, this presentation looks at:

• The relative investment in these media across pharma portfolios
• How pharmaceutical companies are organized to manage this information today
• Quick wins to ensure above-benchmark tactical performance
• A more thorough approach to managing digital marketing analytics at the brand level

The Patient Protection and Affordable Care Act (ACA) contains important provisions relating to the development of new payment and service models. Under the new accountable care models driven by CMS and similar pilots on the commercial payer side, health systems may have an increased interest in understanding and managing the pharmaceutical components of care (including in the outpatient setting). In this presentation, we discuss the types of data sources that are likely to be useful in order to best understand health systems/IDNs, approaches to the integration of various data sources, and analytics that is helpful in identifying potential partners for pharmaceutical companies in this environment.

Data sources are useful ranges from basic census information relating to the geographic area covered by the IDN, increasingly rich data sources available from CMS (e.g., quality data and HTI capabilities, and various subscription data sources that can be integrated into relational databases to can support further analysis. We discuss the process through which this can be done, how IDN/ACOs can be profiled and how Latent class analysis can be leveraged to segment IDN/ACOs. Particular focus is also paid to specifics on the segmentation process from an analytical perspective.

Efficient analysis of large scale Medical Claims, Prescription Records and Clinical Trials data requires the use of appropriate tools that are optimised for the needs of individual departments and research needs. General purpose platforms that promise a generic and universal solution often fall short of expectations due to sub-optimal performance under different workloads. While the scope of research is generally specific to a project, there are a few challenges and data concerns that are broadly shared and is common across the spectrum.

These so-called shared data challenges include: Assessment of Data Quality, Data Cleansing, Data Accuracy, Combining Disparate Data Sources and so on.

The presentation will demonstrate how Research, Sales & Marketing, and Epidemiology departments can analyse medical and Rx data using innovative solutions from the Financial Industry that greatly improve processing speed thus allowing researchers to perform iterative analysis of data that would be otherwise unfeasible. In our present platform, simulations that used to take hours and sometimes days have now been reduced to the order of seconds and minutes respectively.

The key characteristics of the system include:

• Simplicity, Ease of Use, Administration and Maintainability
• Significantly reduced analysis timeframes allowing researchers to run multiple trials and simulations to develop comprehensive results. Productivity increases as the focus shifts from how to get the data to analysing the data
• Extreme Quantitative Data Processing capabilities

Every payer manages access to a brand and the market with various controls (E.g. Prior Authorization, Step-Edit, or Patient Copays), which can be more or less restrictive dependent upon whether or not there is a payer contract in place. Additionally, when a brand decides to not contract with a payer, patients will often deal with an NDC Block or some other denial. By analyzing the full lifecycle of a prescription claim, manufacturers can better understand key concepts such as a patient’s ability to overcome a Prior Authorization, a Step-Edit, or even an NDC Block, or which payers’ benefit design is more or less restrictive, and whether a payer will allow patients to fill the brand without a payer contract. In the “Do you really need to contract with a payer” presentation, the audience will be gain a clearer understanding of how to analyze claims lifecycle data for improved payer contracting strategies.

Few things are more exciting to an industry than the prospect that new data assets can be harnessed to provide insights that can drive true competitive advantage. There has been reasonable hope in the Pharmaceutical industry that social media data could be a major, new and relevant source of brand intelligence to complement, and perhaps even supplement, traditional data sources such as ATUs, syndicated panel-based reporting and the like which are generally more costly. Roughly speaking, social media refers to the active monitoring of social media channels for information contained in posted conversations.

Social media data provides fuel for tracking softer metrics reflecting Consumer, HCP and other key stakeholders sentiment towards a brand overall (akin to brand equity) as well on specific, lower level dimensions such as price, safety, efficacy and has the prospect of providing richer insights. Perhaps most exciting to marketing teams, however, is the prospect that social media data could prove to be a valuable data source as an early warning indicator of brand performance (particularly during early launch) and for going beyond tracking to evaluation of marketing activities and informing actions to calibrate those activities.

The specific objectives of the proposed presentation are to:

• Define a composite summary social-media based metric (brand valence) with desirable mathematical properties that can be used to summarize the overall sentiment about a brand, as well for specific categories of sentiment (brand category valence) with respect to specific product features and benefits such as cost, side-effects, patient support programs…etc by stakeholder.
• Review current uses of social media data for driving actionable Pharmaceutical insights.
• Present information on the extent to which social media-based overall brand valence is an early indicator of the trajectory of brand performance, and whether social media-based brand category valence measures can be used for evaluation of marketing activities and informing actions to calibrate those activities.

Innovations in cancer treatment, especially those linked to precision medicine, are not only improving outcomes and survival rates for patients, but also driving increased competition in the oncology drug market. Data from diagnostic tests has created an opportunity that many life science companies are taking advantage of to develop mechanisms of action that are more targeted and effective. Medivo is putting that data to work through advanced analytics to drive critical insights that identify early treatment opportunities and gaps in care. In this presentation, Medivo will discuss two oncology case studies that predict diagnosis and disease progression based on lab data using advanced analytics methodologies.

The nationwide Medivo lab test database includes over 150 million test results from over 200,000 practices. Our database includes both retrospective (historical) data from 4 years past and current data. We will demonstrate the power of predictive modeling, using de-identified data from disparate lab sources, to reveal in-depth insights critical to inform brand strategies.

Case Studies:

• CML
• Assess likelihood of CML patients achieving remission as a result of mutation testing following clinical relapse
• NSCLC
• Identify clinical factors leading to disease progression in patients with NSCLC
• Predict when diagnosed patients start trending towards danger (in need of intervention).

For many years, pharma companies have relied on sales data for calculating field incentives. Incentive data is typically made up from third party aggregated prescriber sales data. And confidence in that data is well accepted, as are the projection methodologies to fill in any black holes.

Today, longitudinal patient data is robust enough to provide incentive grade data, in fact, it parallels coverage of retail data, but uptake to use patient data within incentive compensation plans is slow. Our presentation will focus on showing a side by side analysis of a contest using traditional NRx data vs. using New to Brand Patient data for the same contest.

What insights are gained by using patient data? We will present that using patient data provides field sales with more actionable and tailored messaging that will resonate with their targets. A different message can be given to prescribers that are truly using Brand A for new patients vs. continuing patients.

Traditionally, a “rational” business development marketplace has been maintained by pharmaceutical manufacturers utilizing a discounted cash flow (“DCF”) approach as the primary determinant of value for an asset. In recent years, valuations have moved above levels historically supported by DCF methods, and by that technique appear to be “irrational.” However, by introducing non-traditional valuation metrics, it is possible to quantify and predict the premiums being paid for life sciences assets. These nontraditional metrics can be used to establish a new definition of “rational” valuations based on (i) attributes of the buyer and (ii) comparative transaction value.

This presentation will investigate the best practices for measuring the premium that is correlated to each non-traditional characteristic and each combination of characteristics. By utilizing a database of direct DCF valuations (with and without risk adjustments) for a basket of deals that occurred over the past 5 years, we can create a baseline of valuations from which comparisons can be made. By analyzing each of the deals in our basket, we can identify which and how many of the non-traditional characteristics were present with the company that won the bid. From these data, conditions that support differential premiums to DCF will be characterized.

We will also provide examples of the comparison of results using DCF versus non-traditional methods and in-depth analysis of how knowledge of these measures could have been used to change the outcome of an organized asset sale process.

Physician targeting and valuation is critical to the success of pharmaceutical manufacturer sales forces. Use of physician-level data that catalogs patients and prescription activities has become a widely accepted means for identifying the most valuable physicians for an upcoming drug launch, seeking additional indications for an existing drug, or to bolster sales for an established brand facing new competition in the market. Despite investing substantial time, effort, and financial resources to target physicians, pharmaceutical manufacturers have largely ignored treatment sequencing when determining which physicians they should pursue. Particularly in the oncology realm, treatment sequencing significantly impacts duration of therapy, and therefore, plays a crucial role in overall drug performance. It is widely accepted that duration of therapy is longer in first-line than in later lines. For pharmaceutical manufacturers with oncology drugs that have indications across lines, the objective is to have their product used earlier in the treatment spectrum. This sequencing information can then be used to weigh first-line regimen use more heavily than later lines when ranking physicians for targeting purposes.

Patients with Epilepsy have a number treatment options available to them, but significant unmet need remains. Even among current approaches to care, there is little clear real-world, independently generated data to help physicians and patients know what the right approach is for them.

By engaging thought leaders and using patient level data, we were able to hypothesize and then demonstrate what factors impact outcomes for Epilepsy patients. Robust analyses were conducted on both open and closed APLD datasets to understand how factors like drug choice, access to specialists, proximity to centers of excellence, deliberate effort to optimize therapy, insurance type, and formulary policy impact patient outcomes, as measured by hospitalizations (both epilepsy related and all-cause).

Results shows that the use of 2nd generation anti-epilepsy drugs resulted in improved outcomes, as did deliberate effort to optimize therapy (as seen via switches and add-ons). Second order effects showed that access to specialists, proximity to centers of excellence, and better formulary coverage led to higher use of 2nd generation therapies, and, thus, better outcomes for patients.

Outputs of the analysis were featured in a peer-reviewed publication, and used to create state “scorecards” that the Epilepsy Foundation used in its advocacy efforts with policymakers.

Sales crediting and eligibility are two of the most common challenges that organizations face in managing their Incentive Compensation programs. Traditionally, these processes have relied on accurate alignment and roster data to determine the right individual to get sales credit, absence of which causes IC & other downstream systems to wonder ‘what happened’ well after ‘it happened’.

As sales organizations evolve in their complexity to respond to a dynamic market place, it has become essential to adopt a nimble sales deployment model where the field leadership, rather than the traditional home office, is empowered to make territory and account coverage decisions. This inevitably increases the frequency (in many cases driven by the situation on the ground) of alignment changes and amplifies the challenges associated with sales crediting and eligibility.

Specifically, in this paper, we discuss four challenges related to incentive compensation that can be traced back to inflexible alignment processes:

• Evolving customer landscape that necessitates a change in sales touch points. (e.g. the ongoing shift in decision-making from individual HCPs towards IDNs)
• Long lead time transactions that involve multiple sales resources, or accounts where both the sales and the service teams are responsible for the growth of the account.
• Standalone Alignment and IC systems/processes with poor integration that result in sales crediting being a reactionary activity, leading to long lead times for incentive payouts.
• Dealing with a diverse set of exceptions to alignment that can wreak havoc on sales crediting and eligibility processes.

While investing in an alignment ecosystem, organizations not only need to ensure that effective controls can be placed on the flow of alignment information into incentive compensation and reporting systems, but also that the field leadership understands the impact of their alignment changes on sales crediting and eligibility.

Successful organizations have addressed both challenges effectively, by adopting the following key approaches:

• Implementation of robust governance processes to ensure alignment changes are in line with defined deployment philosophies, minimize disruption to customer relationships, and are controlled for impact on incentive compensation timelines.
• Enabling insights at the point of decision making to not just include the impact of coverage changes on the business; but also on the sales resources impacted by the change, as it relates to their incentive compensation.
• Tight integration of upstream (customer master, HR systems) and downstream (incentive compensation, reporting, CRM) systems with the alignment system such that the data flow is structured, seamless and automated between the systems.
• This paper presents a case study of how Quest Diagnostics overcame these challenges by adopting a roadmap for success using a blend of insights at the point of decision, powerful technology and good governance processes.

One area within the pharmaceutical industry that is underserviced by analytics community is the distribution channel. In the current environment, having a physician write the prescription for the product is only part of the challenge. Product stocking at the pharmacy, especially in highly competitive and genericized markets, is an important aspect to ensuring the patient receives the prescribed product. Lack of product availability and inadequate pharmacist education can lead to switching and abandonment, making all other marketing efforts go for naught.

ValueCentric and Xenoport partnered to create a successful product launch by creating a data driven strategy to identify high value pharmacies to maximize patient product access. This collaboration resulted in a successful relaunch of their product Horizant, growth coming in part from a focused strategy around retail pharmacies that included Sales, Marketing, Trade, and Managed Markets. The strategy included:

• Creating a target list of pharmacies for stocking efforts
• Providing pharmacy targets to physician sales reps and account directors
• Developing realistic stocking goals, and tracking those goals against ex-factory sales data
• Prioritizing non-personal promotion and consumer programs to pharmacies with the most potential patients
• Post-launch, identifying high potential pharmacies with no product sales

To create these types of segmentation, manufacturers can use a variety of data sources available in the market today. Internal ex-factory sales data is a useful source for larger manufacturers with other products in the therapy class, while national level store sizing databases are also available. Various samples of store level databases are available from store chains, software vendors and data aggregators, each of which requires imputation to develop segments for the full population of pharmacies.

Pharmacy segmentation and targeting can be also taken to further levels by incorporating additional metrics to create a multi-dimensional data set for additional statistical analysis. Relative location to clinics and hospitals that are on formulary can be essential to identify patient traffic and ensure the appropriate pharmacies have stock available. In addition, managed care metrics can be overlaid to prioritize pharmacies for customer related financial promotions. To complete the analysis, post-launch pharmacy level sales can be included to create another important dimension for an annual resegmentation.

For decades, companies have relied on physician volumetric data to gauge market trends and target sales and market initiatives, despite ever-present gaps in coverage. In response, Q2 Metrics has worked with industry to combine volumetric and profiling data in order to fill in the inherent gaps in volumetric data, thus creating new synergistic approaches to enhancing sales and marketing targeting. By combining claims-based procedure or prescription volumes with physician biographic, professional, and academic details, Q2 Metrics can profile physicians and apply predictive statistical models that estimate the procedure or prescription volumes for doctors that are missing from the claims-based analysis.

This approach has been tested across multiple therapeutic specialties by segmenting observations into a training set to determine the best predictors of volume and a qualifying set to test the models created. After the model has been optimized, it is applied to relevant practitioners absent from the initial claims-based analysis in order to estimate their procedure or prescriber volumes. In this presentation, Q2 Metrics will show how applying this type of advanced analytics can unlock value within data sets that a large number of companies may already house internally, allowing users to minimize risk and maximize rewards.

The industry uses a broad range of approaches today for customer valuation, promotion response analytics and sales force resource allocation. With the increasing complexities in the marketplace and availability of richer data, there has also been continuous innovation in analytical methodologies for resource allocation.

On one end of the spectrum is simple economic valuation and resource allocation based on product/market prescription volume & business rules. At the other end of the complexity spectrum is deriving customer valuation based on patient data analytics, promotion response methodologies that incorporate IDN/Payer influence etc.

In this case study, we quantified the difference between simple and complex methodologies for a single product sales force. We compared three different methodologies of resource allocation:

• Market/Brand volume based methodologies
• Promotion response based methodologies
• Advanced promotion response methodologies including Payer/IDN influence quantification

In order to assess the differences in the impact, we used the following metrics:

• Regional Resource allocation: How much do the complex methodologies change the resource allocation
• Brand Impact Assessment: We generated a call plan for the field force for a retrospective time period. We quantified whether conforming to the call plan based on the more sophisticated methodology drove better territory performance.

Over the past few years there has been an influx of new drugs to treat various forms of cancer and many existing brands have gained new indications. Industry representatives wishing to engage in discussions on their new brands/indications have the challenge of obtaining “face time” with physicians, followed by delivering messages with the greatest impact on brand choice, that also meet key quality metrics used to help evaluate sales force effectiveness. To meet this challenge, biopharmaceutical companies in the oncology arena have developed different strategies for sales force deployment. These structures continue to evolve, although they tend to fall within two main categories: (1) a brand-focused, and (2) an indication-focused approach. This study evaluates multiple examples from the Multiple brands with multiple indications (MDMI) structure and the Indication-focused multiple brand (IFMD) structure using sophisticated modeling techniques to provide in-depth insights into the relative impact of each structure, along with recommendations.

One of the key strategies for many companies is to hyper target influential physicians to increase the reach of the marketing message. However, identifying influential physicians remains challenging. In this presentation, we will illustrate how we use the publicly-available co-publication data to build physician network. Through the network analysis, we are able to measure each physician’s degree of connectivity by quantifying how likely a physician is at the most direct route between two people in the network and how fast a physician can reach everyone in the network.

We then perform a cluster analysis to create distinctive influence segments by taking into account each physician’s degree of connectivity within the network, number of publications as the first author, number of publications as co-author, and other relevant factors. We will also present a case study showing how we worked with one of our clients to design a multi-channel contact strategy for influential physicians.

Commercial analytics teams are routinely asked to assess the feasibility of proposed phase IV studies and further challenged to support these studies by devising approaches to locate these specific patient populations according to the trial criteria. These activities can be especially challenging in niche indications, rare/orphan diseases, or in cases of numerous inclusion and/or exclusion criteria.

This approach for the location of patient and physician clusters utilizes multiple data sources, including longitudinal medical and pharmacy claims and organization affiliation data. These clusters are then grouped geographically at the physician and hospital level in order to guide the prioritization and selection of potential trial sites. This approach can result in greater accuracy in gauging potential patient population and enrollment by site thus resulting in improved market planning. A case study will illustrate the prioritization of geographic areas and sites/facilities based on patient volumes, patient volumes by physician, and patient volumes by facility.

Optimizing a drug’s promotional return on investment is a critical task that has a major impact on its revenues. This task has become significantly more complex in recent years as the number and variety of tactics and channel options has exploded. This trend is likely to continue as the pharma industry evolves toward integrated multi-channel, customer-centric selling models. From an ROI measurement perspective, classical marketing mix methods that link individual tactics exclusively to sales are no longer adequate; marketers today also need to determine how tactics interact and impact one another.

In this joint presentation based on recent work between Eisai and ZS Associates, we will share an improved method for measuring the total value of each promotional tactic – both the direct impact on sales as well as on upstream and downstream activities in the treatment decision process. The mutually reinforcing connections between tactics form promotional “pathways” leading to improved sales. Promotional pathways modeling provides:

• A deeper understanding of how promotion impacts the customer journey
• A quantified assessment of how campaigns reinforce each other
• An analytic platform to support promotion investment decisions and inform harmonized touch-points to improve customer engagement

Our presentation will begin by introducing key pathways modeling concepts and the general approach. Data integration is a critical step in the measurement process, and we will share lessons learned around how pharma companies can improve how they track their promotional activities to enable better pathways modeling and ultimately inform better marketing decisions.

Next, we will discuss two recent project examples at Eisai that illustrate the pathways modeling approach and how it enhances resource allocation decisions. The first case example, for an oncology brand, demonstrates how personal promotion activities – sales force detailing, in-office programs and speaker events – interact to improve customer engagement. The second case example, in a retail market, shows how diverse tactics such as sales force detailing and digital paid search are closely linked to consumer-oriented approaches such as vouchers, which encourage patients to talk to their doctors.

We will conclude our presentation by sharing key learnings from our experiences that can help other organizations that are seeking to extract additional value from their promotional investments. These learnings will include best practices for data collection, stakeholder involvement, and coordination across promotional tactics.

In today’s challenging environment, it is difficult for a newly launch brand to meet the forecasted expectation. For a pharmaceutical company, therefore it becomes essential to design optimal and well informed brand strategy ahead in order to capture full value of the brand. Given the high risk of making an un-informed decision it becomes critical in Oncology to understand and react to this dynamic environment.

In this abstract we will showcase a three-fold approach to build a solution that helps brands to capitalize on the full potential of a new launch brand.

The approach includes the following three pieces that work in tandem:

• Understand: What are the patients that can benefit from the brand and where are my leverage points?
• Approach – Analyze the patient pathways using longitudinal patient level data. Identify the leverage points where the brands can make a difference. Look at the intersection of the patients, physicians, accounts and payers to relatively value the leverage points. Find the best way to approach various physicians groups that are treating those patients by designing physician segmentation and preparing a messaging framework.
• Execute: How can we find those patients and activate the physicians and educate them about the benefits of my brand in real time?
• Approach – Design a process to mine the patient level longitudinal data to find the patients which meet the criterion for the leverage. Look at the intersection of the patients and physicians to find the right physicians. Reach out to the physicians in right time with the right message that is most appropriate based on the segmentation and type of leverage. For e.g. physician in “Early Adopter- High Payer Control – Brand Loyalist” segment is about to make treatment decision for a newly diagnosed patients.
• Enhance: How can we understand the brand performance and use the learning to enhance our execution?
• Approach – Track the brand performance by leveraging patient level longitudinal data. Assess the performance against each of the leverage points on the patient pathways. Improve the execution based on the learnings from the performance analysis. Change the messaging based on changing physicians behavior. Inform resourcing and staffing based on the volume of execution and extent of achievement.

For e.g. who are the physicians who keep patients on 8 cycle of treatment for my brand (instead of 6) and achieve better results

Modification of any sales force territory alignment, or any change in the size of sales force, inevitably leads to the problem of deciding which sales representatives should be assigned to which territories. The problem is particularly acute in a down-sizing, where placement of some reps almost inevitably means the dismissal of others. Doing this in a way that is optimal for the company (getting the best available rep into each territory) is critical to the success of any sales force realignment effort. And, in a down-sizing, doing it in a way that is visibly and provably “fair and unbiased” can be critical for morale, and to defend against possible legal action.

Classically, this problem has been addressed by developing business rules about the relative priority of various factors, usually including: Rep tenure, rep performance, distance to be traveled to each possible placement, and target overlap between existing and new territory assignments. These business rules are (typically) used to rank the rep placements in each possible territory, and a sequential assignment is made, starting with the highest-ranked placements.

This presentation will discuss an alternative solution approach based on linear programming, made possible by the advent of powerful but relatively inexpensive desktop solvers. A wide variety of complicated business rules can be expressed as constraints, and/or as variations on objective function weightings. Solutions for realistically-sized problems (300-400 reps, up to 15,000 possible rep-territory placements) can be obtained in only a few minutes.

Advantages (and disadvantages) of each approach will be discussed in the presentation.

BACKGROUND: Diabetic retinopathy is one of the most common complications of diabetes. The screening of patients with diabetes to detect retinopathy is recommended by several professional guidelines but is an underutilized service.

OBJECTIVE: To analyze the relationship between the frequency of retinopathy screening and the cost of care in adult patients with diabetes.

METHODS: Truven Health MarketScan commercial databases (2000-2013) were used to identify the diabetic population aged 18 to 64 years for the performance of a 2001-2013 annual trend analysis of patients with type 1 and type 2 diabetes and a 10-year longitudinal analysis of patients with newly diagnosed type 2 diabetes. In the trend analysis, the prevalence of diabetes, screening rate, and allowed cost per member per month (PMPM) were calculated. In the longitudinal analysis, data from 4 index years (2001-2004) of patients newly diagnosed with type 2 diabetes were combined, and the costs were adjusted to be comparable to the 2004 index year cohort, using the annual diabetes population cost trends calculated in the trend analysis. The longitudinal population was segmented into the number of years of diabetic retinopathy screening (ie, 0, 1-4, 5-7, and 8-10), and the relationship between the years of screening and the PMPM allowed costs was analyzed. The difference in mean incremental cost between years 1 and 10 in each of the 4 cohorts was compared after adjusting for explanatory variables.

RESULTS: In the trend analysis, between 2001 and 2013, the prevalence of diabetes increased from 3.93% to 5.08%, retinal screening increased from 26.27% to 29.58%, and the average total unadjusted allowed cost of care for each patient with diabetes increased from $822 to $1395 PMPM. In the longitudinal analysis, the difference between the screening cohorts’ mean incremental cost increase was $185 between the 0- and 1-4–year cohorts (P <.003) and $202 between the 0- and 5-7–year cohorts (P <.023). The cost differences between the other cohorts, including $217 between the 0- and 8-10–year cohorts (P <.066), were not statistically significant.

CONCLUSIONS: Based on our analysis, the annual retinopathy screening rate for patients with diabetes has remained low since 2001, and has been well below the guideline-recommended screening levels. For patients with type 2 diabetes, the mean increase in healthcare expenditures over a 10-year period after diagnosis is not statistically different among those with various retinopathy screening rates, although the increase in healthcare spending is lower for patients with diabetes who were not screened for retinopathy compared with patients who did get screened.

Traditionally, pharma brands have developed their HCP target lists based on physician prescribing data at both an individual and territory level. Now, new methodologies have emerged that leverage patient-level Rx transactional and behavioral data, to create more qualified cohorts of doctors that empower brands to reconfigure and maximize their field force strategies, and by extension, their sales force plans as well. In this presentation, Crossix will explore the methodology behind this innovative approach and share a case study that demonstrates how a pharma brand leveraged actual patient Rx behavioral data to improve salesforce planning and goal-setting.

Being data driven is good but solely relying on data which assumes rational behavior can create blind spots. In the emergence of active and passive big data data, best decisions are made when human and machine work together. Dextro Analytics will shed light on whole new approach of promotion response modeling including:

• Why we need to go beyond association based and traditional techniques like logistic regressions?
• How to introduce causality in response modeling?
• How to overlay irrationalities and market events to identify optimal promotion strategy?
• How use of artificial intelligence is changing the landscape of promotional mix?

Consumers are routinely targeted through online media campaigns, with each campaign having the potential to reach hundreds of thousands of consumers. Determining the success of these campaigns has historically proven to be difficult. Metrics, such as number of ad views/exposures, are important but don’t tell the entire story. In order to understand if the campaign was truly effective, returned a positive ROI, and is reaching the target audience, one needs to understand the effect the campaign had on the consumers’ behavior. To do this, secondary data assets, including patient level claims for medical, hospital, and retail pharmaceutical brands, needs to be leveraged.

This presentation will illustrate a methodology for measuring online media exposure using patient pathway and health behavior metrics. Online media campaigns have many components, from creative analysis to placement analysis, each requiring proper measurement by analyzing the exposed consumers. This requires a representative control group of non-exposed consumers taking into account online behavior and health profiles. By combining online campaign exposure data with Symphony Health Solutions (SHS) patient level claims data, the methodology yields more accurate insights into a campaign’s true effectiveness by using patient analytics.

Pharmaceutical companies mostly are interested in frequency of the customer visits and number of the customers to reach. They would like to optimize per customer visits quantitatively. On the other hand, qualitative measurement of each visit is the complementary phase of the quantitative analytics. Getting a robust understanding of customer messaging strategy is as important as determining the optimal visits per high potential customers.

The Brand teams track the promotional performance. They also try to understand whether their messaging strategy had been followed or not. Message bundle analysis helps to assess if the presentations are delivered to the right customers, which presentations work best, and what message bundles drive the highest impact. This analysis is helpful to improve the qualitative time of customer visits, and be helpful to provide the right messages. Moreover, it helps to identify the most impactful message bundles for the right customers. Consequently, Message Bundle Analysis is beneficial and complimentary to determining the right frequency and messaging of the customer visiting activities and number of the customers to reach.

Medical Procession Ratio (MPR) and Proportion of Days Covered (PDC) are traditional measures of patient adherence. While insightful when measuring adherence for those on daily-use drugs, for treatment plans more idiosyncratic and customized based on disease activity such measures are biased. As disease-activity driven treatment plans become more integrated into patient care, it is important that we leverage methodologies which are more appropriately suited to the complexity of these treatments.

Symphony Health Solutions (SHS) leverages a proprietary data source of patient level medical office- pharmacy- and hospital claims. The data is used to create baseline factors such as demographic and socioeconomic indicators and customized market-specific factors such as co-morbidities and concomitant therapies. Using patient level claims to measure treatment activity over time and baseline and market-specific factors as inputs we can model persistency using a Kaplan Meier model and adherence using a Cox Proportional Hazard model. These methodologies are more comprehensive as they allow us to determine the significance of these factors in actual adherence to treatment.

Overview: With the trend toward more frequent updates of medical claims data (daily and weekly), Pharma clients are finding new ways to utilize insight from medical claims data to drive swift sales and marketing actions toward physicians.

Why Important: When it comes to physician targeting, medical claims data has often been historically viewed as outdated, particularly when compared to prescription claims data. This made it challenging to use medical claims to support timely business decision making. However, more recently, as healthcare technology evolves, medical claims data is becoming more rapidly available. This shift has opened a door to new and innovative ways of using daily and weekly claims feeds to derive key insights and enable more agile sales and marketing teams.

Supporting Use Cases: In this presentation, we will examine the results of a recent pilot study in which a pharma company gained a competitive advantage in an infused therapy (can we be more specific) market with “near real time” medical claims data. The cases will cover:

• How the pharmaceutical commercial operations function utilized customizable alerts based on defined diagnoses to align with physician targets
• How their pharmaceutical representatives leveraged medical claims data to identify physicians with newly diagnosed patients, but before a treatment regimen has been established.
• An examination of the promotion response and campaign effectiveness through weekly Early Alert trending reports based on medical claims activity (diagnosis and procedure) at the physician/facility/payer level.

The use and application of big data for improving commercial operations has been an exciting challenge across all industries. The Consumer Packaged Goods industry has often led the way in the development of innovative approaches because data volume, velocity, and variety have grown substantially and started earlier than many other industries (with the exception of the Financial and Insurance industries).

The Procter & Gamble Co., a leader in business analytics, has introduced in recent years a new approach to analytics, based on management by exception. The highly successful solution is centered on business processes and promotes agile decision support by pinpointing exceptions, identifying business drivers, and creating focused alerts to issues that really matter. The new approach resulted in a significant improvement in the quality and speed of business decisions.

We found a great commonality between the consumer goods and the pharmaceutical markets, and decided to apply their successful method in business analytics when developing a solution for new drug launches to handle all commercial operations aspects of launching a new drug: brand management, managed care contracts, promotional activity, and of course – sales.

Verix’s Drug Launch application, incorporates P&G’s concepts of management by exception, highlighting trend breaks, alerting on diversions from expectations, and enabling rapid course correction of unfavorable situations and opportunity identification of favorable ones. Bayer HealthCare adopted the application for their launch of two new high profile oncology special pharma drugs. For both new drugs, the results exceeded expectations with smooth operations and tremendously successful product launches.

Personalized Marketing is playing more and more important role in the pharmaceutical industry. It took almost 15 years after one of the first Harvard Business Review articles on the personalized marketing to make it an important and powerful force in the pharmaceutical sales & marketing. Forbes argues that personalization is a key to the future of marketing and many industries already implemented personalized marketing as one of the most important sales & marketing instruments.

With many marketers trying to implement the personalized marketing either on the HCP (health care practitioner) level or on the digital & DTC/patient level it leads us to the question: do we need new and different Data & Analytics to support, measure, and optimize Personalized Marketing campaigns?

This presentation is trying to answer the questions what data and analytics we need to support personalized marketing and is also giving few examples of the personalized marketing analytics.

Differently from the other industries, in the pharmaceutical sales & marketing there are different levels/types of the personalized marketing. For example: HCP/physician level vs. patient & personalized medicine level especially with availability of the APLD, Biomarker & Genomics data.

Personalized Marketing on the individual HCP level requires individual-physician level analytics & data. Historically, most of the pharmaceutical sales & marketing analysis is performed on the zip, territory, or physician segment level and is lucking individualization. Few attempts to build physician-level promotion response curves as a first step of the personalized marketing were done in the past but because of the noisy data it lacked the wide spread acceptance. The HCP-level personalized marketing requires not only physician-level data and analytics but also Patient-Level Data for each individual physician to estimate an impact on the patient behavior of the individual physician.

Digital & DTC Personalized Marketing requires more patient-level data and analytics as well as complex allocation of the DMA-level DTC effort on the physician level. This complexity is coming from the fact that DTC & TV campaigns are run on the DMA (Designated Market Area) level while these campaigns have different personalized impact on the individual patients of the individual physician. And allocation of the DMA level DTC to the individual patient & physician using patient-level data brings extra level of complexity in data and analytics.

To summarize, in this work we discuss new personalized marketing and the paradigm shift in the data & analytics required to support, measure, and optimize personalized marketing campaigns.

The global vaccines market is expected to grow from a $33 billion dollar industry in 2014 to $58 billion by 2019. Manufacturers in the vaccines arena will increasingly require accurate and transparent forecast models to help allocate scarce commercial development and promotional resources. Traditional non-circular prevalence and incidence demand forecasting models in the biopharmaceutical industry do not adequately capture the patient flow dynamics observed in the vaccines marketplace. Traditional models applied to vaccine products can lead to over or under estimation of patient opportunity, compromising forecast accuracy. We will present a modeling approach based on a systems dynamics framework that elegantly solves for patient flow complexities observed in the vaccines market and delivers accurate and transparent results. Specifically, a systems dynamics approach affords accommodation of single-age cohort calculations and associated rollups to age groups while accurately accounting for “age in / age out” dynamics and mortality. These models also allow visibility into a wide range of market metrics that otherwise would be impossible to derive or track. Importantly, this approach also allows transparent benchmarking to cumulative vaccination rate epidemiologic data, such as data published by the National Immunization Survey.

While clinics and physician offices continue to limit access to pharmaceutical sales representatives, the demand for information and support for new products and new indications still grows. Access restriction to reps is often driven by MCOs, corporate IDN management, HCOs, Therapeutic Committees or even legal restrictions, but, in many cases, not by the HCPs themselves. The reality, of course, is that HCPs still want information. So, while traditional access is drying up, demand for new insights is surging.

This demand can most certainly be met through the use of an appropriate mix of channels and tactics, messages, calls to action, cadence and timing. The key is to optimize this mix to ensure that age old marketing principle is achieved, namely, “getting the right message into the right hands via the right medium at the right time”.

The problem is made more complex by the concept of “mix”. The challenge is not to find the “one or two” channels that resonate well with each HCP or HCP segment. Instead, the challenge is to successfully design effective combinations of channels and appropriate sequencing of information delivery based on the massive amounts of information now available about information consumption behavior of HCPs. Today, we can link online and offline information behavior, at the HCP level, to understand content preferences, channel sequencing and related engagement activity. This opens up a vast network of messaging strategy and information delivery, all optimized by data insights.

This presentation will dive deep into identifying, measuring, predicting and weighting cross-channel propensity by forming segments of physicians based on common responsiveness traits – not a pre-hoc criterion such as activity, decile or specialty. The end result is a touch-point analysis that provides an optimal mix of channels, message content and message cadence or flow – all to create effective promotional communications across a defined set of “CRM Targets”.