OR WAIT null SECS
Even though data can single out physicians with high marketing upsides, most pharma companies are doing without such high-value data.
Pharmaceutical companies are not using the best marketing data available. Sure, they spend tens of millions of dollars a year on data, and billions more on marketing campaigns, but they are surprisingly vague about the prescribing habits of the physicians those campaigns target.
Anonymous patient-level data (APLD) could bring a new level of precision to pharma marketing. APLD can differentiate between physicians who are writing mainly refill prescriptions, and those writing mostly new-to-therapy prescriptions. Many companies would reshape their marketing and sales campaigns based on these data alone. Managers understand that physicians who are writing new prescriptions deserve more attention from sales reps. But even though data can single out physicians with high marketing upsides, most pharma companies are doing without such high-value data.
Why is pharma ignoring information that could increase returns? First, acquiring and utilizing the new data is seldom easy. It takes more than money to buy the right information. Not every APLD vendor will have the same data sets available, so companies must begin by shopping for the right vendor. Second, neither the APLD data products nor many of the companies offering them have matured. This creates a complex, shifting vendor landscape, in which it is difficult to sort out competing claims and counter-claims.
Another barrier to entry is the cost of switching from present data sources. Currently, pharma companies rely on prescription data such as IMS' Exponent/NPA and NDC's Source Prescriber to identify high-prescribing physicians. While this data says nothing about which physicians have the most potential prescribing upside (or downside), it is firmly entrenched within most pharma companies. IMS, in particular, does not provide APLD yet with the depth and breadth of other vendors, but its data are used to calculate sales force compensation. And if the cost of switching is high, so is purchasing multiple databases.
Finally, the organization as a whole must change to take full advantage of APLD data. Changes in the way physicians are targeted will lead to changes in the way sales representatives are trained, and probably in how they gather and digest information before calling on doctors.
APLD is healthcare-utilization data that can be linked, in a HIPAA-compliant manner, to individual patients longitudinally. That is, they are data that can track a patient's healthcare utilization over time. This data is captured through similar sources as standard prescription data are, but with two major differences: the number of data fields captured is increased (in some cases), and proprietary linking algorithms that utilize multiple fields (birthdate, zip code, sex) are applied to create a unique encrypted identifier for each patient.
APLD itself is not new. Small data sets have been used for longitudinal analyses in the outpatient market for several years (for example, PMSI, CHAMP, LifeLink). And Solucient (formerly HCIA-Sachs) has made large APLD sets available for hospital inpatients. But only recently have large APLD sets been applied to drug use in the outpatient setting. This latest development presents pharma with opportunities to use APLD to improve ROI.
APLD can improve sales and marketing ROI by delivering differentiated data on physician-prescribing data (see Figure 1). In this hypothetical example, traditional Rx data for a chronic outpatient drug treatment is compared with that using APLD. Based on traditional data, a pharma company would focus resources on Dr. Smith, Dr. Jones, and Dr. Brown, the high-prescribing physicians. However, APLD reveal that Dr. Jenkins and Dr. White may be more important than Dr. Smith and Dr. Jones because they prescribe drugs to more "new-to-therapy" patients. Physicians generally are reluctant to switch treatments when medications are working, so new-to-therapy patients have a higher value to pharma companies than renewing patients. This is particularly critical for companies launching new products in crowded markets. Sometimes, new-to-therapy patients represent the only chance to capture market share.
Figure 1Targeting the Right Physicians
In Figure 2, two physicians may appear to have the same affinity for product X. Unless the sales rep takes time to personally learn a physician's prescribing patterns, he would deliver essentially the same message to each physician. However, APLD reveals that Dr. Smith seems biased against product X for younger patients, while Dr. Jones may be biased against it for patients with cardiovascular co-morbidities. Armed in advance with this knowledge, sales reps can tailor messages to different physician segments.
Figure 2 Quantifying the Untreated
Enhancing forecast accuracy with APLD can improve ROI in new product planning. With traditional data sources, there is no accurate way to size the overall market for diseases when drug penetration within the target patient population is limited. (Consider the case of macular degeneration.) The same problem occurs when existing drugs are used for multiple indications. Chart audits, epidemiology studies such as the National Health and Nutrition Examination Survey (NHANES), and physician surveys, such as the National Drug Therapeutic Index (NDTI) and the Physicians Drug and Diagnosis Audit (PDDA), can provide trends, but their sample sizes are often too small or biased to provide accurate information for any but the most common diseases.
In addition, the most commonly used tools, NDTI and PDDA, only can project number of office visits. They cannot estimate unique patients reliably. Because APLD vendors link diagnosis codes to unique patients for 20 to 25 percent of office visits, they can accurately quantify the number of patients with a given disease (or at least the number of people making at least one office visit each year, which in most cases is the relevant population). A forecast based on a better estimate of unique patients can be 50 to 500 percent more accurate, which is sufficient to change the level of resources allocated to a product.
Even when prescription data can be used to size the overall market, APLD can drive more accurate patient segmentation. For example, assume a company has a fourth-to-market, me-too drug for a chronic disease. The marketers know that male patients over 60 represent the drug's best opportunity to gain market share. To quantify the size of this market segment with traditional data, researchers must build a complex model combining epidemiology and physician-survey data. Even then, the accuracy of the model will be questionable. But because APLD can track unique-patient drug use over time and link it to the patient's age and sex, a fairly accurate number can be generated. Once again, the improvements in forecast accuracy can change go/no-go and resource allocation decisions.
APLD captured through the physician office can link the diagnosis code to drug use. This allows drug companies not only to differentiate utilization of their drugs by indication better—how much calcium channel blocker is prescribed for hypertension as opposed to angina—but also to accurately quantify untreated patient populations and identify the demographics and concomitant conditions of the patient pool.
While these vendors only capture 20 to 25 percent of physicians currently, the data set is rich enough for marketing purposes—to identify and quantify market segments at national and regional levels, and make more robust resource-allocation decisions. The most powerful applications of this data, however, would occur if vendors were able to expand their data access and achieve this level of data on over 50 percent of physicians. Achieving this critical mass will enable drug companies to use the information to influence individual-prescriber behavior by tailoring messages to individual physicians, based on the number and type of untreated patients in their practices (see Figure 3).
Figure 3 Tailoring Messages
Pharma companies do not always understand APLD. Some of the blame lies with vendor personnel, who are perceived as "data people" rather than functional experts who can explain applications cogently to brand managers or sales operations managers. But part of the problem rests with pharma management. Even when companies understand the value of APLD, they may not be able to mobilize the personnel to translate data into actionable business decisions. And when they do—when companies see familiar processes in light of more sophisticated data—the consequences bleed into neighboring fiefdoms within the company. Senior management might see itself breaking a data logjam with a vendor, only to find itself refereeing unanticipated turf wars.
Any large-scale adoption of APLD would affect the way sales operations, brand management, and market research operate on a day-to-day basis. For example, if sales goals and methods were shaped by APLD, sales-force compensation might be linked best to several parameters besides total prescriptions written, such as number of switches effected, number of patients moved from untreated to treated, or number of new prescriptions written. Market research would have to adapt some of its metrics to the new data, while IT, HR, compliance, and a host of other departments would need to buy into the evolving system.
Managers also face the usual change-management issues if they back a move to APLD. In addition to persuading key constituents of the value of the change, executives must deal with the perceived pain and anguish of a new implementation. The most resistance likely will stem from parties that benefit from the existing system and therefore are threatened by a new one.
Companies choosing APLD face high barriers to switching from current data sources, most commonly from industry leader IMS. While new and existing data vendors have been building APLD capabilities faster than IMS, IMS remains the overall leader in the pharma information market due to its strong position in physician-level data. IMS is a formidable competitor because of its sheer size and market share. But the greater challenge for APLD vendors is the level at which IMS data have become entrenched in the business practices of pharma companies.
The most critical entrenched business process is sales-force compensation. While switching to APLD does not necessarily mean that administering sales- force compensation must change (although the new data would allow improvement on the current method), it could require using a different data source (Verispan or NDC physician-level data, for example), which in itself makes sales operations leaders uneasy.
IMS data also is considered standard for reporting to Wall Street. CEOs and CFOs want to remain consistent with that practice. All this contributes to even higher inertial barriers to APLD. As one pharma executive bluntly put it: "No one ever got fired for using IMS data." Unfortunately, continuing to use IMS for compensation, while paying other vendors for APLD, creates another barrier: Two data types are substantially more expensive than one.
Before attempting a large-scale adoption of APLD, pharma companies must work with APLD vendors to ensure better communication. Procurement managers must require vendors to clearly demonstrate the value of APLD. Collectively, vendors have a duty to simplify APLD for their pharma customers. They must put greater emphasis on selling APLD in general, compared with differentiating their individual services. A consolidation in the APLD market also could help clarify the options for customers.
The pharma companies must help vendors crunch the numbers and develop viable return models or business cases for large-scale adoptions in pharma. This involves not only a detailed assessment of the timing, costs, and one-time investments, but also a tangible potential benefit based on improved ROI. Pharma must work with vendors to assess the benefits of better decision-making, more targeted marketing, and improved sales-compensation programs. In addition, it is critical that vendors be required to thoroughly demonstrate in business-case scenarios the statistical validity of the data being used. Vendors must be able to help potential pharma customers assemble the internal business case needed to convince senior management to approve a large investment.
APLD raises personnel issues for both vendors and customers. Vendors need a talent pool that can use the patient-level database, but also understands the different business issues confronting pharma companies. Vendor personnel must not merely know how to manage and mine the database, but also be able to analyze business issues and turn good analyses into actionable insights.
In order to leverage the power of APLD in their businesses fully, pharma companies must assemble a team that can modify analytic engines for business decision-making. In addition, companies must adopt new systems, procedures, and business processes to make effective and efficient use of APLD.
Companies must look beyond the pharmacy to the increasing number of specialty drugs, often biologics, that will be administered in physicians' offices. There are already examples of these, including older drugs such as Leupron and Avonex, and newer ones such as Kinaret, Remicaid, and most oncology drugs. Currently, when it comes to these drugs, companies have no reliable mechanism of measuring individual physicians' prescribing patterns. APLD would improve current, imprecise mechanisms like physician surveys greatly.
Functional managers in pharma companies often feel that implementing a major change in business processes, such as converting to APLD, involves too much risk. It is easier to take the blame for a catastrophe than to get the credit for a coup. Senior executives must take responsibility for implementing APLD by defining the business-case requirements needed—returns required, level of risk, and required demonstration—to justify a sweeping change.
Jeffrey Boschwitz (firstname.lastname@example.org) is principal, Joyjit Saha Choudhury (email@example.com) is senior associate, and Charley Beever (firstname.lastname@example.org) is vice president at Booz Allen Hamilton.