A recent United States Supreme Court ruling protecting the use of data mined information as a guide in developing promotional strategies in the pharmaceutical industry positively impacts all business organizations using similar data collection and analysis. The purpose of this paper is to examine the use of data mining in the pharmaceutical industry, review its criticism, describe the Supreme Court decision, and assess the impact of the Court's holding.
In Sorrell v. IMS Health, Inc. (2011) a data mining health information company won its constitutional challenge of a Vermont statute that restricted the sale and use of data mined information. The Supreme Court held that such information was protected by the First Amendment and invalidated the statute as an infringement of free speech. There are those who see this outcome as an erosion of the privacy interests that might be advanced by keeping that information confidential (Huhn 2011), but the ruling is a victory for any business wishing to gain access to information that will facilitate their commercial transactions (Denniston 2011). Some legal commentators opine that Sorrell represents a new era of protection for commercial speech (Outterson 2011). Although there are differing views of that proposition, there is no doubt that the ruling positively impacts any business that uses information about consumer buying behavior to guide its marketing and sales strategies.
RELEVANT MARKETING LITERATURE
Data mining involves utilizing electronic methods for distilling meaningful information, trends and predictions from large volumes of data collected for some other purpose. The concept has been defined as "the process of secondary analysis of large databases aimed at finding unsuspected relationships that are of interest or value to the database owners" (Klosgen and Zytkow 2002, p. 637). From a marketing perspective, the patterns discovered in the data have value because they can impact the bottom line of a business (Peacock 1998). The end result is new insights and knowledge discovery (Fiske 2002). Businesses either generate their own information from data mining or purchase the gleaned information in order to improve marketing planning and decision making.
As described by Rafalski (2002), the process of data mining begins with "trend analysis and the search for patterns in the underlying data. Once a pattern of interest is identified, statistical analysis is applied to determine whether the pattern is significant. If it is found to be of significance, root cause analysis is applied to determine the cause of the trend" (p. 609). Tools utilized include query tools, descriptive statistics, visualization tools, regression-type models, association rules, decision trees, case-based reasoning, neural networks and genetic algorithms (Peacock 1998).
Data Mining in the Pharmaceutical Industry
This paper focuses on data mining doctors' prescription data which is purchased from participating pharmacies by health information companies such as IMS Health, SDI Health (formerly known as Verispan), and Source Healthcare Analytics (Information Management 2011). The channel of distribution in the pharmaceutical industry is unique in that the intended user of a product cannot access it without a physician, a middleman whose involvement is mandated by law. The original data is generated when a prescription is filled by a patient; the name, dose, and quantity of the drug is collected along with the date of the prescription and the physician's name. The patient's name is not retrieved. The data miner, however, assigns a specific number to the patient for tracking future prescriptions to permit analysis of the patient's prescription history (Orentlicher 2010). In 2007 IMS was reported to have obtained records on over two thirds of prescriptions filled in community pharmacies (Fugh-Berman 2007). Some pharmacies, e.g., WalMart, do not sell their prescription records (Pharmaceutical Sales Jobs 2012).
Data mining companies bundle the pharmacy data with data from other sources, primarily the American Medical Association's Physician Masterfile, a database that includes all (over 600,000) U.S. physicians (Stone 2011). Major data mining techniques utilized in the pharma sector include clustering (e.g., grouping drugs most likely/unlikely to be used), association (e.g., tracking physicians adopting drugs with customer's prescription), and classification and prediction (e.g., predicting the likelihood of success in a drug adoption process) (Ranjan 2009). Data miners' development of integrated proprietary databases has led to a wealth of information that pharmaceutical marketers are eager to access, specifically, profiles of doctors including name, specialty, practice site, and tracking information on prescribing habits for their own and competing drugs (Rafalski 2002). Drug companies observe trends and develop segmentation and marketing strategies based on the information (Fugh-Berman and Ahari 2007). The information is also used to evaluate the impact of sales representatives, samples, and any other marketing efforts on physicians. Some industry salespeople are compensated based on sales data drawn from the data miners' reports (Pharmaceutical Sales Jobs 2012).
Tracking helps a company label and group doctors as heavier or lighter prescribers of their drugs and determine the potential for a representative to effect a change in the doctor's prescribing habits (Hogg 2005). The designations determine how often the representative should call on a doctor as well as the focus of the visit. Tracking is also used by some companies (e.g., Eli Lilly) to categorize doctors by personality type and attitude toward detailers in order to tailor messages to their particular type (Rhee 2009). In addition, Ranjan (2009) indicates that data mined information is utilized in pharmaceutical marketing to help identify the most profitable product(s) and determine the allocation of marketing funds. Without data mining, pharmaceutical marketers' understanding of the target market would be diminished, and the ability to tailor messages for individual physicians would be compromised (Hogg 2005).
Marketers are not the only beneficiaries of data mining. The information produced by data mining is disseminated to insurers, pharmacy benefits managers, academic researchers, health policy officials, law enforcement agents and even the general public. Ultimately, patients benefit from this process. When the FDA approved a new drug to treat a debilitating form of epilepsy in children, the pharmaceutical company that produced the drug was able to use prescriber identifying information to locate the 1,300 health care providers, from a universe of 10,000 to 12, 000, most likely to treat that form of epilepsy. That would not have been possible if statutes like the one in Vermont had been in effect in other states (Brief of Respondents IMS Health Inc. et al., 2011).
Pharmaceutical Sales Representatives and Detailing
The face-to-face visit with the doctor by a sales representative promoting prescription pharmaceuticals and therapeutic devices is known as "detailing." Armed with the knowledge of what, how and when the physician prescribes a drug, the representative can attempt to enhance the effectiveness of each visit by selecting the most appropriate and persuasive information to share. "Traditionally, detailing involved a personable, well-dressed professional visiting a physician's office sharing news, clinical results and product samples with a physician. The meetings involved a good amount of banter and were part of a cordial relationship" (Young 2011). As experts on their drugs, the representatives must possess knowledge of and confidently answer questions about everything from insurance coverage to "the mechanisms of disease, routes of drug metabolism and excretion, side effects, indications and contraindications" (Rhee 2009). Detailing has evolved into a more sophisticated and personalized interaction in which compelling messages are tailored for specific physicians based on the information gained from data mining prescription records.
Doctors benefit from the information shared by detailers. Many physicians like the visits from detailers and enjoy lunch for the office compliments of the pharmaceutical company (Informationweek 2011). The system provides a two-way channel of communication, and recent medical findings can be learned through a quick visit with the detailer. In addition, physicians have a resource for understanding drug interactions, side effects, and insight into difficult cases. For example, using data mined information allows the drug company and the detailer to pinpoint physicians whose patients might benefit most from a new drug for a rare disease. "Without the data, you might visit 1,000 physicians to identify the 10 whose patients might most benefit," Mr. Frankel [Vice President for External Affairs at IMS Health] said. "With the data, you would go to the 10" (Singer 2011, p. 3). Today's physician likely looks to detailers to impart new knowledge that can improve patient care more than the opportunity to get a free lunch. Catering, however, continues as an incentive for the doctor's attention in spite of ethical concerns.
Resistance to Detailers and Data Mined Information
Pharmaceutical companies are supportive of health information companies that data mine physicians' prescription records and defend their right to collect data and analyze it for marketing purposes. However, resistance to sales representatives' detailing and their utilization of data mined information has grown over the years for several reasons. According to Young (2011), the first wave of resistance by doctors began in the late eighties and nineties. Some doctors believed that patient care and the doctor's integrity were compromised by accepting the detailer's complementary items such as pens, pads, lunches, and more...