Turning Dialogue into Data
Contrast this cost-conscious doctor’s animated mission for a Wal-Mart solution with the following discussion between a patient and primary care physician:
Patient: Um, I got a letter from the insurance company about the Zoloft I’m taking, and they suggest...
Patient: Generic, yeah. So I don’t know, I have enough...
Patient: I have two more refills on that.
Patient: So do you want to switch it or...
Doctor: Well, basically it’s a cost issue. Now that there’s a generic available they are going to start charging you the next level up for the name brand.
Patient: Right. It’s not that much difference for me.
Doctor: Well, I probably… I mean, unless it’s… if it’s really not, not going to be that much of a difference… if it were me, I would probably stick with what I have been on.
Doctor: Because by law a generic can vary 20% either way.
Doctor: So as long as, you know, it is not a significant financial hardship I would say if it was me I would stick with the...
The physician in the above example is a pharmaceutical manufacturer’s dream but a managed care organization’s nightmare.
One might chuckle that a patient would think a pharmaceutical manufacturer would send a letter notifying patients that its prescription has gone generic. While this is a small insight, the real value of capturing a patient-physician dialogue on this topic is to get unfettered insight about messages/mediums/sources that trigger a patient to ask, act and achieve therapeutic goals. Such insight may now be possible through a research technique known as “computational linguistics.”
Facilitating the Rx exchange
The currency we deal in is the prescription, and the physician is still the ultimate determinant of the type and quality of the script, very often exerting the most influence on patient treatment. If we think of the manufacturer as dealing in one currency—say Euros—and the patient in another currency—say US dollars—we could think of the doctor as the foreign exchange broker who facilitates the transaction between manufacturer and patient. The doctor could offer favorable exchange terms and make the Rx very desirable for the patient. On the other hand, unfavorable terms would result in very little Rx either taken up by the patient, or used once the prescription is taken home. So how do we understand what goes on at the “foreign exchange” counter? In short, what do we know about the actual discussions that take place between physician and patient?
While industry has done a good job tracking transactions, the dialogue driving these transactions has been shrouded in mystery. What happens between the patient and physician behind the examining room door is often as closely guarded as the confessions within a confessional. In isolation we have a good understanding of all market influencers, but the intersection of these forces during the physician-patient dialogue has been the last bastion of research to elude our reach. This industry has tried very hard to crack that patient-doctor confessional, and computational linguistics has emerged to make it possible, with great marketing implications.
Breaking through the confessional
Traditional forms of research—in-depth interviews, focus groups, observational research—help us get there and are used regularly to hear either party’s takeaway. These are good practices, but they usually capture data in isolation and sequentially. By capturing data at the point of care, however, we are able to minimize or even eliminate the “replay” factor and capture the actual discussion without the frailty of human recall and subjectivity of replay interpretation.
The HIPAA-compliant methodology is intriguing. Physicians who agree to participate in this research use a small, unobtrusive digital recording device to record their conversations with a few patients each month. Along with the recording, physicians provide electronic patient chart information and attitudinal data with regard to the interaction. An example of what the interactive chart captures can be seen in Figure 1.
The information is blinded to names and any other personal identifying information about the patient or physician. By merging the patient chart data with the dialogue of the actual intervention, analysts can now answer the difficult “why” questions that typically accompany the “what” answers provided by conventional forms of research.
Immediate takeaways from leveraging this technology that have previously dogged product managers include:
■ How is my brand being described and positioned by physicians in the patient dialogue?
■ What is said about my competitor’s brands by the patient and by the doctor?
■ What are the discussion topics that influence treatment choice? (And does my sales aid even address any of these?)
■ What conversations occur regarding formularies? (Do doctors understand price or price/value?)
■ How do physicians handle issues of prescription coverage and co-pays? (What can I do to change the texture of this discussion?)
■ Is the patient engaged in the discussion? What tools or support materials are referenced?
■ What words do patients use to communicate their condition?
■ What do patients say about their current medications?
■ How are patients diagnosed in the first place?
■ Do physicians follow therapeutic guidelines or do these vary by therapeutic category or severity of disease?
■ Do physicians alter their approach and behavior for different patient types of the same condition?
We typically rely on a range of tracking tools to measure how well any one of our communication channels (personal selling, interactive e-messaging, tele-detailing, print, etc.) achieves a desired response from a physician. The missing link is understanding how these communication efforts become a part of the interaction between our two primary customers, physicians and patients. By analyzing a number of recordings, analysts can isolate specific patient and brand scenarios and answer the following questions:
■ Which of the brand messages are sticky and actually get conveyed to the patient and in what context?
■ Were the messages patient-initiated or physician-driven?
■ How much time is spent on each sub-message?
■ How were the messages delivered and how were they received?
■ What messages and/or topics are most engaging to the patient?
■ What are the patterns or conditions one can identify in which the marketing message-to-be is played back with the strongest accuracy and credibility?
How might these insights prove useful? The words used in sales aids, journal ads and conventions are a world apart from the words that patients use. Finding the right mix is critical, as the balance of information and “power” has shifted radically since the 1997 FDA Modernization Act catalyzed the pace of DTC promotion. While patients are constantly stimulated to act, research and evaluate new health information, their training does not expose them to the typical physician language. Our messages are still heavily couched in physician terms.
Clearly, the ability to capture their language can vastly improve message reception and—perhaps in the long term—compliance.
Computational linguistics can be used to identify topics important to physicians and patients and the factors which drive changes in behavior (see Figure 2).
Verbal common denominators
An ability to identify the common words used by physicians and patients during office interactions may vastly improve core positioning and messaging (see Figure 3). Furthermore, this tool may be used to dissect differences between patient types or ethnic groups. Used within the diaspora of America, pharma marketing tools and messages may provide less of the intended impact unless tailored for the intended audience. The ability to capture the dialogue in the patient’s own words is increasingly critical to maximize targeted media, leverage niche markets and sharpen one’s competitive edge.
For example in the rheumatoid arthritis (RA) world, patients may talk about pain quite differently from someone experiencing depression pain. In RA, patients talk about “bottoming out,” whereas in depression they talk of being “overwhelmed.” In psychiatry, patients and physicians commonly discuss “stability,” yet this word and theme is missing from many large-budget physician and patient promotional campaigns.
This technique has finally brought the examining room into the realm of research, without breaking the rhythm of the actual discussion. As researchers create a greater database from this methodology, they can provide ever more relevant platforms and messages for all stakeholders.
Carolyn Choh Fleming is a professor in the graduate and undergraduate pharmaceutical marketing and international marketing departments at St. Joseph’s University. The author would like to thank Jeff Kozloff, president of Verilogue Inc. (North Wales, PA), for providing the data for this article. Choh Fleming has no professional ties to Verilogue