Sounding out the diagnosis

Editor's note: Zarif Islam is an analyst at KMK Consulting Inc., a Morristown, N.J., research firm.

Ask anyone in pharmaceutical marketing about their top challenges and most will undoubtedly include the difficulties associated with understanding the physician’s treatment decision process. From a marketing standpoint, this critical task comprises outlining what the decision-making process is as well as identifying points at which it can be influenced in order to drive particular behaviors. This is arguably one of the most fundamental responsibilities of health care market researchers. Is treatment largely a function of patient characteristics? Is it more likely to be shaped by brand perceptions or habit? Or can treatment decisions be better explained by insurance coverage and access?

These questions are not easy to answer but they lie at the heart of marketing research efforts to generate relevant insights that ultimately enable manufacturers to better address the needs of physicians. The challenge, of course, is how best to uncover these insights.

Broad contours

As researchers we rely on a range of quantitative and qualitative methodologies to dissect the physician decision-making process. Qualitative techniques are most typically found at the front lines of these efforts. Although quantitative research utilizing structured questionnaires certainly plays a key role, it is generally through qualitative that the broad contours of physician decision-making are explored, hypotheses developed and subtleties in physician thinking unearthed. One underutilized but very powerful qualitative research technique for penetrating the physician’s decision-making process is cognitive interviewing.

Cognitive interviewing, also known as “think aloud,” involves having research participants “speak aloud any words in their mind as they complete a task” (Charters, 2003). This method has been proven to provide useful information regarding decision-making in a health care setting (Aitken, 2000) as it allows researchers to identify what information is given most importance by physicians when treating or managing a patient and how that information is used to reach specific prescription choices. Researchers then make inferences about the reasoning processes that were used during the task based on “direct verbalization of cognitive processes” (Fonteyn et al, 1993).

Given the practical difficulties of conducting cognitive interviews with physicians as they treat real patients, health care researchers will sometimes approximate cognitive interviewing through one of two means. The most common route is through physician review of anonymized patient charts. The less-commonly applied approach is through having the physician engage with actors who are hired to play the part of patients. 

In the first approach, physicians are asked to gather a few charts of patients they have previously treated. Over the course of the interview, physicians are asked to explain their rationale for the treatment choices they made for the patient in each chart. While this approach is useful in that it avoids the vague generalizations that plague qualitative interviews shorn of any specific patient context, it falls short of true cognitive interviewing. When using patient-charts as a basis for discussion, physicians are temporally removed from their thought process as it occurred when they saw the actual patient. The passage of time, compounded by failures of memory, can impair their recall of their actual experience with the patient. In addition, the physician may be prone to post-hoc rationalization designed to maintain the appearance of unimpeachable and professional decision-making. Perhaps most importantly, it is often difficult for the interviewer to elicit retrospectively the physician’s doubts, perceptions of risk and consideration of alternatives when the future was uncertain and the outcome of treatment unclear.

In the second, less-frequently employed approach, trained actors are brought in for the physician to “treat.” As the physician goes through his interaction with the patient-actor he is asked to verbalize his thoughts. This methodology allows researchers to pinpoint what kind of strategies are being used by physicians in real-time and how they are acquiring and using the information they are being presented with. With proper training of the patient-actors, specific scenarios of interest to researchers can be created. Physician reactions to subtleties of body language and verbal communication can also be examined.

Like the review of patient charts, however, this methodology comes with its own set of challenges. Hiring and training actors can be expensive and time consuming – particularly if the research is to be conducted across central locations in multiple geographies. To the extent that the number of actors deployed is limited, so too are the patient scenarios that can be presented to the physician, unless the realism of the exercise is to be compromised by having a single actor appear as two or more distinct patients. 

Addresses the shortcomings

Simulation offers another, potentially superior approach to leveraging the power of cognitive interviewing with physicians as it addresses many of the shortcomings of patient chart review and patient-actor methods. The use of simulation to conduct cognitive interviews is not new and dates back at least to the 1980s, when it was demonstrated to be an effective means of exploring problem-solving strategies among physicians. In these early exercises, researchers introduced patients to physicians via paper or audio-visual presentations and asked physicians to think aloud as they dealt with the management of these patients. Among the virtues of this approach highlighted in the literature is that simulation allows investigators “to approximate the clinical environment while controlling for other variables found in real-life situations” (Fonteyn et al, 1993). Researchers also demonstrated the ability to integrate treatment outcomes by emulating longitudinal changes in the patient presentation, thus adding another dimension of realism to the simulations (Barrows et al, 1982). Validation studies of these approaches have also been conducted and suggest that the findings from such simulations reflect real-world behavior (Holzemer et al, 1986). 

Since these approaches were published, computer hardware and software improvements have brought us to the point that the gap between artificial laboratory tasks and the real world can be further reduced. Leveraging these technological capabilities into next-generation digital simulation platforms built around the treatment of virtual patients present intriguing possibilities for broadening the practical use of cognitive interviewing with physicians. Early work we have done with a novel virtual patient simulator affirms the potential of simulation-based cognitive interviewing in probing physician treatment decision-making.

This particular simulation platform is built around the examination and treatment of simulated patients. It resembles a simple, electronic medical record system.

The simulation is divided into three tasks: 1) the patient examination, 2) treatment selection and 3) scheduling of follow-up visits. In examining the patient, the physician can access information regarding patient history, clinical presentation and laboratory tests. Commentary and concerns expressed by the virtual patient or virtual caregiver can also be integrated into this task. All information is then programmed to update on subsequent “visits,” allowing researchers to explore physician response to treatment outcomes. In the treatment choice screen, physicians select from a range of actions, including prescribing of medication, surgery, referral or ordering of diagnostic tests. Once one or more actions have been selected, the physician then indicates when the virtual patient should return for a follow-up visit. 

Our pilot tests of this simulation platform were conducted through central location, one-on-one interviews in which the physician respondents were given an iPad loaded with the virtual patient simulation. As the physicians engaged with the simulation to examine and treat the virtual patients, they were asked to think out loud, as in standard cognitive interviewing protocols. This process resulted in useful verbalization of thought processes, as illustrated in this quote from a physician treating a virtual patient suffering from schizophrenia:

“I’m going to go back to the patient because I want to review her history. She’s currently on Haldol and Lexapro but she’s quite, she has a lot of positive symptoms of psychosis and these are the symptoms that are typically most responsive to anti-psychotic medication. So I have to assume that either the dosage of medicine she is on is inadequate or, or she is non-compliant with medicine, or the medicine is just not working for her. So, I have a couple of different possibilities there that are going through my mind…

“I’ve reached the conclusion now that the problem is non-compliance. We have a medication that has helped her in the past so we want to try giving it to her and let’s say he [the caregiver] comes back in a week or something, says he’s giving it to her orally and she does indeed, she is indeed somewhat better. Then okay, I would propose to them that we give this as a once monthly injection. Let’s give the decanoate as, let’s try Haldol decanoate, that way we don’t have to take it every day.”

We can see that the physician is clearly “treating” a patient in real-time while considering potential outcomes and consequences, much in the way she would in a real practice setting. Through the course of the research, physicians like this one tended to speak in the present tense. They also tended to voice uncertainty and consideration of differing treatment options – along with their perception of the pros and cons associated with these options. Our sense is that the digital nature of the simulation supported a “suspension of disbelief” in the artificial nature of the task, as well as greater sense of engagement, as physicians believed that their actions would have consequences for the virtual patient in subsequent visits. 

This idea that computer-based simulation is perceived as more realistic has also been demonstrated in other studies conducted in health care settings (Zary et al, 2006). In sum, we have found that digital simulation in cognitive interviewing can produce valuable insights into physician decision-making processes more cost-effectively than patient-actor approaches and with potentially greater subtlety and depth than is typically obtained through retrospective patient chart reviews. 

A solid foundation

Given the importance of understanding medical treatment decision-making – in this case, from the perspective of the physician – health care marketing researchers may want to reexamine the potential of cognitive interviewing. As outlined here, many of the traditional obstacles to effective cognitive interviewing with physicians can be overcome through the use of digital simulation support. Moreover, while the use of digital simulation can be considered innovative, it rests on a solid foundation of historical research and experimentation in the world of health care. Accordingly, it is a low-risk approach positioned to address some of the challenging research issues confronted by health care market researchers.