Kaplan and colleagues [1] add a substantial brick to the edifice of potential self-remediation by showing the importance of physician-initiated participatory decision making in general medical settings. As part of the much-celebrated Medical Outcomes Study (which has been recognized with the Glaser award of the Society of General Internal Medicine and the Pew Primary Care Award, among others), Kaplan and colleagues asked three questions: "If there were a choice between treatments, would this physician ask you to help make the decision? How often does this physician make an effort to give you some control over your treatment? How often does this physician ask you to take some of the responsibility for your treatment?" They found that physicians who scored lowest (in the bottom quartile) according to their patient's ratings (with myriad correct and confusing corrections and controls) lost one third of their patients during the study period; the physicians most likely to encourage patients to participate in their own care (those who scored in the top quartile) lost only 15% of their patients. Physicians who had had primary care training or training in interviewing skills, whose practices were lower in volume, and who were more satisfied with their own autonomy were more likely to be in this higher participatory decision-making group.

Other data support and expand these findings and provide a context for them. Kaplan and colleagues [1] cogently assume that participatory decision-making style is a marker for higher-quality interpersonal care. Recently, we have come to view the interview as a core clinical skill and a major determinant of quality of care, including such important outcomes as physician and patient satisfaction, biological outcome, quality of life, cost-effectiveness, and cost-efficiency. The medical interview is conceptualized as having 12 structural elements, such as greeting, surveying problems, negotiating a priority problem, and closing [2]. It also has three functions: 1) determining and monitoring the nature of the problem, 2) developing and maintaining [and sometimes concluding] a therapeutic relationship, and 3) patient education and implementing plans. Each of the structural elements and functions is associated with specific behaviors that, when properly executed, relate to better process or outcomes. Knowing these structures and functions and possessing the skill to execute them well is increasingly a standard of care.

A robust body of literature relates to the importance of the second function, developing and maintaining a therapeutic relationship. Patients disclose more information to physicians they trust, and trust relates more to behavior during the interview than to anything else (perceived competence, an element of trust, also derives largely from behavior during the interview). The outcomes of patient education—recall of and adherence to diagnostic and therapeutic plans—also relate in larger part to trust than to the perceived risks associated with noncompliance [3]. Less critical outcomes, perhaps, such as patient satisfaction and return rates, also relate to the use of excellent and congruent (that is, consistent with the real self of the practitioner) relationship techniques. These have been shown to include the use of reflection ("I hear that you are angry"), empathy ("You seem to be feeling sad"), understanding ("I can understand how you might feel that way"), legitimation ("Under the circumstances, feeling that way makes a lot of sense"), and support ("I want you to know that I care about you and I plan to work with you on this until we get it resolved"). The most common single complaint of patients, one that is sometimes true and sometimes not, is that physicians don't listen to them. By this, they mean that physicians interrupt, don't pay attention, and seem preoccupied, and that even when they do listen, they don't understand. Thus, common courtesy, checking understanding, summarizing main points, and asking "what else?" (one of the most important questions in medicine) add up to improved relationships and, thereby, to improved outcomes [4].

One of the most important acts of the physician has been dubbed "activation of the patient," which means involving the patient actively in his or her own care. Kaplan and colleagues have documented improved biological outcomes in patients with hypertension, diabetes, and peptic ulcer disease when these patients are taught simply to ask their physicians questions [5, 6]. Hogbin and Fallowfield [7] show that giving patients an audiotape of the "bad news" session in which they learned that they had breast cancer led to improved quality of life. How these effects occur is not known and is the subject of considerable speculation. The answer may be partly related to patient empowerment or self-efficacy (persons who feel more effective become more effective) or to the social support provided by the physician who is perceived as caring. It has been shown extensively [8-10] that persons with greater social support do better in managing diabetes, depression, asthma, the development of arteriosclerotic disease, and more.

Various objections to the findings of Kaplan and colleagues will be heard. How valid is their measure? Would it not have been better to have actually observed physician behavior to document style and relate it to questionnaire response? Might their results be a "halo effect," a proxy for the nice physicians with more time? We have observed actual physician behavior (Roter D, Stewart M, Putnam SM, Lipkin M, Stiles W, Inui T. Personal communication) and have found, indeed, that interaction styles that emphasize patient participation occur about as often and correlate about as well with satisfaction as they do in the study by Kaplan and colleagues [1].

More importantly, so what? Can physicians change their behavior? Isn't physician behavior a matter of personality and nature, or the arteriosclerotically engraved habit of a lifetime? Evidence clearly shows that the communication styles of physicians can change [11], that they improve with proper teaching [12, 13], and that such changes and improvements endure [14]. The key seems to be a proper teaching model that works not only on knowledge or skills but also on knowledge, skills, and attitudes in an integrated manner [15]. There is no quick fix here, but if a substantial percentage of a group's physicians can lose 50% fewer patients, if fewer physicians burn out or quit (currently, physicians in health maintenance organizations stay about 3 years), and if continuity is thereby enhanced and fewer defensive tests or longer first interviews are needed, then investment in communication skills is worthwhile. Also, other factors argue for such investment in communication skills by individual persons and groups; these factors include more efficient and complete diagnosis, less testing and redundancy of procedures, greater physician and patient satisfaction, improved outcomes of care, and fewer malpractice lawsuits. The necessary training is increasingly available through organizations such as the American Academy on Physician and Patient, the Society of General Internal Medicine, and the American College of Physicians. I believe that in the future, regular work on interviewing skills will be an expected part of internist self-maintenance just as it is already part of performance monitoring in several managed care organizations.

How, then, do the findings of Kaplan and colleagues relate to getting back up on the horse and down from the mountain? Simply put, changing our own interviewing skills for the better improves our own and our patients' satisfaction with our work, improves our quality and efficiency, and increases our patient retention. Some may wish this to be an exaggerated or absurd claim. But, after all, medical encounters are the thing we do most as practicing physicians: We have between 160 000 and 300 000 of them in our professional lifetimes. Not surprisingly, if we want to feel better about what we do, if we want to improve our quality, we must attend to what we do most. We now know how to do this. It is time to start.

Mack Lipkin Jr., MD

New York University Medical Center

New York, NY 10016