Ginsburg, Shiphra MD, MEd; Bernabeo, Elizabeth MPH; Ross, Kathryn M. MBE; Holmboe, Eric S. MD
Previous research has shown that context has a critical influence on individuals’ behavior and is essential to understanding lapses in professionalism.1 As accrediting bodies and organizations have, in recent years, begun to mandate the teaching, learning, and evaluation of professionalism, much of the research in this area has focused on the education setting.2,3 However, the medical community’s understanding of professionalism in practicing physicians is limited.
Papadakis and colleagues’ research4,5 has been instrumental in developing an understanding of what sorts of disciplinary actions physicians face in practice, but very little is known about why these lapses occur. Research by Campbell and colleagues6 has illustrated that gaps exist between what physicians endorse as important to professionalism and how they actually react in practice. In one study, the reasons physicians cited for not acting in accordance with their own self-endorsed values (e.g., reporting an impaired colleague) included a belief that someone else would take care of the issue, a belief that nothing would change, and fear of retribution.7 In another survey study, women, nonwhite, and foreign graduate physicians were more likely to disapprove of relationships with patients that may represent conflicts of interest, indicating that context, including the respondent, affects views on professional behavior.8 These are important findings, but by its very nature, survey research can provide only limited glimpses into the myriad competing factors that may contribute to less-than-ideal professionalism in practice.
Qualitative research offers an alternative, and previous researchers have used it extensively to examine professionalism, but largely in medical education settings. As an example, several investigators have used standardized “professional dilemmas,” in written or video format, to explore how students conceptualize appropriate or inappropriate behavior, and how they come to decisions about how they might act.9–12 These studies have involved medical students and attending physicians in education settings; their findings suggest that individuals are motivated to act according to principles of professionalism (e.g., patient care, disclosure, honesty), on the basis of “affect” (feelings or beliefs), or by considering potential consequences of their actions (to the patient, others, or self). The investigators conducting these studies explicitly developed scenarios to be relevant to medical students in clinical settings, and they found that the scenarios were potent triggers for discussion and reflection. We felt that such an approach might be equally useful in practicing physicians.
The purpose of the current study was to explore how and why practicing physicians respond to a set of professional challenges relevant to the context of practicing in the specialties of internal medicine. Our goal was to develop a better understanding of how physicians view these challenges, how they come to decisions about how to act, and what factors influence these decisions.
Although a more complete description of the development of the scenarios is available elsewhere,13 we also highlight it briefly here.
We chose the Physician Charter on Professionalism’s definition of professionalism as a basis for our work because over 100 professional organizations have endorsed or adopted it and because it has direct relevance to practicing physicians.14 We then used the aforementioned studies by Papadakis and colleagues4,5 and by Campbell and colleagues6,7 to select domains in which disciplinary actions against physicians are most common (e.g., negligence, physician impairment, or incompetence) and areas where gaps exist between the charter and how physicians report (or propose) behaving in actual practice. We then created a content blueprint to guide our scenario development. Over several weeks, we developed and refined our scenarios in an iterative process, after which we subjected them to cognitive testing with four internists. In a single, confidential, one-on-one interaction, one of us (E.B.) guided each internist through a reading of and response to each scenario. Specifically, we probed for understanding of each scenario, for relevance to practice, and to determine whether the intended key features were raised. Brief descriptions of the scenarios are available in Box 1.
We chose focus groups to examine professionalism among practicing physicians (as opposed to one-on-one interviews) as a means to stimulate discussion, debate, and reflection amongst participants. An external agency (Focus Suites, Philadelphia, Pennsylvania) that maintains an independent database of physicians handled recruitment. In March 2011, we contracted with the agency to recruit, by e-mail, five focus groups of up to 10 practicing internists each from the Philadelphia area. The agency recruited two groups of primary care physicians (one hospital-based, one ambulatory) and two groups of specialists (one hospital-based, one ambulatory). These were homogeneous in terms of practice type in order to stimulate relevant, safe discussion and so that we could explore potential differences between groups. The final group was mixed, consisting of primary care and specialist physicians who worked in ambulatory and hospital settings.
The same investigator who conducted the cognitive testing interviews (E.B.), plus one other member of the research team (S.G. or K.R.), facilitated each focus group. We selected five to six scenarios for each group to reflect that group’s makeup (e.g., hospital-based primary care physicians, specialists in ambulatory practices, mixed). During each focus group, the facilitators presented participants with one scenario at a time and asked them what they would do if that particular situation happened to them. The facilitators guided the discussion to ensure a broad range of responses and to probe respondents, when necessary, to explain why they proposed or discounted certain actions. The facilitators also challenged the participants to consider what, if anything, might change their minds about a proposed action once it had been stated.
In the tradition of grounded theory, in which data are analyzed as they are collected, we made changes to some of the scenarios based on each focus group’s reactions.15 In some cases, we developed different versions of a scenario to reflect different practice types (hospital versus ambulatory); in some cases, the original scenario appeared vague or did not resonate with participants, so we altered it; and, in some instances, participants volunteered a more relevant scenario along the same topic.
We collected demographic data through paper questionnaires that participants completed immediately before each focus group. All focus group discussions were transcribed, and the focus group members were anonymized. The Essex Institutional Board provided ethical approval, and all participants gave informed consent to participate and be audio-taped. Participants received a small financial incentive.
We used a constructivist approach to grounded theory; that is, we brought sensitizing concepts to the analysis while we conducted an open, inductive analysis.16 In this case the sensitizing concepts arose, a priori to analysis, from a framework derived from the literature9–12 (as described above), in which participants’ motivations to act are based on principles of professionalism, internal affect, or potential implications of their actions.9 For purposes of this report, we focused on what appeared to be the guiding principles for behavior, along with factors that appeared to modify suggested behaviors. Two authors (S.G. and E.B.) conducted the analysis by individually reading, in depth, one transcript at a time, and using a method of constant comparison in an iterative fashion. They discussed emerging themes and codes (labels assigned to themes or subthemes) on a frequent and regular basis and as each new transcript was read. They refined the coding in an iterative process, resolving disagreements through extensive discussion and consensus.
We used NVivo software (version 9.2, QSR International Pty. Ltd., Doncaster, Victoria, Australia) to organize and code the transcripts, and to keep records of modifications as coding proceeded. During this process, the extended research team met at several points to review and challenge the emerging analysis. Once coding was complete, we met once more to review, challenge, and approve the final framework.
In May 2011, we conducted five 90- to 120-minute focus groups, consisting of, collectively, 40 practicing internists. Overall, 32 participants (80%) were male, and 17 (43%) practiced general internal medicine. They were experienced physicians, having been in practice a mean of 18 years (standard deviation [SD] 8.1) and having spent an average of 94.7% (SD 7.7) of their time in direct patient care. On average, the participants had spent 65% (SD 25) of their time in private practice.
All of the scenarios were effective in stimulating discussion and debate, resulting in an average of 45 codes applied per scenario (range 34—77). The participants perceived the scenarios as authentic; they endorsed facing identical or very similar dilemmas on a regular basis. Through our analysis, we identified two main sets of factors that appeared to underpin participants’ responses to the dilemmas: guiding principles and modifiers. As our findings will illustrate, these did not always represent mutually exclusive categories because the application of differing principles could also at times modify responses. Further, participants often referenced many principles and modifiers together in a single response to a given scenario, and at times the principles and modifiers could be seen to conflict with each other. Because of space constraints, we cannot explore each theme in detail; rather, we have chosen several to illustrate key findings.
We have provided numbers with Table 1 and Supplementary Digital Tables 1 and 2 (http://links.lww.com/ACADMED/A110) to give the reader a sense of how commonly and in what patterns the individual themes and codes arose, but these counts should not be interpreted as indicators of any theme’s importance. Participants may have mentioned one theme less frequently than another; nonetheless, that less-discussed theme may be more salient to participants.
We identified 14 guiding principles through our analysis. Table 1 lists these principles along with their definitions, and Supplementary Digital Table 1 (http://links.lww.com/ACADMED/A110) presents these as major (and sub-) themes along with their prevalence by scenario. With one exception, each principle arose in all five focus group discussions to some degree (concern for colleague did not arise in the mixed group).
One of the most pervasive themes across all scenarios related to patient welfare, which, for our participants, encompassed attending to matters of clinical and patient care, putting the patient first, and relieving patient stress and anxiety. Participants commonly raised patient welfare issues in their discussion of Scenario 3 and Scenario 4, both of which focused on what to do in the case of a physician colleague who is suspected of some sort of impairment. One participant stated, “From my perspective, your responsibility is to the patient, not to the colleague.” Another remarked, “You have an obligation to your patients to assure that they get the best care.” Importantly, simply articulating these principles did not necessarily lead the participants to suggest what might be considered the ideal behavior (i.e., in this case, reporting the physician colleague). Rather, patient welfare was often one of several principles the participating physicians considered and weighed as they crafted their responses. During discussions of Scenarios 3 and 4 in particular, participants also considered—among many other principles—the nature of their relationship with their physician colleague, local or state regulations, policies around reporting colleagues, and the potential nature of the perceived impairment. Each one of these factors, to any expressed degree, could have the effect of iteratively changing the nature of participants’ responses.
Also common was a theme we have referred to as keeping patients happy, and although it, too, related to patients, the focus here was not on clinical care. Two important subthemes emerged: (1) helpfulness or niceness and (2) protecting or supporting the practice or institution. The participants expressed a desire to be seen as nice or helpful particularly during their discussions of Scenario 5, in which a neighbor asks for medical advice: “For most of us, we probably went into medicine to help people. And I definitely have a sense of guilt if I don’t.” This theme also frequently arose in response to Scenario 8, which focused on which (if any) patients physicians might give their e-mail address to. Some respondents felt this additional access was helpful to the physician–patient relationship, as illustrated in the comment “Because people appreciate the added service.”
However, keeping patients happy could also take on a slightly different tone. Participants often noted the importance of building or protecting a practice or institution. For example, one participant noted in response to Scenario 2a, in which a patient insists on antibiotics, “One of the reasons that we end up yielding to the patient is the reputation of the institution. If you won’t do it, the patient is not going to come back.” Others noted that patients who are not satisfied may go “doctor shopping, and are going to find somebody else” including nonphysicians who will do what they wish. This theme was also prominent in discussions of Scenario 1, in which physicians must decide how to absorb a sudden bolus of new patients into their practice. Protecting or supporting the practice or institution also often arose with another set of guiding principles—namely, financial and reimbursement issues. Some were quite clear that running a business was not only reasonable but worthy: “If you’re in private practice, you’ve got to pay your bills.”
The principle of being available or accessible to one’s patients was also prevalent, especially, as mentioned, in response to Scenario 8, which is about providing e-mail access to patients. This scenario often evoked tension, as participants articulated a desire to be available when truly needed while not wanting to be inundated with more communications than they could handle.
The theme of whether or not to follow evidence-based medicine was particularly relevant to Scenarios 2a, 2b, and 2c, which we designed expressly to elicit a discussion on this topic. Participants often referred to the evidence-based guidelines that were relevant to each particular scenario, but they expressed significant difficulties in sticking to them. Sometimes they questioned the legitimacy of guidelines in general (one stated that “the evidence is only 15% of what we do”). Some participants expressed frustration either when guidelines were unclear or when more than one, perhaps conflicting, guideline existed for a particular diagnosis (e.g., mammograms for breast cancer screening). They also expressed frustration at areas in which the literature and evidence are frequently in flux (e.g., as with cancer screening). Many expressed trying their best to stick to guidelines and to convince and educate patients that they represent the best approach, although they frequently noted that doing this was futile.
As the analysis for guiding principles proceeded, we began to identify factors that appeared to modify how participants interpreted and acted on the principles (see Table 1, Modifiers). That is, participants articulated how they might treat situations quite differently depending on certain factors, the most important of which were the type of patient and the nature of the illness or request. Supplementary Digital Table 2 (http://links.lww.com/ACADMED/A110) depicts the interactions between the guiding principles and modifiers.
Regarding the type of patient, participants considered such factors as whether, in their perception, the patient was compliant or challenging, had simple or complex problems, and did or did not respect the physician’s time and boundaries. They also considered whether they personally liked or bonded with that patient. These factors were particularly important in response to Scenario 1 (admitting new patients to one’s practice) and 8 (providing patients with one’s e-mail address); respondents admitted to considering issues such as whether the patients were particularly challenging, time-consuming, or disrespectful as they formulated their approaches. Type of patient was also a major consideration in deciding whether or not to prescribe antibiotics for what was presumed to be a viral illness (Scenario 2a), as illustrated by the following:
They’ve gotten to you, psychologically, in a sense where they’re not respecting your opinion, no matter what you say. They want their antibiotics. There’s a sense of entitlement. They’re disrespecting the professional that you are…. How are you going to handle this? Maybe in each patient you kind of know that particular individual…. You say to yourself, it’s not worth it. I’m going to get several calls. And you give in. You move to the next patient.
Similarly, participants were swayed in many instances by considering the nature of the illness, diagnosis, or request. For example, respondents distinguished between physical conditions and mental illness or addiction issues, especially when deciding whether to take on new patients, whether to treat a friend or neighbor, or whether or how to intervene if a colleague seems unwell. Regarding the last case, we created two different versions of a similar situation; in Scenario 3, the physician is “slowing down” cognitively, whereas in Scenario 4 the physician shows weight loss and a change in behavior. Unexpectedly, these two scenarios evoked quite different responses among our participants. They discussed legalities, policies for reporting colleagues, systems issues, and availability more often in their reaction to Scenario 3, whereas Scenario 4 triggered more concern for the colleague in difficulty. Even regarding Scenario 4, though, participants expressed hesitation and reluctance to approach a colleague if mental illness or addiction—rather than a “medical”—diagnosis was possible. Participants also seemed to draw a hard line at the issue of narcotics, treating requests for these sorts of medications and the patients who made these requests quite differently than requests for or patients requesting other medications.
Participants also often reconsidered their responses if the diagnosis was potentially more serious (e.g., cancer) compared with something more minor. Other modifiers included the participant’s relationship or role with a colleague in difficulty and the risk of danger or harm to a patient (especially the potential to miss a diagnosis). Still another “it depends” factor (especially in a case like Scenario 5) was familiarity or comfort with the particular request (e.g., did it fall within the physician’s usual scope of practice?). Of note, the physician’s relationship with the patient arose as quite an important modifier in discussion of Scenario 8; participants often made decisions about how much access they would grant patients depending on how long they had known the patient, whether trust had developed, and—as one respondent put it—if the patient had already “passed the ‘normal’ test.”
Interactive nature of principles and modifiers
Notably, as we alluded to above, some guiding principles served additionally as modifiers. For example, the principle of availability and accessibility served to modify responses to Scenario 3, when participants were discussing whether or not to continue to refer patients to the colleague who was slowing down. Consider the following exchange among three focus group participants:
It’s a very simple case.
Is it looking for a little polyp or a little GI bleed? No big deal. You might be able to. But, again, if it’s a bigger story, there’s more to the history, and what is it? Has he killed three people on the lab table, when he’s doing colonoscopies? Again, you need more facts.
You really do not refer to somebody who you feel is incompetent. If that’s the question, do you refer to an incompetent doctor just because he’s free, the answer’s no. Unless you’re in a rural town, I’m sure there are more than a handful of GI specialists that can see your patients sooner. So, if you know they’re incompetent, would you send?
Here the participants first reference modifiers related to the nature of the patient’s illness (simple versus complex, serious versus minor) and to the nature of the request (here a scope is considered a simple procedure). On the basis of these, this participant feels that continuing to refer might be okay. But then he or she considers the nature of the risk of danger or harm to the patient and would perhaps respond differently if the physician has “killed three people.” The next respondent picks up this potential mitigating information, articulating the guiding principle of patient welfare (do not refer to somebody who you feel is incompetent), and then quickly modifies his or her initial response by considering the availability of relevant specialists.
Another example is a response to Scenario 2c, in which a patient is requesting a non-evidence-based test. The dynamic interplay among several factors is articulated in rapid succession, by a single respondent. The participant begins to explain:
I’d tell [the patient] why I don’t think you need the scan. I’d explain it to him, and educate the patient. If the patient’s still insistent that they want the scan done in three months, first of all, if I say “no,” they’re going to go doctor shopping and are going to find somebody else to do [it].
We coded this portion of the participant’s response under the principles of transparency and informed choice as well as keeping patients happy, in this case to protect the practice. The respondent then went on to clarify that “the patient’s not a stupid person, nine times out of ten. And they’re probably going to circumvent the guidelines on their own,” reflecting a sense that all of the explanations provided will ultimately be futile. Next, the participant continues,
But if you lay the cards on the table and do the best you can, and the patient still insisted, I guess you have to fall back to the position that not having the scan’s going to adversely affect the patient’s mental health, maybe, that they’re going to [go] nuts, that they’re going to die of a cancer.
Now the physician is expressing a belief that doing the scan may outweigh the desire to follow evidence-based medicine, but for a different reason—to reduce patient anxiety (coded under patient welfare). Then, toward the end of his response, this physician shifts focus and references yet a third set of principles, related to a fear of getting sued:
Chances are, I don’t know what I’m doing anyway. And there’s a good chance that the stinking thing will turn out to be a cancer, in a year and a half, and I’ll get sued. So, I think I’d rather mitigate and choose my risks, and send the patient for the unnecessary scan.
This lengthy response also demonstrates a physician’s consideration of the modifiers related to the nature of the illness (serious versus minor), and the risk of danger or harm to the patient.
These examples illustrate both the complexity of these interdependent guiding principles and modifiers and the reality that even principles that seem like hard and fast rules are highly malleable and can easily be broken. Further, they highlight the difficulties physicians may have in trying to delineate which factor(s) is (are) more or less important in a given case, or which may be more responsible for a given decision.
Finally, making direct comparisons between groups (e.g., primary care versus specialists, or hospital-based versus ambulatory practice) is difficult because of the study’s design. We used different scenarios (or different versions of the same scenarios) for different focus groups, and each scenario or modification evoked different sorts of discussions; however, we can see key differences in participants’ reactions to the scenarios that were presented to more than one group. Consider the responses to Scenario 1, for example, which we presented to two ambulatory-based groups (one composed of primary care physicians, the other composed of specialists). The scenario sparked much discussion in the primary care group, whose participants debated issues of availability and accessibility, reimbursement, and type of patient at great length (59 codes applied); however, in the specialist group the discussion was much more brief, focusing quickly on reimbursement and relevant expertise (12 codes applied). Responses to Scenario 8 serve as another example. We presented the scenario to both of the primary care groups and both of the specialist groups, evoking similar degrees of discussion (in terms of quantity, at least; we applied 114 and 128 codes, respectively). The discussions, however, differed qualitatively. The primary care groups showed somewhat more concern than specialists about availability, keeping patients happy, and efficiency, whereas the specialists focused a bit more on patient welfare, their relationship with and potential risk to the patient, and work–life balance.
It is important to develop a more in-depth appreciation of the professional challenges that physicians face in their day-to-day practices. Doing so will help the medical community understand why physicians may (or may not) endorse elements of the physician charter and—importantly—why they may endorse them in theory but not act in accordance with them in actual practice. Our study adds significantly to this discussion by exploring how practicing internists respond to a series of crafted scenarios depicting professionalism dilemmas that our participants found to be realistic, common, and challenging. Physicians were influenced by a number of guiding principles which were subject to modification, depending heavily on the situation and the players involved. Thus, focusing not just on behaviors but also on the context in which they occur is critically important for understanding professionalism in practicing physicians.
One of our key findings is that although physicians frequently made reference to principles to guide their behavior, their responses were often quite malleable. We noted that participants’ responses shifted depending on their consideration of other factors or on discussion and challenges that arose from the group or the facilitators. These findings support the current discourse of professionalism as a complex adaptive system,17 in which multiple interdependent factors operate simultaneously. These systems, considered to be “moving targets,” change and reorganize to adapt to the problems imposed by the surroundings.18 We observed that our participants expressed few hard and fast rules, and even where absolutes appeared to exist they were admittedly broken on numerous occasions depending on specific circumstances. “It depends” was a ubiquitous refrain.
In parallel to previous work studying medical students’ and academic attending physicians’ responses to professional dilemmas,9,12,19,20 the principles and modifiers presented here can be classified as avowed, unavowed, or disavowed. For example, the principle of patient welfare or the modifiers concerning potential risk of danger or harm to a patient can be considered “avowed”—that is, what physicians are supposed to consider in the face of a dilemma. Avowed factors align with what physicians profess. Other factors might be considered “unavowed”—that is, factors that are not explicitly endorsed but are apparently widely recognized and implicitly condoned (e.g., the principles of managing time and efficiency in one’s practice, keeping patients happy to protect or support one’s practice or institution). However, several factors or principles can be considered “disavowed”—that is, principles the members of the profession would eschew (e.g., choosing patients on the basis of their ability to pay, or avoiding [or treating differently] patients who are challenging, less compliant, or disrespectful of time and/or boundaries). Some researchers have found this avowed–unavowed–disavowed framework to be applicable in other practice settings,21 and it represents one potentially useful approach toward reflection, allowing physicians to consider their responses in light of current accepted standards.
The finding that physicians may be swayed in their decision making by nonmedical factors is certainly not new. However, these influences are usually ignored in discussions of professionalism, as if physicians are, in their clinical settings, not also human beings and, in that space, not subject to the same sorts of biases as everyone else. Our findings that nonclinical factors influence physicians align with those of a recent study reporting that physicians’ judgments about the complexity of their patients were often more influenced by behavioral and social factors than by standard scales of illness complexity.22 So the point is not that physicians’ assessments and reactions are “wrong” and need to be fixed; rather, these reactions may be normal. Most people would presumably agree that physicians should be allowed to make a living or to prefer friendly patients who adhere to their advice.23 What is more important is how these factors are weighed and balanced and how they interact in a given situation to ultimately produce professional behavior.
Of course, from this sort of study we can know and report only what physicians say they would do, not what they would actually do in practice. As illustrated in our findings, what a physician actually does may have many competing rationales behind it (each of which may be modified by multiple other factors), and—importantly—determining which one or ones ultimately lead to a particular decision may not be possible. Still, although these factors might not seem so unexpected, we were surprised to see them discussed so openly in a group setting. Indeed, the focus group format appeared to be conducive to this sort of reflective discussion in that participants both openly discussed their own challenges and admitted to mistakes while at the same time challenging and accepting the challenges of their peers.
Recently, an international working group suggested that the evaluation of professionalism must be viewed from three different, yet complementary, perspectives: that of individuals, their myriad relationships, and the organizations and societies in which they function.24 Others have also encouraged the community to view professionalism as more of a systems issue, with systems-levels evaluations and solutions, rather than simply focusing on the physician.25 Further, Lucey and Souba17 noted that “despite the fact that the most widely disseminated anecdotes about unprofessional behavior have practicing physicians as their subjects, most … organizations have chosen to look upstream, to the educational process, to fix the problems,” perhaps because there are 129 MD-granting medical schools in the United States versus hundreds of thousands of physicians in practice. These articles underscore the urgent need for the medical community not only to deepen its understanding of the context in which most physicians practice and the real pressures they face on a daily basis but also to explore their decision-making processes in the face of these sorts of challenges. Focusing on the training environment, although more efficient, will not be sufficient.
Our research has several limitations to consider. Although our participants worked in varied practice settings both in primary care and specialties, they were all volunteers from a single geographic area, so we cannot be certain about how transferable our findings would be to other groups or settings. Participants were also aware that the study was funded by (and conducted in part by employees of) the American Board of Internal Medicine, which might have had an influence on the nature of their responses. Further, the nature of focus group methodology precludes comparisons among individuals, although it does allow for robust discussion and debate within each group.
Our study begins to shed light on the contextual issues that influence professional behavior in practice and suggests some potential avenues worth exploring in an attempt to improve professionalism in practice. For example, it seems clear that in addition to the basic principles of professionalism that physicians recognize and refer to, multiple other interdependent forces are at play in each unique situation, which makes physicians’ approaches to professional dilemmas difficult to predict or assess. Our participants indicated that these scenarios are quite common in practice, so perhaps it would be worthwhile to teach physicians how to recognize situations in which they are at risk of acting in a way that is inconsistent with endorsed principles. They might uncover their own particular triggers or idiosyncrasies, and they might learn strategies to anticipate and overcome them. In addition, helping physicians examine their own practices might be particularly fruitful in determining which practice-based elements might trigger lapses in professional behavior. Finally, we were struck by the degree of open and supportive yet challenging discussion within the focus groups, and we feel there is potential to further explore the use of critical reflection among peers as a way to foster and reinforce professionalism in practice. Many of these issues will be the subject of future research, along with a more broad-based study to determine transferability of our findings.
Acknowledgments: The authors would like to acknowledge Ben Chesluk, Sidd Reddy, and Robin Guille at the American Board of Internal Medicine for their assistance in creating the vignettes and critiquing the emerging analysis.
Funding/Support: Dr. Ginsburg was funded in part as the Kimball Scholar, American Board of Internal Medicine (ABIM) Foundation, during the course of the study. Funding for the study was provided by the ABIM Foundation. The funding agency had no influence on the study design, data collection, analysis, or reporting.
Other disclosures: None.
Ethical approval: Ethical approval was granted by the Essex Institutional Board, Lebanon, New Jersey.
Previous presentations: These results have not yet been published, but abstracts were accepted at the Ottawa Conference, Kuala Lumpur, March 2012, and the Society of General Internal Medicine, Orlando, Florida, May 2012.