Physician Self-disclosure in Primary Care Visits: Enough About You, What About Me?

Background  The value of physician self-disclosure (MD-SD) in creating successful patient-physician partnerships has not been demonstrated.

Methods  To describe antecedents, delivery, and effects of MD-SD in primary care visits, we conducted a descriptive study using sequence analysis of transcripts of 113 unannounced, undetected, standardized patient visits to primary care physicians. Our main outcome measures were the number of MD-SDs per visit; number of visits with MD-SDs; word count; antecedents, timing, and effect of MD-SD on subsequent physician and patient communication; content and focus of MD-SD.

Results  The MD-SDs included discussion of personal emotions and experiences, families and/or relationships, professional descriptions, and personal experiences with the patient’s diagnosis. Seventy-three MD-SDs were identified in 38 (34%) of 113 visits. Ten MD-SDs (14%) were a response to a patient question. Forty-four (60%) followed patient symptoms, family, or feelings; 29 (40%) were unrelated. Only 29 encounters (21%) returned to the patient topic preceding the disclosure. Most MD-SDs (n = 62; 85%) were not considered useful to the patient by the research team. Eight MD-SDs (11%) were coded as disruptive.

Conclusions  Practicing primary care physicians disclosed information about themselves or their families in 34% of new visits with unannounced, undetected, standardized patients. There was no evidence of positive effect of MD-SDs; some appeared disruptive. Primary care physicians should consider when self-disclosing whether other behaviors such as empathy might accomplish their goals more effectively.

Effective physician-patient communication appears to improve health outcomes,^1-4 but the empirical evidence regarding how best to create healing relationships is largely lacking.⁵ In particular, physician self-disclosure (MD-SD), when the physician shares personal information and/or experiences, has generated controversy. Despite seeming to be a way to strengthen the patient-physician relationship,⁶ recent evidence has called this into question. In the only systematic study of MD-SD to our knowledge, Beach et al^7,8 found that MD-SD was common. They noted that MD-SDs in surgical visits were associated with greater patient satisfaction and reports of warmth, friendliness, comfort, and reassurance. In contrast, in primary care visits, MD-SDs were associated with lower ratings on all of those scales. These findings highlight unanswered questions about the content, antecedents, and consequences of MD-SD, and when and how physicians should use self-disclosure during medical visits.

To describe patterns of MD-SD comprehensively, we applied sequence analysis to an existing database of patients' first visits to primary care physicians. During prior qualitative analyses of these data focusing on physician responses to ambiguous symptoms, our research team⁹ incidentally noted the frequency and content of MD-SDs. In the present study, we examine the presence and impact of the MD-SD on the nature of communication, the patient, and the process of care. This analysis is part of a larger study, approved by the University of Rochester research subjects review board, that used covert audio recordings of office visits to study the relationship of patient-centered communication and health outcomes. The methods and physician sample have been reported in detail elsewhere.^9-11

In late 1999, 12 physician recruiters identified and approached 297 eligible primary care physicians in the Rochester, NY, region (population, 1.1 million people) until a total of 100 physicians were enrolled. Physicians gave written informed consent to participate in a study of patient care and outcomes. They agreed to complete surveys and have 2 unannounced, surreptitiously audio-recorded standardized patient visits in 2000 or 2001. Physicians were reimbursed $400 to cover the cost of their time for the visits and completion of surveys.

We used unannounced standardized patients to provide physicians uniform stimuli, avoiding confounding factors such as the Hawthorne effect, case-mix, mutual accommodation to each others' communication styles, and self-selection of physicians by patients.^12,13 Thus, by using unannounced standardized patients, we focused on the physician's contribution to communication, and in particular for this study, the use of MD-SD.

The standardized patients were trained to portray specific patient roles realistically and reproducibly, such that their communication, symptom presentation, clinical signs, and affect were indistinguishable from those of actual patients. Detailed clinical biographies were developed for 2 middle-aged white patient presentations: gastrointestinal reflux disease and medically unexplained symptoms.^9-11 The standardized patients were not specifically trained to respond to MD-SDs.

Each physician was randomly assigned 2 standardized patients, 1 man and 1 woman, representing each illness condition. The first standardized patient visit was randomly assigned and stratified by standardized patient illness condition and sex; the second visit was by a standardized patient of the other sex who portrayed the other illness condition. The standardized patients made appointments as if they were real patients wishing to establish care with a new physician and who complained of “chest pain for a couple of weeks” that presented as an acute, nonurgent concern. Two days after the visit, a fax was sent to the physician to determine whether, when prompted, the physician could identify the patient as a standardized patient. If detected, physicians rated the realism of the standardized patient portrayal, whether they changed their practice style in response, and when the detection occurred (before, during, or after the visit).

Each standardized patient visit was recorded using a hidden digital audio-disc recorder, transcribed, and reviewed for verbal content and intonation. Although prior quantitative analyses of these data showed no differences between communication or outcomes of undetected visits compared with detected visits,^10,11,14 we did not want to risk that subtle differences might distinguish them. Thus, we excluded the 3 MD-SDs that occurred in detected visits.

After all authors reviewed a subsample of MD-SDs, self- disclosures were defined as physician statements about his or her own personal or professional experience. Personal experience included statements about the physician's own family members, living situation, health problems, travel experience, feelings, and political beliefs. Professional experience included statements about the physician's practice conditions and interactions with other health care professionals or patients.

All 113 visits that were part of this study were reviewed by one of us (J.S.) to identify MD-SDs. Two of us (S.H.M. and D.M.) reviewed the visits and confirmed MD-SDs. Disagreements were brought to the full group and resolved by consensus.

Phenomenology and sequence analysis were used to analyze visit transcripts. Phenomenology uses an inductive, discovery-oriented approach to examine a behavior or experience of interest.¹⁵ To understand how the MD-SDs arose and affected the interaction, we used sequence analysis¹⁶ to describe the sequence of discourse immediately preceding and following the MD-SD.

We used an iterative process to create a coding system for the sequences and subsequently for the observed effects of the MD-SDs. We noted key words and phrases that influenced the form and quality of the visit and discussed them in consensus meetings. Coding development continued until no new codes emerged from examining additional data (saturation).^15,17 With each revision, all previously coded interviews were recoded and checked by a minimum of 2 researchers on the team. Codes encompassed timing, antecedents, content, patient response, concordance of patient and physician content, and subsequent physician communication. Any differences in coding were brought to the larger group for verification or resolution.¹⁷ Disagreements were resolved by consensus.

The 100 participating physicians had 193 standardized patient visits (7 physicians saw only 1 standardized patient). Four recordings could not be used owing to poor technical quality. Of the remaining 189 visits, 76 (40%) were excluded from further analysis because the physician, after prompting, suspected the standardized patient was not a real patient before the end of the visit.

All 113 remaining visit recordings were reviewed. A total of 73 MD-SDs were identified. Thirty-eight visits (34%) to 32 physicians contained at least 1 MD-SD (range, 1-12). These physicians were demographically similar to nondisclosing physicians (Table 1). Among the physicians who self-disclosed, the median number of self-disclosures was 1 (interquartile range, 1-2). There was no association in the number of MD-SDs with visit length, standardized patient sex, or role.

The MD-SDs were found throughout the visit, with the greatest number (n = 28; 38%) occurring during the history and the information-gathering phases before the physical examination (Table 2). The MD-SDs were most often preceded by patient statements. Examples of the most common topics are symptoms, personal and/or emotional disclosures, and family issues (n = 49; 67%). The MD-SDs related to the content of the preceding patient statement 60% of the time (n = 44). The MD-SD was seldom a response to a direct patient solicitation for personal information (n = 10; 14%); instead, the physician often assumed patient interest in his or her experience. The following exchange between physician 16121 and a patient will serve as an example (please note, each example used in this article is from a different physician, and participants gave explicit permission to publish the quotations; the text is taken verbatim from the study transcripts except that, to protect anonymity, we changed all names and identifying characteristics):

Patient: It's not [my son’s] first year of teaching but it's his first time [at School A]. . . . 

Physician: Yeah, it's a nice place. We live over next to there. I think [my daughter], June, will be going there next year. . . . It's a nice school.

Patient: Well, she might have him. Does she like math?

Physician: She's pretty good but . . . she wouldn't TELL ya that she likes it.

Patient: Yeah.

Physician: [laughter] I don't think it's cool to like math.

Patient: I suppose if you're a girl.

Four disclosures (5%) were unsolicited endorsements of other clinicians in response to a patient statement about seeing the other practitioner.

Most often (n = 51; 70%), patients acknowledged or continued the topic of the MD-SD rather than returning to his or her concerns that preceded the MD-SD (n = 21; 29%). Rarely, patients explicitly expressed opinions about the MD-SD. Twice (4%), a patient abruptly changed the topic, suggesting disapproval of the MD-SD. This finding was experienced as caretaking in a hierarchal relationship, a deference to power, by 2 female members of the team, while the male members of the team rated the response to the MD-SD as indicative of patient interest. The team was blinded to the sex of the patients during the data analysis.

There was no example of a physician making a transition statement leading back to the patient's concern and no example of a patient stating explicitly that the MD-SD was helpful in any way.

The MD-SD sometimes ended with the physician's transition to a completely new topic; in the following case of physician 10151, the transition was abrupt, giving the patient no chance to respond to the MD-SD:

Physician: No partners recently?

Patient: I was dating for a while and that one just didn't work out. . . . about a year ago.

Physician: So you're single now.

Patient: Yeah. It's all right.

Physician: [laughing] It gets tough. I’m single as well. I don't know. We're not at the right age to be dating, I guess. So, let's see. No trouble urinating or anything like that?

Table 3 lists the coders' ratings of the foci and effects of the MD-SD sequence. Patient focus was defined as explicitly related to the patient's concerns or prior statements, while physician focus was defined as a thought or concern relating only to the physician, though it could be triggered by a word the patient used. In total, none of the MD-SDs were patient focused, and 44 (60%) were physician focused. For example, 1 physician, after learning that he and the patient had a relative with the same first name, talked at length about his family's immigration story.

Twenty-nine (40%) of MD-SDs were social, such as superficial comments on the weather. No MD-SDs were coded as focused primarily on patients' concerns. The vast majority (n = 62; 85%) were considered not useful to the patient, including all of the patient-initiated MD-SDs, such as the following from physician 12432:

Patient: I was concerned when I saw your “For Rent” sign here. I thought you might be moving. Are you going to stay here?

Physician: Oh yeah. There are two different offices upstairs. They are both rented right now, but one of the people will be moving out in September. It's very hard to rent sometimes. Sometimes you're lucky and somebody just comes along, and other times you're just not so lucky. Why, do you want to rent?

Patient: Nope.

Physician: Give you a good price.

Patient: No, I just wanted to find out what's going on, really.

Physician: Okay, what can I do for you?

Only 3 MD-SDs (4%) were coded as useful—providing education, support, explanation, or acknowledgment, or prompting some indication from the patient that it had been helpful. In all 3 useful disclosures, the physician mentioned that he or she had the same medical condition as the patient, such as this exchange with physician 11212:

Physician: . . . I suffer from it myself.

Patient: Oh, you're kidding me?

Physician: Absolutely.

Patient: So what I’m saying to you is . . . what you have? The exact same thing?

Physician: Exactly.

Patient: Oh, my gosh!

Physician: Exactly.

Patient: What a coincidence.

Physician: Basically . . . the acids in the stomach are coming back up That's the reflux. And that really irritates the lining of the esophagus.

Eight MD-SD's (11%) were coded as disruptive. We defined disruptive as detracting in some way from the physician-patient relationship. Physician self-focus involved talking about the physician's own experience to the exclusion of the patient's concern, or extended monologues that distracted from the patient's concern, and used up valuable visit time. Expression of personal or political viewpoints that did not take the patient’s perspective into account is illustrated in the following exchange with physician 16052:

Physician: Were you in the service?

Patient: Nope.

Physician: How’d you get out of that?

Patient: Pardon?

Physician: How’d you get out of that?

Patient: Didn't get my number.

Physician: You lucky guy, you!

Patient: I agree.

Physician: I had to work harder than that, okay? [laughter]

Patient: What did you do?

Physician: I went to medical school, okay? [laughter]

Patient: Oh. All right. Cool.

Inadvertent competition with the patient was noted in this exchange with physician 11942:

Patient: I’m six feet, and she just told me I was 204.

Physician: Is that up a little bit for you weight-wise?

Patient: . . . it might be up a few pounds . . . I used to jog and . . . I just haven’t. . . . 

Physician: See, ‘cause I’m weighing more like 172, 173, and I’m six foot . . . and I’m still running. . . . I’m still doing the 5 and 10 and 15 Ks. The half-marathons and . . . 

Patient: So . . . I’m 30 pounds heavier than you?

Physician: Right now, yeah.

Finally, some physicians made implicit or explicit requests for patient support of the physician, as in the following exchange with physician 14641:

Physician: Deep breaths. . . . Lay down for me. I don't know. I think I’ll be panicked about the college time. I’m not sure I’m ready for that in any way. My daughter went to camp this week and everybody's saying, “Isn't it so great?” And I’m like, “No! I hate it! It's quiet! I don't have any friends!”

Patient: I hear ya. Right, right, right. Yep.

Physician: No buddies to do things with!

Patient: Yup, yup.

Physician: . . . All right. We don't like it! Now, did they give you an appointment for the stress test?

Table 4 summarizes the influence of the length of the MD-SD on the judged effect. While the numbers are not sufficient for quantitative analysis, the data suggest longer MD-SDs tended to be not useful or disruptive.

To our knowledge, this is the first effort to understand the sequence of interaction that results in MD-SD and to describe the consequence of MD-SD on subsequent discourse, visit content, and process. Our data support the findings of Beach et al^7,8 that use of disclosure to strengthen patient-physician relationships is rarely successful in primary care. We found that MD-SDs were often non sequiturs, unattached to any discussion in the visit, and focused more on the physician's than the patient's needs. Longer disclosures, both not useful and disruptive, interrupted the flow of information exchange and expended valuable patient time in the typically time-pressured primary care visit.

Though stimulated recall with standardized patients and physicians would be necessary to confirm our observations, we saw no evidence in our textual or audio recording analyses that any of the MD-SDs focused primarily on patients' needs. Had the MD-SD been patient focused, we expect that the patients would have elaborated further on their symptoms, acknowledged physician empathy (“Oh, so you really know what I am talking about”), or indicated that the MD-SD resonated in some way.

The frequency of MD-SD in these first visits is surprising. One could postulate that MD-SD is primarily a function of longer-term relationships. However, our incidence of MD-SD (34%) is actually higher than that found by Beach et al (17% of medical and 14% of surgical visits),^7,8 who did not use first visits or standardized patients. The reason for the difference in percentages is not clear. It is possible that physicians work harder to build a relationship with a new patient and believe that self-disclosure is a tool in this regard. However, given that the MD-SD–containing interviews were not significantly longer than others, these MD-SDs may take time away from important patient concerns. We saw no evidence that it helped build rapport.

Some MD-SDs appeared to be attempts at relationship building through the sharing of personal beliefs or experiences. We found little evidence that this strategy improved relationships. Sharing strong beliefs or emotions without understanding the patient's perspective seems risky; a practitioner may unknowingly infuse the dialogue with his or her needs without carefully tying them to the patients' needs. We noted a few cases in which physicians lost focus on the patient and focused exclusively on their own issues. These digressions may compromise the quality of patient care, be perceived as distancing, distracting, or uncaring, and, in some cases, make patients feel pressured to provide caring and support to a distressed physician.

Although the physicians may have intended it to be helpful, we have concerns that even the “I’ve got it too” category of MD-SD is rarely helpful. Although intended to reduce patient worry, this approach runs the risk of premature reassurance or advice without fully understanding the patient's experience,¹⁸ which can paradoxically raise patient anxiety. Furthermore, it may increase the chance of a misdiagnosis if the physician does not distinguish the patient's clinical situation from his or her own (eg, assuming a benign explanation for symptoms based on his/her own outcome).

In numerous MD-SDs, physicians spontaneously talked about their children and other life experiences without knowing about the patient's family experiences. In these initial visits, infusing dialogue with statements about the physician's family, personal habits, or professional life runs the risk of competing with the patient or raising negative or emotionally charged reactions without providing opportunity for the patient to process the reactions.

Most MD-SDs likely reflect positive intentions, attempts to show the physician's human side. However, there are other, more reliably helpful ways in which the physician's personal experience can be called on as a resource to support patients, particularly through the use of empathy, understanding, and compassion.¹⁹

To the extent that practitioners are disclosing to receive support, there are more helpful venues that would not risk diverting the focus of the medical encounter. Physician self-awareness and well-being are critical to shaping successful practices and can be accomplished in many ways.^20-25 Examples are professional support groups, mindfulness training, or conversations with friends and family. Although it is understandable that practitioners who spend considerable time each week caring for patients might find themselves wanting to share thoughts and feelings with patients, we hope that our findings encourage practitioners to explicitly consider whether self-disclosure is in the patient's best interest and the best use of visit time.

Using standardized patients has strengths and weaknesses. Standardized patients reduce patient variation, allowing research to focus more on physician factors. The standardized patients in this study were trained for several months using detailed scripts and multiple test interviews before they participated in the study.

This study had a higher detection rate than other standardized patient studies,^26-29 likely owing to 2 factors: (1) more than half of the primary care practices in Rochester, NY, were closed to new patients at this time; and (2) participating physicians were prompted by a fax 2 days after the visit, a much shorter time lag than in other studies. The physicians who detected the standardized patients rated them as very believable. Although prior quantitative analyses from this study suggested that standardized patient detection was not associated with alterations in communication behavior or outcomes,^10,11,30 we excluded these visits from our analyses to avoid any subtle effects of detection.

While textual and audio recording analyses allowed us to describe the phenomenon of physician self-disclosure, we cannot draw definitive conclusions about physician intent or patient experience. A stimulated recall study would be a valuable next step in furthering our understanding of the effect of MD-SD. Finally, our data only apply to primary care physicians. Others have reported that, in contrast to primary care physicians, surgeons who self-disclosed were rated more positively by patients than those who did not.³¹ Further study might clarify whether the context, content, or presentation of surgeons' MD-SDs contributed to the higher patient ratings, or if the MD-SDs are markers for other behaviors that build those relationships.

Our observations of initial visits to primary care physicians by unannounced standardized patients confirm that MD-SDs are common and exhibit a wide spectrum of content, frequency, and effect. Our analysis suggests that MD-SD usually is of little value and, occasionally, can actually impair the physician-patient relationship. Primary care physicians may wish to make explicit decisions about any use of self-disclosure and consider using empathy and other ways of demonstrating support and building relationships.

Correspondence: Susan H. McDaniel, PhD, Department of Family Medicine, 777 S Clinton Ave, Rochester, NY 14620 (susanh2_mcdaniel@urmc.rochester.edu).

Accepted for Publication: February 28, 2007.

Author Contributions: All authors had full access to all of the data in the study and take equal responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: McDaniel, Beckman, Morse, and Seaburn. Acquisition of data: Silberman and Epstein. Analysis and interpretation of data: McDaniel, Beckman, Morse, Silberman, and Epstein. Drafting of the manuscript: McDaniel, Beckman, Morse, Silberman, and Epstein. Critical revision of the manuscript for important intellectual content: McDaniel, Beckman, Morse, Silberman, and Seaburn. Statistical analysis: Morse. Obtained funding: Epstein. Administrative, technical, and material support: McDaniel, Beckman, Silberman, Seaburn, and Epstein. Study supervision: Epstein.

Financial Disclosure: None reported.

Funding/Support: This research was supported by a grant from the University of Rochester Family Research Roundtable Small Grants Program (Dr McDaniel) and grant R01-HS1610-01A1 from the Agency for Healthcare Research and Quality, Rockville, MD.

Acknowledgment: We thank the physicians who participated in the study and courageously submitted their interviews for analysis. Their examples helped us gain insight into a sensitive and important area for all practitioners. We also acknowledge Sean Meldrum, MA, and Sarah Schneider, BA, for their help with data analysis. We thank the members of the University of Rochester Wynne Center for Family Research as well as Arlene Katz, PhD, Stephen Lurie, MD, PhD, and Fred Wamboldt, MD, for their thoughtful comments on earlier drafts of this article.