CONTEXT: Physician self-disclosure has been viewed either positively or negatively, but little is known about how patients respond to physician self-disclosure.
OBJECTIVE: To explore the possible relationship of physician self-disclosure to patient satisfaction.
DESIGN: Routine office visits were audiotaped and coded for physician self-disclosure using the Roter Interaction Analysis System (RIAS). Physician self-disclosure was defined as a statement describing the physician's personal experience that has medical and/or emotional relevance for the patient. We stratified our analysis by physician specialty and compared patient satisfaction following visits in which physician self-disclosure did or did not occur.
PARTICIPANTS: Patients (N= 1,265) who visited 59 primary care physicians and 65 surgeons.
MAIN OUTCOME MEASURE: Patient satisfaction following the visit.
RESULTS: Physician self-disclosure occurred in 17% (102/589) of primary care visits and 14% (93/676) of surgical visits. Following visits in which a primary care physician self-disclosed, fewer patients reported feelings of warmth/friendliness (37% vs 52%; P = .008) and reassurance/comfort (42% vs 55%; P = .027), and fewer reported being very satisfied with the visit (74% vs 83%; P = .031). Following visits in which a surgeon self-disclosed, more patients reported feelings of warmth/friendliness (60% vs 45%; P = .009) and reassurance/comfort (59% vs 47%; P= .044), and more reported being very satisfied with the visit (88% vs 75%; P = .007). After adjustment for patient characteristics, length of the visit, and other physician communication behaviors, primary care patients remained less satisfied (adjusted odds ratio [AOR], 0.45; 95% confidence interval [CI], 0.24 to 0.81) and surgical patients more satisfied (AOR, 2.22; 95% CI, 1.12 to 4.50) after visits in which the physician self-disclosed.
CONCLUSIONS: Physician self-disclosure is significantly associated with higher patient satisfaction ratings for surgical visits and lower patient satisfaction ratings for primary care visits. Further study is needed to explore these intriguing findings and to define the circumstances under which physician self-disclosure is either well or poorly received.
Physician self-disclosure can be defined broadly as any statement made to a patient that describes the physician's personal experience,1 and has been alternatively described as either a boundary transgression2 or a way of fostering trust and rapport in the physician-patient relationship.3–6 Yet, while many aspects of doctor-patient communication have been explored in great detail, the impact of physician self-disclosure on patients has received relatively little attention.
This study aims to explore the possible relationship of physicians’ self-disclosure to patient satisfaction. We used an existing data set of audiotaped routine office visits with primary care physicians and surgeons7 to determine the prevalence of physician self-disclosure in routine office visits and to explore its possible relationship to patient satisfaction. Because previous studies have demonstrated that physician communication patterns differ across specialties,7 we hypothesized that these relationships may differ between primary care and surgical visits.
We analyzed data originally collected in 1993 to compare the routine communication styles of physicians with and without a history of malpractice claims.7 The study was approved by the Institutional Review Board (IRB) of Legacy Good Samaritan Hospital, Portland, Oregon and by the IRB of Johns Hopkins School of Hygiene and Public Health, Baltimore, Maryland.
Physicians were selected for participation in the original study from the databases of two large insurance companies according to their history of malpractice claims. Physicians were eligible for inclusion if they were in active practice in Colorado or Oregon and had graduated from medical school at least 13 years prior to the study. Eligible primary care physicians included general internists and family practitioners. Eligible surgeons included general surgeons and orthopedic surgeons. Overall, 81% of physicians agreed to participate. The final sample of 124 physicians included 59 primary care physicians and 65 surgeons.
Patients were recruited by research assistants from the physicians’ waiting rooms. The research assistant attempted to recruit at least 10 patients for each physician (range in final sample 8 to 12 patients/physician). Patients were eligible for inclusion if they were older than 18 years, spoke English, and were not in any acute distress. Primary care patients were eligible only if they had had at least 2 prior visits with the physician. Surgical patients were not eligible if they were scheduled for a procedure at that visit. All patients gave informed consent and 80% of eligible patients agreed to participate. Characteristics of the patient and physician samples are summarized in Table 1.
Table 1. Sample Characteristics
|Race, % White||93||95|
|Gender, % female*||12|| 2|
|Mean years in practice†||20.2||16.6|
|Mean hours/week providing patient care†||45.6||55.0|
|Mean patient age, y†||54.4||50.8|
|Patient gender, % female||59||56|
|Patient race, % White*||82||88|
|Patient education, % college||28||26|
|Patient income, % > $60,000||17||17|
|Self-rated physical status, % very good/excellent||39||41|
|Self-rated emotional status, % very good/excellent||46||51|
There were 2 primary sources of data for our analysis: patient surveys and audiotaped patient-physician encounters. Each patient completed an extensive survey providing demographic data and ratings of the doctor and visit. These ratings were ranked on a 5-point scale and included overall satisfaction (very dissatisfied to very satisfied), the likelihood that they would recommend their doctor to a friend (strongly agree to strongly disagree), how well the patient felt their doctor knew them (not at all to very much), and the extent to which the patient experienced warmth/friendliness and reassurance/comfort during the visit (not at all to very much). Patients also were asked if they had any concerns that were not discussed in the encounter (yes/no).
Each of the audiotaped patient visits (N= 1,265) was coded for content by one of three trained coders using the Roter Interaction Analysis System (RIAS). In this extensively studied and well-validated system,8–11 each statement or complete thought made by either physician or patient is coded into one of 38 mutually exclusive and exhaustive categories. Although the purpose of the original study was not focused on physician self-disclosure explicitly, the RIAS routinely codes it. The operational definition from the RIAS coding manual is “Statements which describe the physician's personal experience in areas which have medical and/or emotional relevance for the patient.”1 Coders were trained in the RIAS at Johns Hopkins using a coding manual with detailed definitions, annotated examples, and training tapes. Intercoder reliability was calculated on 121 double-coded tapes and the reliability coefficient for physician self-disclosure was 0.94.
Because previous analyses have confirmed that communication patterns differ across specialties, we stratified all our analyses by physician specialty.7 First, we examined the frequency with which physician self-disclosure occurred in routine primary care and surgical office visits both at the patient level (N= 1,265) and at the physician level (N= 124).
Next, we analyzed patient satisfaction ratings of visits (N= 1,265) in which the physician did or did not disclose. Our independent variable was the presence or absence of physician self-disclosure in the medical encounter as measured by the RIAS. Because of the strong positive skew typical of satisfaction ratings, the ratings were dichotomized so that the highest-ranking category was compared with all other responses using χ2 tests in bivariate analyses and logistic regression in multivariate analyses. All multivariate estimates used the generalized estimating equation (GEE) to account for clustering of patients within physicians. We adjusted for potentially confounding variables that were related either to physician self-disclosure or to patient satisfaction in our bivariate analysis, including other potentially confounding physician communication behaviors (such as positive talk, affective talk, and personal talk, which we hypothesized might be related to satisfaction and to self-disclosure) as coded by the RIAS.
In addition, we performed several subsidiary analyses. We stratified our analyses by the number of prior visits between physician and patient (<2, 3 to 10, 11 to 20, and >20) and evaluated the consistency of the associations between physician self-disclosure and patient satisfaction across strata. We also stratified the analysis by the presence of malpractice claims to determine the consistency of association between physician self-disclosure and patient satisfaction across malpractice strata.
Finally, in order to help explain the results of our initial analysis, we used χ2 tests to determine whether particular types of physician self-disclosure (further categorized in a qualitative analysis described elsewhere12) occurred more often during surgical or primary care visits, or whether particular types of self-disclosure were more or less likely to be associated with a high degree of patient satisfaction.
Prevalence of Physician Self-disclosure
Physician self-disclosure occurred in 15.4% of routine office visits, and there was no statistically significant difference in the prevalence of self-disclosure at the visit level between surgery and primary care (17%[102/589] of primary care visits and 14%[93/676] of surgical visits). Overall, at the physician level, 71% of physicians disclosed in at least one of their audiotaped visits and there were no significant differences in this rate by specialty. There were also no significant differences in the prevalence of self-disclosure between physicians who had and those who did not have a history of malpractice claims.
Although the number of self-disclosing statements made by the physicians who disclosed ranged between 1 and 69 statements per visit, the median number of self-disclosing statements made by the physician was 2. Most physicians (96%) who disclosed during a patient visit made 3 or fewer self-disclosing statements. Both primary care and surgical visits in which physician self-disclosure occurred were significantly longer than visits without physician self-disclosure (21.8 vs 15.4 minutes [P= .001] and 16.4 vs 13.2 minutes [P= .005], respectively).
Patient Evaluation of Disclosure Visits
The results of the unadjusted analysis of the relationship of physician self-disclosure to patient evaluations of an office visit are summarized in Table 2. In the primary care setting, patients were significantly less likely to report that their physician knows them “very much,” significantly less likely to report feeling “very much” warmth/friendliness and “very much” reassurance/comfort, and significantly less likely to report being “very satisfied” following disclosure visits. Following visits in which the primary care physician self-disclosed, patients were not significantly more or less likely to report having had concerns that were not discussed.
Table 2. Patient Evaluation of Office Visits
|Patients reporting that their doctor knows them “very much”, %||42.9||58.1†||36.7||29.9|
|Patients reporting that they had concerns that were not discussed, %|| 1.0|| 1.6|| 4.4|| 4.3|
|Patients reporting experiencing:|
| “very much” warmth/friendliness, %||36.7||51.7†||59.8||45.0*|
| “very much” reassurance/comfort, %||42.4||54.7*||58.7||47.3*|
|Patients who “strongly agree” that they would recommend doctor, %||72.0||80.7||89.1||80.4|
|Patients reporting “very satisfied” with visit, %||73.7||83.4*||87.9||75.2*|
In contrast to primary care, a greater percentage of surgical patients reported feeling “very much” warmth/friendliness and “very much” reassurance/comfort, and significantly more reported being “very satisfied” following visits in which the surgeon self-disclosed. In the surgical setting, there was no statistically significant difference in the percent of patients who reported that their physician knows them “very much,” the percent who had concerns that were not discussed, or the percent who strongly agreed that they would recommend their physician.
In the multivariate analysis (Table 3), primary care patients remained significantly less likely to report being “very satisfied” and surgical patients remained significantly more likely to report being “very satisfied” following a visit in which physician self-disclosure occurred, after controlling for patient variables (race, self-rated physical and emotional health, number of prior visits with the physician), number of hours/week worked by the physician, and the length of the visit. When we included other potentially confounding physician communication behaviors (physician personal talk, physician positive talk, physician affective talk) in the model, the findings were essentially unchanged; primary care patients remained significantly less satisfied and surgical patients remained more satisfied following disclosure visits. Additional adjustment for physician gender and race did not substantively change the results.
Table 3. Odds of Patient Being Very Satisfied Following an Office Visit in Which Physician Self-discloses
|Unadjusted||0.56||0.36 to 0.93||2.40||1.24 to 4.63|
|Adjusted for patient race, number of prior visits, patient self-rated physical and emotional health, length of visit, and number of hours/week physician works||0.49||0.20 to 0.87||2.49||1.29 to 4.80|
|Adjusted for patient race, number of prior visits, patient self-rated physical and emotional health, length of visit, number of hours/week physician works, doctor personal talk, doctor positive talk, and doctor affective talk||0.45||0.24 to 0.81||2.22||1.12 to 4.50|
The results were unchanged after stratification by the number of prior visits; primary care patients were less satisfied and surgical patients more satisfied after visits in which their physician self-disclosed regardless of the number of prior visits between patient and physician. Finally, the results were also unchanged after stratification by malpractice claims.
Patient Evaluations Following Specific Subtypes of Physician Self-disclosure
Definitions and examples of physician self-disclosure subtypes have been described previously.12 There were no statistically significant differences in the frequency of self-disclosure subtypes across physician specialty. As shown in Table 4, the absolute number of each physician self-disclosure subtype was small, and there were almost no statistically different patient satisfaction ratings following visits in which particular types of physician self-disclosure occurred.
Table 4. Percentage of Patients Very Satisfied Following Physician Self-disclosure Subtypes
|Any PSD||99||73||83*||91|| 87||75*|
| Reassurance ||30||67||82||39|| 90||77*|
| Counseling||28||82||82||22|| 86||77|
| Rapport||26||85||82||14|| 93||77|
| Intimacy|| 9||78||82|| 4||100||77|
| Extended narrative|| 5||60||82|| 6||100||77|
| Casual||15||67||83||14|| 93||77|
Our study found that physician self-disclosure occurred in 15% of all routine outpatient visits and that most physicians (71%) self-disclosed to at least one patient among their audiotaped encounters. These data suggest that physician self-disclosure is not an uncommon behavior. Our study also found that, following outpatient visits in which physician self-disclosure occurred, surgical patients were more satisfied and primary care patients were less satisfied. Several interpretations for these disparate findings among primary care and surgical patients can be suggested.
One possible explanation is that differences in patient expectations and emotional state may lead patients to respond differently to the self-disclosures of surgeons and primary care physicians. Facing an invasive procedure with an unknown outcome and inherent risk, surgical patients may be more acutely anxious and vulnerable than primary care patients. Within this context, every form of self-disclosure seemed to be appreciated by patients, but especially evident was the higher satisfaction of surgical patients (in contrast to primary care patients) when the disclosure was characterized as reassuring. Self-disclosure from a surgeon may function as a sign of personal interest and emotional support.
Primary care physicians, by contrast, are more involved in chronic disease management in which cure is often not a realistic goal. Studies have demonstrated that, in a pediatric primary care context, physician empathy had a measurable effect on reducing parental anxiety, while statements of reassurance had no effect.13 It is not hard to imagine that reassurance in the primary care context, while well-intentioned, might be heard as premature and promising something that cannot easily be delivered. Perhaps in the context of chronic illness, reassuring disclosures appear dismissive or to invalidate a patient's concerns. This interpretation is also consistent with a recent qualitative study, which found that typical reassurance patterns of consultant rheumatologists (those emphasizing mildness or earliness of symptoms) were unsuccessful.14 Insofar as primary care physicians serve a gate-keeping role, patients may be more suspicious of this sort of reassurance. Also, primary care patients and their physicians have had longer standing relationships in which physician self-disclosures might be repeated. A repeated story not only loses its emotional currency but in fact may be resented as insincere and impersonal.
It is also possible that primary care physicians use self-disclosure instrumentally to increase their own credibility when faced with circumstances of interpersonal strain. If so, the decreased satisfaction ratings of primary care patients could either be a result of the underlying strained relations that led to the disclosure, or of the disclosure itself. Unfortunately, there is no way of knowing this from our data. Additional qualitative research techniques such as video review and participant expertise to compare responses of surgical and primary care patients to physician disclosure may be helpful in exploring this phenomenon in the future.15,16
Our study did not have enough power to detect significant differences in patient ratings of care following specific subtypes of disclosure; however, several observations might be made and pursued with future research. In our study, primary care patients did not appear to be less satisfied following a visit in which the physician made a counseling self-disclosure. Although these patients were not more satisfied, this finding is not inconsistent with previous work showing that counseling disclosures increased patient evaluation of the physician's ability to motivate patients and establish credibility as a role model.6 Primary care patients’ satisfaction also appears unaffected by rapport-building disclosures, suggesting that self-disclosing statements that are intended to be humorous, express empathy, or legitimate a patient's emotions may not be poorly received by primary care patients.
Although it has been suggested that patients may find sharing health concerns difficult in an encounter where a physician self-discloses,2 we found that patients were no less likely to report having had concerns that were not discussed following disclosure visits. Our data suggest that, although primary care patients are less satisfied after a visit in which physician self-disclosure occurs, the mechanism for this dissatisfaction does not appear to be that the patients felt uncomfortable discussing their own concerns. Similarly, although self-disclosure has been negatively viewed as a boundary violation, we found little cause for alarm in our reading of the transcripts, as extreme examples of intimacy and self-preoccupation were absent.
These results should be interpreted in light of several limitations. First, physician self-disclosure was often only one or two statements in the context of an entire visit. As in any correlational study, we cannot prove a causal relationship between these statements and subsequent patient satisfaction. Nevertheless, given the strength of the association with either high or low patient satisfaction ratings, physician self-disclosure is either a direct or a proxy indicator for some important characteristic of the visit. This is evident in the increased length of the disclosure visits (by 3 to 6 minutes), which could not simply have been due to one or two self-disclosing statements made by the physician (estimated to take no longer than 15 seconds).
Second, these encounters occurred approximately 10 years ago and, although we have no particular reason to believe that the prevalence or impact of physician self-disclosure may have changed over this period of time, it is possible that physician self-disclosure is either more or less prevalent today. Because of the cross-sectional nature of the database, we were not able to assess the impact of physician self-disclosure on the development of physician-patient relationships over time. Also, we have treated physician self-disclosure as a behavior independent of conversational sequencing. It seems likely that most physician self-disclosures are not free standing, but are rather responses to actions of the patient immediately prior to the self-disclosure. Careful analysis of the sequential context in which physician self-disclosure occurs in primary care and surgical visits may help shed light on the current findings.
Finally, the generalizability of our findings is limited in several respects. Our data are limited to experienced primary care physicians and surgeons practicing in two states (Colorado and Oregon) of the United States, and may not be generalizable to other areas or to less experienced physicians. Also, physician gender and racial diversity is limited in our sample, and we did not assess how gender, race, or age concordance between physician and patient may impact physician self-disclosure or patient satisfaction. Such concordance has been shown to affect communication patterns between patients and physicians in other studies.17
Despite these limitations, we believe that these results are valid and intriguing. We studied a large number of physicians and a relatively large number of patients for each physician. We were able to control for a variety of patient, physician, and visit characteristics, including many other potentially confounding physician communication behaviors. Last, the same relationships were found for a variety of evaluative measures; for almost every measure, primary care patients perceived disclosure visits more negatively whereas surgical patients perceived disclosure visits more positively, or at least equivalently. To our knowledge, this is the first study that attempts to empirically evaluate patient satisfaction following physician self-disclosure. Our results are surprising and should lead to further discussion and research.
In conclusion, our previous qualitative work demonstrated that physician self-disclosure is a complex and varied communication behavior.12 This current study demonstrates that, depending on the circumstances, self-disclosure can be significantly associated with either higher or lower patient satisfaction ratings. These results highlight the importance and complexity of physician self-disclosure and suggest that physicians should not assume that self-disclosure is uniformly going to lead to a more or less satisfactory visit from the patient's perspective. Until we know more about how patients respond to different types of self-disclosure, we recommend that physicians practice patient-centered interviewing techniques,18,19 maintain self-awareness of their motivation for and the impact of self-disclosure on the patient,20 be sensitive to how a patient responds when self-disclosure occurs, and self-disclose only when their intention is to meet a patient's needs.
The authors would like to thank Susan Larson for her help in analyzing the audiotapes and Hong Thi Vu for her statistical assistance. These data were presented in part at the 23rd annual meeting of the Society of General Internal Medicine, Boston, Massachusetts, May 4–6, 2000.
This work was supported by grant 99-504 from the Bayer Institute for Health Care Communication and by grant R01 07289 from the Agency for Healthcare Policy Research.