Patient Education and Counseling 95 (2014) 313–318
Contents lists available at ScienceDirect
Patient Education and Counseling journal homepage: www.elsevier.com/locate/pateducou
Communication Study
Physician gender, physician patient-centered behavior, and patient satisfaction: A study in three practice settings within a hospital Judith A. Hall a,*, Pa˚l Gulbrandsen b,c, Fredrik A. Dahl c a
Department of Psychology, Northeastern University, Boston, USA Institute of Clinical Medicine, University of Oslo, Oslo, Norway c Akershus University Hospital, Lørenskog, Norway b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 27 November 2013 Received in revised form 4 March 2014 Accepted 16 March 2014
Objective: To compare male and female physicians on patient-centeredness and patients’ satisfaction in three practice settings within a hospital; to test whether satisfaction is more strongly predicted by patient-centeredness in male than female physicians. Methods: Encounters between physicians (N = 71) and patients (N = 497) in a hospital were videotaped and patients’ satisfaction was measured. Patient-centeredness was measured by trained coders. Results: In the outpatient setting, female physicians were somewhat more patient-centered than male physicians; patient satisfaction did not differ. In the inpatient and emergency room settings, female physicians were notably more patient-centered than male physicians; satisfaction paralleled these differences. Nevertheless, there was some, though mixed, evidence that patient-centeredness predicted satisfaction more strongly in male than female physicians, suggesting that patients valued patientcentered behavior more in male than female physicians. Conclusion: Even though satisfaction mirrored the different behavior styles of male and female physicians in the inpatient and emergency room settings, in all settings male physicians got somewhat more credit for being patient-centered than female physicians did. Practice implications: If female physicians do not consistently receive credit for high patientcenteredness in the eyes of patients, this could lead female physicians to reduce their patient-centered behavior. ß 2014 Elsevier Ireland Ltd. All rights reserved.
Keywords: Physician gender Patient-centeredness Patient satisfaction Gender bias Outpatient Inpatient Emergency room
1. Introduction Female physicians are, on average, more patient-centered than their male counterparts. Female physicians and/or female medical students spend longer with patients, behave more positively verbally and nonverbally, talk more about psychosocial and emotional issues, use more partnership statements, offer more nonverbal encouragement to talk, have more skill in interpersonal perception, listen better, have greater self-reported and perceived empathy, offer more expressions of respect or praise, and put a greater focus on prevention [1–11]. Female physicians have more humanistic and patient-centered attitudes about patient care [12], and patients generally desire a patient-centered style [13–15]. Yet, female physicians do not receive much more—usually, no more—satisfaction from patients [16]. This lack of a difference
* Corresponding author at: Department of Psychology, Northeastern University, Boston, MA 02115, USA. Tel.: +1 617 373 3790; fax: +1 617 373 8714. E-mail address: [email protected] (J.A. Hall). http://dx.doi.org/10.1016/j.pec.2014.03.015 0738-3991/ß 2014 Elsevier Ireland Ltd. All rights reserved.
suggests a bias when considered against the research showing that female physicians better fulfill the goals of patient-centered medicine. Most of the research relating physicians’ gender to their behavior and to patient satisfaction has been conducted in primary care [1,16]. Our first goal was to compare the patient-centered behaviors of male and female physicians in three settings in one general teaching hospital—outpatient, inpatient, and emergency room. The second goal was to compare patients’ satisfaction between male and female physicians within each practice setting. In the only study we know of from an emergency room, female physicians received significantly higher satisfaction ratings from patients than male physicians did, but only among female patients [17]. That study could not ascertain whether this effect was due to the lens through which patients viewed physicians or to behavioral differences between male and female physicians. The present study allowed an investigation of this question. The third goal was to test a hypothesis about gender bias in patients’ satisfaction that has been suggested in previous research.
314
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
The meta-analysis finding only a negligible tendency for patients to be more satisfied with female than male physicians [16] suggests such a bias because it does not fit logically with female physicians’ behavior patterns, the fact that patients generally want patient-centeredness, and the fact that they report more satisfaction when the physician is more patient-centered [18]. However, since most studies have a preponderance of male physicians, the latter result may obscure a different pattern for female physicians. Indeed, two studies have found that the correlation of either competence or satisfaction with objectively measured patientcenteredness was stronger for male than female providers. Among medical students interacting with a standardized patient [19], for female medical students there was no correlation between patient-centeredness as scored by trained coders [20] and analog patients’ competence ratings, whereas there was a positive correlation for male medical students. In a laboratory experiment using scripted high and low patient-centered behavior enacted by actor-physicians [21], patient-centeredness strongly predicted analog patients’ satisfaction with male physicians, but it predicted satisfaction with female physicians either not at all or more weakly. These results clearly suggest a bias in patients’ evaluations. This could result from the overlap between the behaviors and skills considered patient-centered and the behavior and skills of women in general, which are well studied, ubiquitous, and reflected in strong societal stereotypes. According to research in nonclinical populations women are, and are stereotyped to be, more empathic, more warm and personable, more interested in relationships, more status leveling, more comfortable with emotional self-disclosure, and more skilled in nonverbal communication than men [22–29]. All of these expected and real female traits have their counterpart in the collection of attributes called patient-centeredness [30]. Because of this overlap, high patient-centered behavior in female physicians may be discounted as a marker for being a good physician, because such behavior is viewed simply as expected, even exemplary, female behavior but not necessarily a manifestation of high quality medical performance. Such behavior may be appreciated, while not adding to the patient’s appraisal of the physician as a fine professional. On the other hand, high patientcenteredness by a male physician is not part of the male stereotype and cannot be assimilated to a male schema. Instead, it is assimilated to a ‘‘good doctor schema’’ and the male physician gets a disproportionate boost in the patient’s eyes. Thus, patients may have a bias when evaluating the professional performance of physicians, with the result that a significant aspect of female physicians’ performance is not recognized as professionally relevant. This kind of subtle bias has also been seen in other organizational settings, where job-relevant ‘‘female’’ behaviors by women are not credited as contributing to high job performance when they should be [31–33]. Both of the studies described above showing stronger correlations between patient-centeredness and evaluations of male clinicians [19,21] were artificial in various ways. The first study was on medical students interacting with a standardized patient in a videotaped clinical examination, and the competence ratings were made by undergraduate analog patients. The second study used videos in which male and female actors played physicians and the satisfaction data were again collected from undergraduate analog patients. Although the analog patient method has received validational support [34], research must be conducted with practicing physicians and real patients. In light of the foregoing research, we would predict greater patient-centeredness in female than male physicians, though the comparison of such differences may vary across the different settings. We also would expect that satisfaction with female physicians would not be appreciably different from satisfaction
with male physicians, at least in the outpatient setting where most existing research has been done. Finally, in keeping with the experimental research relating patient-centeredness to satisfaction, we would expect this correlation to be stronger for male than female physicians. 2. Methods 2.1. Participants and procedure The data stem from a randomized controlled trial [35,36]. It consisted of 497 videotaped physician–patient interactions in the outpatient, inpatient, and emergency room settings of a general teaching hospital in Lørenskog, Norway, with 71 non-psychiatrist physicians who each saw between 1 and 8 patients. Twenty-one physicians were observed in more than one setting. The participation rate of the physicians was 69% (71 of 103 invited). After the interaction, patients filled in a questionnaire evaluating their interaction with the physician. All videos were included, as there was no significant difference in change of behavior between female and male physicians following the intervention [35]. Physicians were informed that they were due for observation shortly before filming and could not adjust their patient lists. Patients were recruited consecutively. Of 553 eligible patients, 519 (94%) consented [37]. Technical difficulties with videotapes left 497 encounters for analyses. Literate children completed questionnaires themselves, assisted by their parents when needed. 2.2. Encounters Table 1 provides descriptive information on the physicians and patients, as well as visit characteristics. 2.3. Measures 2.3.1. Roter Interaction Coding System Two hundred seven of the videotapes were rated by trained and reliable coders using the Global Ratings from the Roter Interaction Coding System (RIAS), a widely used descriptive system for provider–patient interaction [38]. After watching the interaction, coders made global ratings of the physician on the following scales: interested, friendly, engaged, sympathetic, dominant, anxious, and angry. The first five are relevant to the patient-centered concept (dominant in an inverse relation) and are analyzed in the present article. Anxious is not relevant to patient-centeredness, while the rating of anger, though relevant to patient-centeredness (in an inverse relation), had too limited variance to be useful. Examination of the correlations among the five patient-centered ratings Table 1 Physician and patient characteristics. Physicians (N = 71)
Patients (N = 497)
Age, mean (SD) Sex
40.3 (8.6) Females 30 (42%), males 41 (58%)
46.3 (24.6) Females 256 (52%), males 241 (48%)
Specialty
Internal medicine 25 (35%) Surgical disciplines 18 (26%) Anesthesiology 5 (7%) Neurology 8 (11%) Pediatrics 8 (11%) Gynecology 7 (10%) Residents 33 (47%) Consultants 38 (53%) 56 26 13
Position Settinga
a
Outpatient 375 (76%) Inpatient 81 (16%) Emergency room 41 (8%)
Twenty-one physicians were observed in more than one setting.
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
revealed strong correlations among the first four (Cronbach’s alpha = .86), while dominant was negligibly related to them (median r = .08). Therefore the ratings of interested, friendly, engaged, and sympathetic were averaged into a composite called the Global Ratings Patient-Centered Composite (M = 4.13, SD = .81, range = 2.25–6.00, possible range = 1–6), and dominant was kept separate (M = 3.28, SD = 1.13, range = 1–6, possible range = 1–6). 2.3.2. Four Habits Coding Scheme Physicians’ patient-centeredness was also scored with the Four Habits Coding Scheme [20] in which recorded provider–patient interactions are rated by trained coders using a 23-item scale, with higher scores indicating more patient-centered behavior. The habit scores are: I, Invest in the Beginning, 6 items; II, Elicit the Patient’s Agenda, 3 items; III, Demonstrate Empathy, 4 items; and IV, Invest in the End, 10 items. Four experienced students educated in psychology were trained on the system. Videotapes were rated in groups of 20 until acceptable interrater reliability (IRR) (Pearson r > .70) was achieved [39]. The four habits were analyzed separately as well as in a total score (M = 60.10, SD = 14.45, range = 29–108, possible range = 23–115). 2.3.3. Satisfaction Patient satisfaction was assessed by the global item ‘‘Using any number from 0 to 10, where 0 is the worst possible consultation with a doctor and 10 is the best possible consultation with a doctor, what number would you use to rate this consultation?’’ taken from The Consumer Assessment of Health Care Providers and Systems (http://www.ahrq.gov) (M = 8.76, SD = 1.32, range = 2–10). 2.4. Statistical analysis One set of analyses examined relations between variables in the whole sample and in the three practice settings without taking account of the nesting of patients within physicians. Although significance tests are compromised when patients who are nested within physicians are considered as separate units of analysis, estimates of effect size (i.e., correlation coefficients) are not biased and give a good picture of the relative magnitudes of effects. Pearson correlations and analyses of variance were used. Gender differences were expressed as the point-biserial correlation, which is the Pearson correlation between gender (coded female = 0, male =1) and another study variable. Positive correlations mean male physicians were higher on the variable than female physicians; negative correlations mean the reverse. In addition, a two-level model was employed in examining physician gender differences in order to take account of the nesting of patients within physicians. When examining relations between patients’ satisfaction and the patient-centeredness of male versus female physicians, Pearson correlations were calculated, and also two-level models with a term for the physician gender patient-centeredness interaction. This interaction was predicted to show a stronger relation between patient-centeredness and satisfaction for male than female physicians. An additional analytic method which also controls for nesting was used to test this interaction, by calculating the correlation between patient-centeredness and patient satisfaction for each physician separately, so that each correlation was based on only that physician’s patients. Thus, each physician had a correlation between his/her patient-centeredness and his/her patients’ satisfaction, thereby putting the analysis at the level of the individual physicians. The question was then posed whether these correlations were stronger for male than female physicians. A meta-analytic fixed-effects contrast between male and female physicians was then performed, which is conceptually analogous to an independent-samples t-test [40]. This method was used by
315
Table 2 Pearson correlations between physician gender and RIAS Global Ratings. RIAS scale
Interested Friendly Engaged Sympathetic Patient-Centered Composite Dominant
All
Outpatient
Inpatient
Emergency
(207)
(145)
(41)
(21)
.11 .26*** .02 .20** .18** .04
.03 .17* .02 .11 .09 .06
.27+ .40** .18 .38* .37* .31*
.28 .54** .30 .45* .47* .02
Notes: Gender is coded 0 = female, 1 = male. N is given in parentheses. Negative correlations mean female physicians scored higher than male physicians. + p .10. * p .05. ** p .01. *** p .001.
Hall and colleagues in a study to show that physicians liked their healthier patients more than their less healthy patients [41]. In the present study, only physicians who had four or more patients coded for patient-centeredness and who had variance across patients for both patient-centeredness and satisfaction were used in this analysis (n = 19). 3. Results 3.1. Physician gender differences for RIAS Global Ratings Table 2 shows physician gender differences for the RIAS Global Ratings (n = 207), for both the whole sample and for the three settings. The gender differences are expressed as point-biserial correlations; these correlations carry the identical p-value as ttests but add more information because they are expressed in a readily understood effect size metric (the Pearson correlation, denoted as r). For the sample as a whole, female physicians were rated by the trained coders as significantly more friendly and sympathetic than male physicians, contributing to a significant Global Ratings Patient-Centered Composite as shown in the table. However, there were notable differences between settings. Gender differences were weak in the outpatient setting. The only significant correlation indicated that female physicians were rated by the trained coders as more friendly than male physicians. The inpatient setting showed stronger gender differences than the outpatient setting; in the inpatient setting, females were significantly more friendly and sympathetic, and marginally more interested, contributing to a significant Global Ratings PatientCentered Composite as shown in the table. The gender difference became even stronger in the emergency room setting for friendliness, sympathy, and the Global Ratings Patient-Centered Composite. Controlling patient gender, patient age, and the length of the encounter using partial correlations had no impact except in the inpatient condition. In that condition, all of the physician gender differences actually became stronger when these variables were controlled: interested, r = .36, p < .05; friendly, r = .53, p < .001; engaged, r = .30, p < .07; sympathetic, r = .45, p < .01; and Global Ratings Patient-Centered Composite, r = .49, p < .01 (all partial correlations, df = 36). Turning to physician dominance, in the inpatient setting, male physicians were significantly more dominant than female physicians (Table 2). Controlling for patient gender, patient age, and the length of the encounter had no impact on the dominance correlations in any of the settings. In the two-level models for the whole sample, female physicians remained more friendly (p < .01) and sympathetic (p < .05) than male physicians, and the Global Ratings PatientCentered Composite was marginally significant (p = .07). In the
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
316
Table 3 Mean levels of the Global Ratings Patient-Centered Composite, Four Habits Total Score, and patient satisfaction for male and female physicians. Physician gender
Outpatient
Global Ratings Patient-Centered Composite Male 4.09 Female 4.25 Four Habits Total Score Male 58.73 Female 60.47 Patient satisfaction Male 8.91 Female 8.96
Inpatient
Table 4 Pearson correlations between physician gender and Four Habits Coding Scheme. Four Habits scale
Emergency
3.90 4.49
3.50 4.27
59.69 66.89
57.00 62.56
7.40 8.68
7.86 8.83
outpatient setting, effects were lower in significance with only the physician gender difference in friendliness approaching significance (p < .09). In the inpatient setting, no effects were significant. However, in the emergency room the difference in friendliness and the Global Ratings Patient-Centered Composite both were marginally significant (p < .06 and = .10, respectively), in spite of the very small sample size. This speaks to the strength of the correlations as shown in Table 2. The correlations shown in Table 2 express the magnitude of gender differences in terms of standardized effect sizes, but they do not give insight into the actual levels of performance of male and female physicians. A one-way analysis of variance (ANOVA) comparing the three settings on the Global Ratings PatientCentered Composite was performed separately for male and female physicians (Table 3, top panel). Female physicians were consistent across the three settings, F(2, 77) < 1.0. On the other hand, male physicians behaved differently across the three settings, F(2, 124) = 2.58, p < .09. A linear trend showed that male physicians’ behavior was progressively less patient-centered from outpatient to inpatient to emergency room, p < .05. For the dominance rating, there were no significant effects, but for male physicians there was a marginally significant (p < .10) quadratic trend showing higher dominance in the inpatient setting than in the other two settings, creating the notable gender difference for dominance in the inpatient setting. 3.2. Physician gender differences for the Four Habits Coding Scheme Table 4 shows gender differences for the Four Habits Coding Scheme, expressed as correlations. It is evident that female physicians scored higher than male physicians, although the correlations sometimes were not significant in the separate settings. The strongest gender effects occurred on Habit I (Invest in the Beginning, with items such as greetings and expansion of concerns) and Habit III (Demonstrate Empathy, with items such as expression of emotion, acceptance of feelings, and empathy). These are the two habits that pertain mainly to socioemotional behavior, in contrast to Habits II and IV, which are more about task functions. The correlations did not change when patient gender, patient age, and the length of the encounter were controlled. After applying the two-level model to the whole sample to correct for interdependence among patients within physicians, the effects for Habits I and III remained significant (p .05). There were no significant gender differences within the separate settings. To examine actual levels of Four Habits performance, ANOVAs like those described above were performed for the Four Habits Total Score and the individual habits. For Total Score, there were no effects for male physicians, F(2, 307) < 1.0. However, female physicians’ performance varied over settings, F(2, 184) = 3.76, p < .05, with a significant quadratic trend indicating that Total Score was highest in the inpatient setting, p < .05 (Table 3, middle panel). For female physicians, this quadratic trend was also significant for both Habits I and III.
I: Invest in the Beginning II: Elicit the Patient’s Perspective III: Demonstrate Empathy IV: Invest in the End Total Four Habits
All
Outpatient Inpatient Emergency
(497)
(375)
.15*** .01 .16*** .07 .12**
.11* .04 .10* .01 .06
(81) .21+ .12 .30** .21+ .27**
(41) .19 .10 .10 .20 .19
Notes: Gender is coded 0 = female, 1 = male. Sample sizes are given in parentheses. Negative correlations mean female physicians scored higher than male physicians. + p .10. * p .05. ** p .01. *** p .001.
3.3. Physician gender differences for patient satisfaction Consistent with previous research [16], patients were not more satisfied overall with either gender of physician; the overall correlation between physician gender and satisfaction was r(495) = .07, p < .13. This overall lack of a difference was mainly due to the outpatient visits, for which the relation was r(373) = .02, p < .69. Gender differences were notable in the other settings, however. Patients were more satisfied with female than male physicians both in the inpatient setting, r(79) = .35, p < .001, and in the emergency room, r(39) = .44, p < .01 (Table 3, lower panel). The correlations did not change when patient gender, patient age, and the length of the encounter were controlled. Also, there was no evidence, unlike the earlier published emergency room study [17], that the physician gender difference in the emergency room was limited to female patients. The two-level models similarly did not reveal any physician gender effects either overall or in the outpatient setting, but in the inpatient setting female physicians received higher satisfaction, p < .01, and in the emergency room this was also the case, p < .06. ANOVAs were conducted to examine satisfaction levels across the three settings. Satisfaction with male physicians was much lower for inpatient and emergency room than for outpatient, F(2, 307) = 24.91, p < .001. There was no difference across settings for female physicians, F(2, 184) < 1.0. 3.4. Is there a gender-related ‘‘double standard’’ for patient satisfaction? According to the logic put forth in the Introduction, indicators of patient-centeredness were predicted to be more strongly predictive of patients’ satisfaction for male than female physicians. Table 5 (top panel) shows these correlations for the Global Ratings Patient-Centered Composite; the pattern for the separate RIAS ratings was very similar. The prediction that the male physicians’ correlation would be stronger than the female physicians’ correlation was supported, more so for inpatients and emergency room than the outpatient setting. There were no significant interaction effects when examining this same question in the twolevel models. However, the meta-analysis based on physicians found that the average correlation between the Global Ratings Patient-Centered Composite and satisfaction was both strong and significant for male physicians (r = .50, p < .01, n = 12) while the corresponding correlation for female physicians was negligible (r = .06, p = .76, n = 7). Furthermore, the meta-analytic contrast comparing male and female physicians on this correlation was significant, p < .03. Thus, this analysis found significant support for the hypothesis, but note must be made that only 19 of the 61 physicians who had at least one patient scored for the RIAS could
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318 Table 5 Correlations of patient satisfaction with Global Ratings Patient-Centered Composite and Four Habits Total Score, separately for male and female physicians. Variable
All patients
Outpatients
Inpatients
Emergency
(207)
(145)
(41)
(21)
Composite .18+ (97) .25+ (48)
.41+ (21) .09 (20)
.70* (9) .05 (12)
.26*** (260) .12 (115)
.14 (36) .22 (45)
.34 (14) .24 (27)
Global Ratings Patient-Centered Male .26** (127) Female .15 (80) Four Habits Total Score Male .21*** (310) Female .13+ (187)
Note: Sample sizes are given in parentheses. p .10. * p .05. ** p .01. *** p .001.
+
be included due to the exclusion of physicians who did not have four or more patients. Table 5 (bottom panel) shows that for male physicians the correlation between Four Habits Total Score and patient satisfaction was stronger than the corresponding correlation for female physicians for all patients and outpatients; correlations for the individual habits showed the same pattern. However, the interactions as tested using either the two-level model or the meta-analytic contrast were not significant. 4. Discussion and conclusion 4.1. Discussion The relations among physician gender, patient-centered behavior, and patient satisfaction were examined in three practice settings within a hospital. As in past research [1], female physicians performed in a more patient-centered way than male physicians did, but this difference varied with the setting: although there were some differences in the outpatient setting, the strongest differences occurred with inpatients and in the emergency room. For the RIAS, the difference was due to the male physicians in those two settings being notably less patient-centered than in the outpatient setting. For the Four Habits, the difference was due to female physicians being notably more patient-centered in the inpatient setting compared to the other two settings. Because the inpatient and emergency room situations would generally present a higher level of emotional stress for both physician and patient, these differences in physician performance may reflect gender-linked responses to emotional demand, consistent with documented physician gender differences in their tendencies to talk about emotions and psychosocial issues with patients [1]. Possibly, female physicians are more likely to meet the challenge through heightened or sustained patient-centered behavior while male physicians are more likely to display a more emotionally disconnected, task-oriented manner. We can be sure this is an effect of the settings and not self-selection, since in Norwegian hospitals physicians are assigned and do not make their own choices about these assignments. However, in a correlational study there are inevitably factors that confound clear interpretation of study results. The settings could not be equated on patient characteristics because of very real differences that probably exist in patient characteristics across the settings. Also, in this study there was an unbalanced number of patients across settings. For patient satisfaction, there was no physician gender difference in the outpatient setting, consistent with earlier research [16], even though there was evidence that the female physicians were somewhat more patient-centered in that setting, suggestive of a gender bias in evaluations. On the other hand,
317
female physicians earned significantly higher satisfaction than male physicians in the inpatient and emergency room settings. Thus, female physicians were accorded the most satisfaction, compared to male physicians, in the settings in which females’ performance was in fact the most different from males’ in terms of patient-centeredness. Thus it appears that in these settings satisfaction was reflecting the female physicians’ higher patientcenteredness. We also tested for a gender bias in a different way, by calculating the correlation of patient-centeredness with satisfaction separately for male than female physicians. This withingender analysis addresses the gender bias question in a different way from the mean difference comparisons discussed above. In terms of the correlations, the pattern clearly supported the hypothesis that satisfaction and patient-centeredness would be more strongly associated for male than female physicians, with the most dramatic differences appearing for the RIAS Global Ratings and in the inpatient and emergency room settings. However, though the meta-analytic gender comparison was significant for the Global Ratings Patient-Centered Composite, we could not demonstrate significant analogous interaction effects of physician gender and patient-centered behavior on patient satisfaction using the two-level model. The most likely explanations for this are (a) high within physician variability both for patient reported and coder reported outcomes and (b) high between individual physicians variability that is much more dominant than the gender effect. Hence, we found indications that patients might hold a double standard when they rate their satisfaction with their physicians as suggested by previous research [19,21], but the effect in the present study tended to be overshadowed by individual patient and physician differences. Aside from the limitations already mentioned, the present study was limited geographically, and was limited to only one hospital setting within Norway. However, the results gain generality by being consistent with previous meta-analytic and experimental results regarding physician gender, behavior, satisfaction, and the different relations between satisfaction and behavior for male versus female physicians. 4.2. Conclusion This study continues to show that male and female physicians’ practice styles are different in ways that have implications for quality of care and patient satisfaction. Partial support emerged for previous research suggesting that patients might not evaluate male and female physicians using the same standards. If this phenomenon is widespread either among patients or among those who socialize young physicians into the profession, it would be very unfortunate, indeed ironic, owing to the fact that the patientcentered model of care is one that requires both competence and warmth, a combination that is optimal according to more than one theoretical framework. Both the stereotype-content model [42] and conceptions of psychological androgyny [43] predict the best outcomes when both competence and warmth are high. Patientcentered medicine requires this very combination of stereotypically ‘‘masculine’’ qualities of competence, authority, expertise, independence, and self-confidence, and stereotypically ‘‘feminine’’ qualities of warmth, sensitivity, a caring attitude, a relationship orientation, and interpersonal responsiveness. An equitable evaluation of all physicians according to these dual standards is necessary. In the present study, all physicians were hospital staff, working in teams most of their time and with no private practice. If female physicians under such conditions do not get credit for their patient-centeredness, they might over time develop a more malestyle communication, particularly in outpatient settings where less is at stake.
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
318
4.3. Practice implications Gender differences in practice style appear to be more prominent in settings where the patients are more vulnerable. In these settings female physicians were more patient-centered than their male peers, and this paralleled the pattern of patient satisfaction in those settings. However, there was still evidence that patient-centered behavior in male physicians is more highly valued than patientcentered behavior in female physicians. If female physicians do not consistently receive sufficient credit for their patient-centeredness in the eyes of patients (or their peers or superiors), this could lead female physicians to adopt less patient-centered behavior if their ‘‘female’’-like patient-centered behaviors are not sufficiently appreciated. This would not only be a loss in terms of patient care but also for female physicians, who may suffer not only loss of regard and self-confidence (which is already lower in female than male medical students [44]), but even monetary losses considering patient satisfaction is increasingly used in determinations of financial compensation for physicians [45]. Role of funding The data collection was funded by The Regional Health Enterprise for South-East Norway (Helse Sør-Øst RHF), which played no role in the conduct, analysis, or decisions about dissemination/publication of this research. Conflict of interest None declared. Acknowledgments We are in debt to Ba˚rd Fossli Jensen, MD, PhD for his efforts to collect this rich dataset, and to our co-investigator on that project Arnstein Finset. The data collection was funded by The Regional Health Enterprise for South-East Norway (Helse Sør-Øst RHF). References [1] Roter DL, Hall JA, Aoki Y. Physician gender effects in medical communication: a meta-analytic review. J Amer Med Assoc 2002;288:756–64. [2] Meeuwesen L, Schaap C, Van der Staak C. Verbal analysis of doctor patient communication. Soc Sci Med 1991;32:1143–50. [3] Roter DL, Lipkin Jr M, Korsgaard A. Gender differences in patients’ and physicians’ communication during primary care medical visits. Med Care 1991;29:1083–93. [4] Wasserman RC, Inui TS, Barriatua RD, Carter WB, Lippincott P. Pediatric clinicians’ support for parents makes a difference: an outcome-based analysis of clinician–parent interaction. Pediatrics 1984;74:1047–53. [5] Mendez A, Shymansky JA, Wolraich M. Verbal and non-verbal behaviour of doctors while conveying distressing information. Med Educ 1986;20:437–43. [6] Hall JA, Irish JT, Roter DL, Ehrlich CM, Miller LH. Gender in medical encounters: an analysis of physician and patient communication in a primary care setting. Health Psychol 1994;13:384–92. [7] Zandbelt LC, Smets EMA, Oort FJ, Godfried MH, de Haes HCJM. Determinants of physicians’ patient-centered behaviour in the medical specialist encounter. Soc Sci Med 2006;63:899–910. [8] Lurie N, Slater J, McGovern P, Ekstrum J, Quam L, Margolis K, et al. Preventive care for women: does the sex of the physician matter? N Engl J Med 1993;329:478–82. [9] Hall JA, Roter DL, Blanch DC, Frankel RM. Nonverbal sensitivity in medical students: implications for clinical interactions. J Gen Intern Med 2009;24:1217–22. [10] Chen DCR, Kirshenbaum DS, Yan J, Kirshenbaum E, Aseltine RH. Characterizing changes in student empathy throughout medical school. Med Teach 2012;23:305–11. [11] Berg K, Majdan JF, Berg D, Veloski J, Hojat M. Medical students’ self-reported empathy and simulated patients’ assessments of student empathy: an analysis by gender and ethnicity. Acad Med 2011;86:984–8. [12] Krupat E, Rosenkranz SL, Yeager CM, Barnard K, Putnam SM, Inui TS, et al. The practice orientations of physicians and patients: the effect of doctor–patient congruence on satisfaction. Patient Educ Couns 2000;39:49–59.
[13] de Boer D, Delnoij D, Rademakers J. The importance of patient-centered care for various patient groups. Patient Educ Couns 2013;90:405–10. [14] Deledda G, Moretti F, Rimondini M, Zimmermann C. How patients want their doctor to communicate: a literature review on primary care patients’ perspective. Patient Educ Couns 2013;90:297–306. [15] Swenson SL, Buell S, Zettler P, White M, Ruston DC, Lo B, et al. Patient-centered communication: do patients really prefer it? J Gen Intern Med 2004;19: 1069–79. [16] Hall JA, Blanch-Hartigan D, Roter DL. Patients’ satisfaction with male versus female physicians: a meta-analysis. Med Care 2011;49:611–7. [17] Derose KP, Hays RD, McCaffrey DF, Baker DW. Does physician gender affect satisfaction of men and women visiting the emergency department? J Gen Intern Med 2001;16:218–26. [18] Hall JA, Roter DL, Katz NR. Meta-analysis of correlates of provider behavior in medical encounters. Med Care 1988;26:657–75. [19] Blanch-Hartigan D, Hall JA, Roter DL, Frankel RM. Gender bias in patients’ perceptions of patient-centered behaviors. Patient Educ Couns 2010;80:315–20. [20] Krupat E, Frankel R, Stein T, Irish J. The Four Habits Coding Scheme: validation of an instrument to assess clinicians’ communication behavior. Patient Educ Couns 2006;62:38–45. [21] Hall JA, Roter DL, Blanch-Hartigan D, Schmid Mast M, Pitegoff CA. How patientcentered do female physicians need to be?: patient satisfaction with male and female physicians’ identical behaviors; 2013 [submitted for publication]. [22] Hall JA, Gunnery SD. Gender differences in nonverbal communication. In: Hall JA, Knapp ML, editors. Handbook of nonverbal communication. Berlin: De Gruyter Mouton; 2013. p. 639–69. [23] Kring AM, Gordon AH. Sex differences in emotion: expression, experience, and physiology. J Pers Soc Psychol 1998;74:686–703. [24] Eagly AH, Johnson BT. Gender and leadership style: a meta-analysis. Psychol Bull 1990;108:233–56. [25] Su R, Rounds J, Armstrong PI. Men and things, women and people: a metaanalysis of sex differences in interests. Psychol Bull 2009;135:859–84. [26] Dindia K, Allen M. Sex differences in self-disclosure: a meta-analysis. Psychol Bull 1992;112:106–24. [27] O’Brien E, Konrath SH, Gru¨hn D, Hagen AL. Empathic concern and perspective taking: linear and quadratic effects of age across the adult life span. J Gerontol B Psychol Sci Soc Sci 2013;68B:168–75. [28] Briton NJ, Hall JA. Beliefs about female and male nonverbal communication. Sex Roles 1995;23:79–90. [29] Hall JA, Carter JD, Horgan TG. Gender differences in the nonverbal communication of emotion. In: Fischer AH, editor. Gender and emotion: social psychological perspectives. Paris: Cambridge University Press; 2000. p. 97– 117. [30] Mead N, Bower P. Patient-centredness: a conceptual framework and review of the empirical literature. Soc Sci Med 2000;51:1087–110. [31] Heilman ME, Okimoto TG. Why are women penalized for success at male tasks?: the implied communality deficit. J Appl Psychol 2007;92:81–92. [32] Heilman ME, Chen JJ. Same behavior, different consequences: reactions to men’s and women’s altruistic citizenship behavior. J Appl Psychol 2005;90:431–41. [33] Heilman ME. Description and prescription: how gender stereotypes prevent women’s ascent up the organizational ladder. J Soc Issues 2001;57: 657–74. [34] Blanch-Hartigan D, Hall JA, Krupat E, Irish JT. Can naı¨ve viewers put themselves in the patients’ shoes?: reliability and validity of the analogue patient methodology. Med Care 2013;51:16–21. [35] Fossli Jensen B, Gulbrandsen P, Dahl FA, Krupat E, Frankel RM, Finset A. Effectiveness of a short course in clinical communication skills for hospital doctors: results of a crossover randomized controlled trial (ISRCTN22153332). Patient Educ Couns 2011;84:163–9. [36] Gulbrandsen P, Benth JS, Dahl FA, Fossli Jensen B, Finset A, Hall JA. Specialist physicians’ sensitivity to patient affect and satisfaction. Med Care 2012;50:290–3. [37] Gulbrandsen P, Fossli Jensen B. Post-recruitment confirmation of consent by SMS. J Med Ethics 2010;36:126–8. [38] Roter DL, Larson S. The Roter Interaction Analysis System (RIAS): utility and flexibility for analysis of medical interactions. Patient Educ Couns 2002;46:243–51. [39] Fossli Jensen B, Gulbrandsen P, Dahl FA, Saltyte-Benth J, Krupat E, Finset A. The interrater reliability of the Four Habits Coding Scheme as part of a randomized controlled trial. Patient Educ Couns 2010;80:405–9. [40] Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive meta-analysis version 2. Englewood, NJ: Biostat; 2005. [41] Hall JA, Epstein AM, De Ciantis M, McNeil BJ. Physicians’ liking for their patients: further evidence for the role of affect in medical care. Health Psychol 1993;12:140–6. [42] Fiske ST, Cuddy AJC, Glick P, Xu J. A model of (often mixed) stereotype content: competence and warmth respectively follow from perceived status and competition. J Pers Soc Psychol 2002;82:878–902. [43] Taylor MC, Hall JA. Psychological androgyny: theories, methods, and conclusions. Psychol Bull 1982;92:347–66. [44] Blanch DC, Hall JA, Roter DL, Frankel RM. Medical student gender and issues of confidence. Patient Educ Couns 2008;72:374–81. [45] Rosenthal MB, Landon BE, Normand ST, Frank RG, Epstein AM. Pay for performance in commercial HMOs. N Engl J Med 2006;355:1895–902.
Contents lists available at ScienceDirect
Patient Education and Counseling journal homepage: www.elsevier.com/locate/pateducou
Communication Study
Physician gender, physician patient-centered behavior, and patient satisfaction: A study in three practice settings within a hospital Judith A. Hall a,*, Pa˚l Gulbrandsen b,c, Fredrik A. Dahl c a
Department of Psychology, Northeastern University, Boston, USA Institute of Clinical Medicine, University of Oslo, Oslo, Norway c Akershus University Hospital, Lørenskog, Norway b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 27 November 2013 Received in revised form 4 March 2014 Accepted 16 March 2014
Objective: To compare male and female physicians on patient-centeredness and patients’ satisfaction in three practice settings within a hospital; to test whether satisfaction is more strongly predicted by patient-centeredness in male than female physicians. Methods: Encounters between physicians (N = 71) and patients (N = 497) in a hospital were videotaped and patients’ satisfaction was measured. Patient-centeredness was measured by trained coders. Results: In the outpatient setting, female physicians were somewhat more patient-centered than male physicians; patient satisfaction did not differ. In the inpatient and emergency room settings, female physicians were notably more patient-centered than male physicians; satisfaction paralleled these differences. Nevertheless, there was some, though mixed, evidence that patient-centeredness predicted satisfaction more strongly in male than female physicians, suggesting that patients valued patientcentered behavior more in male than female physicians. Conclusion: Even though satisfaction mirrored the different behavior styles of male and female physicians in the inpatient and emergency room settings, in all settings male physicians got somewhat more credit for being patient-centered than female physicians did. Practice implications: If female physicians do not consistently receive credit for high patientcenteredness in the eyes of patients, this could lead female physicians to reduce their patient-centered behavior. ß 2014 Elsevier Ireland Ltd. All rights reserved.
Keywords: Physician gender Patient-centeredness Patient satisfaction Gender bias Outpatient Inpatient Emergency room
1. Introduction Female physicians are, on average, more patient-centered than their male counterparts. Female physicians and/or female medical students spend longer with patients, behave more positively verbally and nonverbally, talk more about psychosocial and emotional issues, use more partnership statements, offer more nonverbal encouragement to talk, have more skill in interpersonal perception, listen better, have greater self-reported and perceived empathy, offer more expressions of respect or praise, and put a greater focus on prevention [1–11]. Female physicians have more humanistic and patient-centered attitudes about patient care [12], and patients generally desire a patient-centered style [13–15]. Yet, female physicians do not receive much more—usually, no more—satisfaction from patients [16]. This lack of a difference
* Corresponding author at: Department of Psychology, Northeastern University, Boston, MA 02115, USA. Tel.: +1 617 373 3790; fax: +1 617 373 8714. E-mail address: [email protected] (J.A. Hall). http://dx.doi.org/10.1016/j.pec.2014.03.015 0738-3991/ß 2014 Elsevier Ireland Ltd. All rights reserved.
suggests a bias when considered against the research showing that female physicians better fulfill the goals of patient-centered medicine. Most of the research relating physicians’ gender to their behavior and to patient satisfaction has been conducted in primary care [1,16]. Our first goal was to compare the patient-centered behaviors of male and female physicians in three settings in one general teaching hospital—outpatient, inpatient, and emergency room. The second goal was to compare patients’ satisfaction between male and female physicians within each practice setting. In the only study we know of from an emergency room, female physicians received significantly higher satisfaction ratings from patients than male physicians did, but only among female patients [17]. That study could not ascertain whether this effect was due to the lens through which patients viewed physicians or to behavioral differences between male and female physicians. The present study allowed an investigation of this question. The third goal was to test a hypothesis about gender bias in patients’ satisfaction that has been suggested in previous research.
314
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
The meta-analysis finding only a negligible tendency for patients to be more satisfied with female than male physicians [16] suggests such a bias because it does not fit logically with female physicians’ behavior patterns, the fact that patients generally want patient-centeredness, and the fact that they report more satisfaction when the physician is more patient-centered [18]. However, since most studies have a preponderance of male physicians, the latter result may obscure a different pattern for female physicians. Indeed, two studies have found that the correlation of either competence or satisfaction with objectively measured patientcenteredness was stronger for male than female providers. Among medical students interacting with a standardized patient [19], for female medical students there was no correlation between patient-centeredness as scored by trained coders [20] and analog patients’ competence ratings, whereas there was a positive correlation for male medical students. In a laboratory experiment using scripted high and low patient-centered behavior enacted by actor-physicians [21], patient-centeredness strongly predicted analog patients’ satisfaction with male physicians, but it predicted satisfaction with female physicians either not at all or more weakly. These results clearly suggest a bias in patients’ evaluations. This could result from the overlap between the behaviors and skills considered patient-centered and the behavior and skills of women in general, which are well studied, ubiquitous, and reflected in strong societal stereotypes. According to research in nonclinical populations women are, and are stereotyped to be, more empathic, more warm and personable, more interested in relationships, more status leveling, more comfortable with emotional self-disclosure, and more skilled in nonverbal communication than men [22–29]. All of these expected and real female traits have their counterpart in the collection of attributes called patient-centeredness [30]. Because of this overlap, high patient-centered behavior in female physicians may be discounted as a marker for being a good physician, because such behavior is viewed simply as expected, even exemplary, female behavior but not necessarily a manifestation of high quality medical performance. Such behavior may be appreciated, while not adding to the patient’s appraisal of the physician as a fine professional. On the other hand, high patientcenteredness by a male physician is not part of the male stereotype and cannot be assimilated to a male schema. Instead, it is assimilated to a ‘‘good doctor schema’’ and the male physician gets a disproportionate boost in the patient’s eyes. Thus, patients may have a bias when evaluating the professional performance of physicians, with the result that a significant aspect of female physicians’ performance is not recognized as professionally relevant. This kind of subtle bias has also been seen in other organizational settings, where job-relevant ‘‘female’’ behaviors by women are not credited as contributing to high job performance when they should be [31–33]. Both of the studies described above showing stronger correlations between patient-centeredness and evaluations of male clinicians [19,21] were artificial in various ways. The first study was on medical students interacting with a standardized patient in a videotaped clinical examination, and the competence ratings were made by undergraduate analog patients. The second study used videos in which male and female actors played physicians and the satisfaction data were again collected from undergraduate analog patients. Although the analog patient method has received validational support [34], research must be conducted with practicing physicians and real patients. In light of the foregoing research, we would predict greater patient-centeredness in female than male physicians, though the comparison of such differences may vary across the different settings. We also would expect that satisfaction with female physicians would not be appreciably different from satisfaction
with male physicians, at least in the outpatient setting where most existing research has been done. Finally, in keeping with the experimental research relating patient-centeredness to satisfaction, we would expect this correlation to be stronger for male than female physicians. 2. Methods 2.1. Participants and procedure The data stem from a randomized controlled trial [35,36]. It consisted of 497 videotaped physician–patient interactions in the outpatient, inpatient, and emergency room settings of a general teaching hospital in Lørenskog, Norway, with 71 non-psychiatrist physicians who each saw between 1 and 8 patients. Twenty-one physicians were observed in more than one setting. The participation rate of the physicians was 69% (71 of 103 invited). After the interaction, patients filled in a questionnaire evaluating their interaction with the physician. All videos were included, as there was no significant difference in change of behavior between female and male physicians following the intervention [35]. Physicians were informed that they were due for observation shortly before filming and could not adjust their patient lists. Patients were recruited consecutively. Of 553 eligible patients, 519 (94%) consented [37]. Technical difficulties with videotapes left 497 encounters for analyses. Literate children completed questionnaires themselves, assisted by their parents when needed. 2.2. Encounters Table 1 provides descriptive information on the physicians and patients, as well as visit characteristics. 2.3. Measures 2.3.1. Roter Interaction Coding System Two hundred seven of the videotapes were rated by trained and reliable coders using the Global Ratings from the Roter Interaction Coding System (RIAS), a widely used descriptive system for provider–patient interaction [38]. After watching the interaction, coders made global ratings of the physician on the following scales: interested, friendly, engaged, sympathetic, dominant, anxious, and angry. The first five are relevant to the patient-centered concept (dominant in an inverse relation) and are analyzed in the present article. Anxious is not relevant to patient-centeredness, while the rating of anger, though relevant to patient-centeredness (in an inverse relation), had too limited variance to be useful. Examination of the correlations among the five patient-centered ratings Table 1 Physician and patient characteristics. Physicians (N = 71)
Patients (N = 497)
Age, mean (SD) Sex
40.3 (8.6) Females 30 (42%), males 41 (58%)
46.3 (24.6) Females 256 (52%), males 241 (48%)
Specialty
Internal medicine 25 (35%) Surgical disciplines 18 (26%) Anesthesiology 5 (7%) Neurology 8 (11%) Pediatrics 8 (11%) Gynecology 7 (10%) Residents 33 (47%) Consultants 38 (53%) 56 26 13
Position Settinga
a
Outpatient 375 (76%) Inpatient 81 (16%) Emergency room 41 (8%)
Twenty-one physicians were observed in more than one setting.
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
revealed strong correlations among the first four (Cronbach’s alpha = .86), while dominant was negligibly related to them (median r = .08). Therefore the ratings of interested, friendly, engaged, and sympathetic were averaged into a composite called the Global Ratings Patient-Centered Composite (M = 4.13, SD = .81, range = 2.25–6.00, possible range = 1–6), and dominant was kept separate (M = 3.28, SD = 1.13, range = 1–6, possible range = 1–6). 2.3.2. Four Habits Coding Scheme Physicians’ patient-centeredness was also scored with the Four Habits Coding Scheme [20] in which recorded provider–patient interactions are rated by trained coders using a 23-item scale, with higher scores indicating more patient-centered behavior. The habit scores are: I, Invest in the Beginning, 6 items; II, Elicit the Patient’s Agenda, 3 items; III, Demonstrate Empathy, 4 items; and IV, Invest in the End, 10 items. Four experienced students educated in psychology were trained on the system. Videotapes were rated in groups of 20 until acceptable interrater reliability (IRR) (Pearson r > .70) was achieved [39]. The four habits were analyzed separately as well as in a total score (M = 60.10, SD = 14.45, range = 29–108, possible range = 23–115). 2.3.3. Satisfaction Patient satisfaction was assessed by the global item ‘‘Using any number from 0 to 10, where 0 is the worst possible consultation with a doctor and 10 is the best possible consultation with a doctor, what number would you use to rate this consultation?’’ taken from The Consumer Assessment of Health Care Providers and Systems (http://www.ahrq.gov) (M = 8.76, SD = 1.32, range = 2–10). 2.4. Statistical analysis One set of analyses examined relations between variables in the whole sample and in the three practice settings without taking account of the nesting of patients within physicians. Although significance tests are compromised when patients who are nested within physicians are considered as separate units of analysis, estimates of effect size (i.e., correlation coefficients) are not biased and give a good picture of the relative magnitudes of effects. Pearson correlations and analyses of variance were used. Gender differences were expressed as the point-biserial correlation, which is the Pearson correlation between gender (coded female = 0, male =1) and another study variable. Positive correlations mean male physicians were higher on the variable than female physicians; negative correlations mean the reverse. In addition, a two-level model was employed in examining physician gender differences in order to take account of the nesting of patients within physicians. When examining relations between patients’ satisfaction and the patient-centeredness of male versus female physicians, Pearson correlations were calculated, and also two-level models with a term for the physician gender patient-centeredness interaction. This interaction was predicted to show a stronger relation between patient-centeredness and satisfaction for male than female physicians. An additional analytic method which also controls for nesting was used to test this interaction, by calculating the correlation between patient-centeredness and patient satisfaction for each physician separately, so that each correlation was based on only that physician’s patients. Thus, each physician had a correlation between his/her patient-centeredness and his/her patients’ satisfaction, thereby putting the analysis at the level of the individual physicians. The question was then posed whether these correlations were stronger for male than female physicians. A meta-analytic fixed-effects contrast between male and female physicians was then performed, which is conceptually analogous to an independent-samples t-test [40]. This method was used by
315
Table 2 Pearson correlations between physician gender and RIAS Global Ratings. RIAS scale
Interested Friendly Engaged Sympathetic Patient-Centered Composite Dominant
All
Outpatient
Inpatient
Emergency
(207)
(145)
(41)
(21)
.11 .26*** .02 .20** .18** .04
.03 .17* .02 .11 .09 .06
.27+ .40** .18 .38* .37* .31*
.28 .54** .30 .45* .47* .02
Notes: Gender is coded 0 = female, 1 = male. N is given in parentheses. Negative correlations mean female physicians scored higher than male physicians. + p .10. * p .05. ** p .01. *** p .001.
Hall and colleagues in a study to show that physicians liked their healthier patients more than their less healthy patients [41]. In the present study, only physicians who had four or more patients coded for patient-centeredness and who had variance across patients for both patient-centeredness and satisfaction were used in this analysis (n = 19). 3. Results 3.1. Physician gender differences for RIAS Global Ratings Table 2 shows physician gender differences for the RIAS Global Ratings (n = 207), for both the whole sample and for the three settings. The gender differences are expressed as point-biserial correlations; these correlations carry the identical p-value as ttests but add more information because they are expressed in a readily understood effect size metric (the Pearson correlation, denoted as r). For the sample as a whole, female physicians were rated by the trained coders as significantly more friendly and sympathetic than male physicians, contributing to a significant Global Ratings Patient-Centered Composite as shown in the table. However, there were notable differences between settings. Gender differences were weak in the outpatient setting. The only significant correlation indicated that female physicians were rated by the trained coders as more friendly than male physicians. The inpatient setting showed stronger gender differences than the outpatient setting; in the inpatient setting, females were significantly more friendly and sympathetic, and marginally more interested, contributing to a significant Global Ratings PatientCentered Composite as shown in the table. The gender difference became even stronger in the emergency room setting for friendliness, sympathy, and the Global Ratings Patient-Centered Composite. Controlling patient gender, patient age, and the length of the encounter using partial correlations had no impact except in the inpatient condition. In that condition, all of the physician gender differences actually became stronger when these variables were controlled: interested, r = .36, p < .05; friendly, r = .53, p < .001; engaged, r = .30, p < .07; sympathetic, r = .45, p < .01; and Global Ratings Patient-Centered Composite, r = .49, p < .01 (all partial correlations, df = 36). Turning to physician dominance, in the inpatient setting, male physicians were significantly more dominant than female physicians (Table 2). Controlling for patient gender, patient age, and the length of the encounter had no impact on the dominance correlations in any of the settings. In the two-level models for the whole sample, female physicians remained more friendly (p < .01) and sympathetic (p < .05) than male physicians, and the Global Ratings PatientCentered Composite was marginally significant (p = .07). In the
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
316
Table 3 Mean levels of the Global Ratings Patient-Centered Composite, Four Habits Total Score, and patient satisfaction for male and female physicians. Physician gender
Outpatient
Global Ratings Patient-Centered Composite Male 4.09 Female 4.25 Four Habits Total Score Male 58.73 Female 60.47 Patient satisfaction Male 8.91 Female 8.96
Inpatient
Table 4 Pearson correlations between physician gender and Four Habits Coding Scheme. Four Habits scale
Emergency
3.90 4.49
3.50 4.27
59.69 66.89
57.00 62.56
7.40 8.68
7.86 8.83
outpatient setting, effects were lower in significance with only the physician gender difference in friendliness approaching significance (p < .09). In the inpatient setting, no effects were significant. However, in the emergency room the difference in friendliness and the Global Ratings Patient-Centered Composite both were marginally significant (p < .06 and = .10, respectively), in spite of the very small sample size. This speaks to the strength of the correlations as shown in Table 2. The correlations shown in Table 2 express the magnitude of gender differences in terms of standardized effect sizes, but they do not give insight into the actual levels of performance of male and female physicians. A one-way analysis of variance (ANOVA) comparing the three settings on the Global Ratings PatientCentered Composite was performed separately for male and female physicians (Table 3, top panel). Female physicians were consistent across the three settings, F(2, 77) < 1.0. On the other hand, male physicians behaved differently across the three settings, F(2, 124) = 2.58, p < .09. A linear trend showed that male physicians’ behavior was progressively less patient-centered from outpatient to inpatient to emergency room, p < .05. For the dominance rating, there were no significant effects, but for male physicians there was a marginally significant (p < .10) quadratic trend showing higher dominance in the inpatient setting than in the other two settings, creating the notable gender difference for dominance in the inpatient setting. 3.2. Physician gender differences for the Four Habits Coding Scheme Table 4 shows gender differences for the Four Habits Coding Scheme, expressed as correlations. It is evident that female physicians scored higher than male physicians, although the correlations sometimes were not significant in the separate settings. The strongest gender effects occurred on Habit I (Invest in the Beginning, with items such as greetings and expansion of concerns) and Habit III (Demonstrate Empathy, with items such as expression of emotion, acceptance of feelings, and empathy). These are the two habits that pertain mainly to socioemotional behavior, in contrast to Habits II and IV, which are more about task functions. The correlations did not change when patient gender, patient age, and the length of the encounter were controlled. After applying the two-level model to the whole sample to correct for interdependence among patients within physicians, the effects for Habits I and III remained significant (p .05). There were no significant gender differences within the separate settings. To examine actual levels of Four Habits performance, ANOVAs like those described above were performed for the Four Habits Total Score and the individual habits. For Total Score, there were no effects for male physicians, F(2, 307) < 1.0. However, female physicians’ performance varied over settings, F(2, 184) = 3.76, p < .05, with a significant quadratic trend indicating that Total Score was highest in the inpatient setting, p < .05 (Table 3, middle panel). For female physicians, this quadratic trend was also significant for both Habits I and III.
I: Invest in the Beginning II: Elicit the Patient’s Perspective III: Demonstrate Empathy IV: Invest in the End Total Four Habits
All
Outpatient Inpatient Emergency
(497)
(375)
.15*** .01 .16*** .07 .12**
.11* .04 .10* .01 .06
(81) .21+ .12 .30** .21+ .27**
(41) .19 .10 .10 .20 .19
Notes: Gender is coded 0 = female, 1 = male. Sample sizes are given in parentheses. Negative correlations mean female physicians scored higher than male physicians. + p .10. * p .05. ** p .01. *** p .001.
3.3. Physician gender differences for patient satisfaction Consistent with previous research [16], patients were not more satisfied overall with either gender of physician; the overall correlation between physician gender and satisfaction was r(495) = .07, p < .13. This overall lack of a difference was mainly due to the outpatient visits, for which the relation was r(373) = .02, p < .69. Gender differences were notable in the other settings, however. Patients were more satisfied with female than male physicians both in the inpatient setting, r(79) = .35, p < .001, and in the emergency room, r(39) = .44, p < .01 (Table 3, lower panel). The correlations did not change when patient gender, patient age, and the length of the encounter were controlled. Also, there was no evidence, unlike the earlier published emergency room study [17], that the physician gender difference in the emergency room was limited to female patients. The two-level models similarly did not reveal any physician gender effects either overall or in the outpatient setting, but in the inpatient setting female physicians received higher satisfaction, p < .01, and in the emergency room this was also the case, p < .06. ANOVAs were conducted to examine satisfaction levels across the three settings. Satisfaction with male physicians was much lower for inpatient and emergency room than for outpatient, F(2, 307) = 24.91, p < .001. There was no difference across settings for female physicians, F(2, 184) < 1.0. 3.4. Is there a gender-related ‘‘double standard’’ for patient satisfaction? According to the logic put forth in the Introduction, indicators of patient-centeredness were predicted to be more strongly predictive of patients’ satisfaction for male than female physicians. Table 5 (top panel) shows these correlations for the Global Ratings Patient-Centered Composite; the pattern for the separate RIAS ratings was very similar. The prediction that the male physicians’ correlation would be stronger than the female physicians’ correlation was supported, more so for inpatients and emergency room than the outpatient setting. There were no significant interaction effects when examining this same question in the twolevel models. However, the meta-analysis based on physicians found that the average correlation between the Global Ratings Patient-Centered Composite and satisfaction was both strong and significant for male physicians (r = .50, p < .01, n = 12) while the corresponding correlation for female physicians was negligible (r = .06, p = .76, n = 7). Furthermore, the meta-analytic contrast comparing male and female physicians on this correlation was significant, p < .03. Thus, this analysis found significant support for the hypothesis, but note must be made that only 19 of the 61 physicians who had at least one patient scored for the RIAS could
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318 Table 5 Correlations of patient satisfaction with Global Ratings Patient-Centered Composite and Four Habits Total Score, separately for male and female physicians. Variable
All patients
Outpatients
Inpatients
Emergency
(207)
(145)
(41)
(21)
Composite .18+ (97) .25+ (48)
.41+ (21) .09 (20)
.70* (9) .05 (12)
.26*** (260) .12 (115)
.14 (36) .22 (45)
.34 (14) .24 (27)
Global Ratings Patient-Centered Male .26** (127) Female .15 (80) Four Habits Total Score Male .21*** (310) Female .13+ (187)
Note: Sample sizes are given in parentheses. p .10. * p .05. ** p .01. *** p .001.
+
be included due to the exclusion of physicians who did not have four or more patients. Table 5 (bottom panel) shows that for male physicians the correlation between Four Habits Total Score and patient satisfaction was stronger than the corresponding correlation for female physicians for all patients and outpatients; correlations for the individual habits showed the same pattern. However, the interactions as tested using either the two-level model or the meta-analytic contrast were not significant. 4. Discussion and conclusion 4.1. Discussion The relations among physician gender, patient-centered behavior, and patient satisfaction were examined in three practice settings within a hospital. As in past research [1], female physicians performed in a more patient-centered way than male physicians did, but this difference varied with the setting: although there were some differences in the outpatient setting, the strongest differences occurred with inpatients and in the emergency room. For the RIAS, the difference was due to the male physicians in those two settings being notably less patient-centered than in the outpatient setting. For the Four Habits, the difference was due to female physicians being notably more patient-centered in the inpatient setting compared to the other two settings. Because the inpatient and emergency room situations would generally present a higher level of emotional stress for both physician and patient, these differences in physician performance may reflect gender-linked responses to emotional demand, consistent with documented physician gender differences in their tendencies to talk about emotions and psychosocial issues with patients [1]. Possibly, female physicians are more likely to meet the challenge through heightened or sustained patient-centered behavior while male physicians are more likely to display a more emotionally disconnected, task-oriented manner. We can be sure this is an effect of the settings and not self-selection, since in Norwegian hospitals physicians are assigned and do not make their own choices about these assignments. However, in a correlational study there are inevitably factors that confound clear interpretation of study results. The settings could not be equated on patient characteristics because of very real differences that probably exist in patient characteristics across the settings. Also, in this study there was an unbalanced number of patients across settings. For patient satisfaction, there was no physician gender difference in the outpatient setting, consistent with earlier research [16], even though there was evidence that the female physicians were somewhat more patient-centered in that setting, suggestive of a gender bias in evaluations. On the other hand,
317
female physicians earned significantly higher satisfaction than male physicians in the inpatient and emergency room settings. Thus, female physicians were accorded the most satisfaction, compared to male physicians, in the settings in which females’ performance was in fact the most different from males’ in terms of patient-centeredness. Thus it appears that in these settings satisfaction was reflecting the female physicians’ higher patientcenteredness. We also tested for a gender bias in a different way, by calculating the correlation of patient-centeredness with satisfaction separately for male than female physicians. This withingender analysis addresses the gender bias question in a different way from the mean difference comparisons discussed above. In terms of the correlations, the pattern clearly supported the hypothesis that satisfaction and patient-centeredness would be more strongly associated for male than female physicians, with the most dramatic differences appearing for the RIAS Global Ratings and in the inpatient and emergency room settings. However, though the meta-analytic gender comparison was significant for the Global Ratings Patient-Centered Composite, we could not demonstrate significant analogous interaction effects of physician gender and patient-centered behavior on patient satisfaction using the two-level model. The most likely explanations for this are (a) high within physician variability both for patient reported and coder reported outcomes and (b) high between individual physicians variability that is much more dominant than the gender effect. Hence, we found indications that patients might hold a double standard when they rate their satisfaction with their physicians as suggested by previous research [19,21], but the effect in the present study tended to be overshadowed by individual patient and physician differences. Aside from the limitations already mentioned, the present study was limited geographically, and was limited to only one hospital setting within Norway. However, the results gain generality by being consistent with previous meta-analytic and experimental results regarding physician gender, behavior, satisfaction, and the different relations between satisfaction and behavior for male versus female physicians. 4.2. Conclusion This study continues to show that male and female physicians’ practice styles are different in ways that have implications for quality of care and patient satisfaction. Partial support emerged for previous research suggesting that patients might not evaluate male and female physicians using the same standards. If this phenomenon is widespread either among patients or among those who socialize young physicians into the profession, it would be very unfortunate, indeed ironic, owing to the fact that the patientcentered model of care is one that requires both competence and warmth, a combination that is optimal according to more than one theoretical framework. Both the stereotype-content model [42] and conceptions of psychological androgyny [43] predict the best outcomes when both competence and warmth are high. Patientcentered medicine requires this very combination of stereotypically ‘‘masculine’’ qualities of competence, authority, expertise, independence, and self-confidence, and stereotypically ‘‘feminine’’ qualities of warmth, sensitivity, a caring attitude, a relationship orientation, and interpersonal responsiveness. An equitable evaluation of all physicians according to these dual standards is necessary. In the present study, all physicians were hospital staff, working in teams most of their time and with no private practice. If female physicians under such conditions do not get credit for their patient-centeredness, they might over time develop a more malestyle communication, particularly in outpatient settings where less is at stake.
J.A. Hall et al. / Patient Education and Counseling 95 (2014) 313–318
318
4.3. Practice implications Gender differences in practice style appear to be more prominent in settings where the patients are more vulnerable. In these settings female physicians were more patient-centered than their male peers, and this paralleled the pattern of patient satisfaction in those settings. However, there was still evidence that patient-centered behavior in male physicians is more highly valued than patientcentered behavior in female physicians. If female physicians do not consistently receive sufficient credit for their patient-centeredness in the eyes of patients (or their peers or superiors), this could lead female physicians to adopt less patient-centered behavior if their ‘‘female’’-like patient-centered behaviors are not sufficiently appreciated. This would not only be a loss in terms of patient care but also for female physicians, who may suffer not only loss of regard and self-confidence (which is already lower in female than male medical students [44]), but even monetary losses considering patient satisfaction is increasingly used in determinations of financial compensation for physicians [45]. Role of funding The data collection was funded by The Regional Health Enterprise for South-East Norway (Helse Sør-Øst RHF), which played no role in the conduct, analysis, or decisions about dissemination/publication of this research. Conflict of interest None declared. Acknowledgments We are in debt to Ba˚rd Fossli Jensen, MD, PhD for his efforts to collect this rich dataset, and to our co-investigator on that project Arnstein Finset. The data collection was funded by The Regional Health Enterprise for South-East Norway (Helse Sør-Øst RHF). References [1] Roter DL, Hall JA, Aoki Y. Physician gender effects in medical communication: a meta-analytic review. J Amer Med Assoc 2002;288:756–64. [2] Meeuwesen L, Schaap C, Van der Staak C. Verbal analysis of doctor patient communication. Soc Sci Med 1991;32:1143–50. [3] Roter DL, Lipkin Jr M, Korsgaard A. Gender differences in patients’ and physicians’ communication during primary care medical visits. Med Care 1991;29:1083–93. [4] Wasserman RC, Inui TS, Barriatua RD, Carter WB, Lippincott P. Pediatric clinicians’ support for parents makes a difference: an outcome-based analysis of clinician–parent interaction. Pediatrics 1984;74:1047–53. [5] Mendez A, Shymansky JA, Wolraich M. Verbal and non-verbal behaviour of doctors while conveying distressing information. Med Educ 1986;20:437–43. [6] Hall JA, Irish JT, Roter DL, Ehrlich CM, Miller LH. Gender in medical encounters: an analysis of physician and patient communication in a primary care setting. Health Psychol 1994;13:384–92. [7] Zandbelt LC, Smets EMA, Oort FJ, Godfried MH, de Haes HCJM. Determinants of physicians’ patient-centered behaviour in the medical specialist encounter. Soc Sci Med 2006;63:899–910. [8] Lurie N, Slater J, McGovern P, Ekstrum J, Quam L, Margolis K, et al. Preventive care for women: does the sex of the physician matter? N Engl J Med 1993;329:478–82. [9] Hall JA, Roter DL, Blanch DC, Frankel RM. Nonverbal sensitivity in medical students: implications for clinical interactions. J Gen Intern Med 2009;24:1217–22. [10] Chen DCR, Kirshenbaum DS, Yan J, Kirshenbaum E, Aseltine RH. Characterizing changes in student empathy throughout medical school. Med Teach 2012;23:305–11. [11] Berg K, Majdan JF, Berg D, Veloski J, Hojat M. Medical students’ self-reported empathy and simulated patients’ assessments of student empathy: an analysis by gender and ethnicity. Acad Med 2011;86:984–8. [12] Krupat E, Rosenkranz SL, Yeager CM, Barnard K, Putnam SM, Inui TS, et al. The practice orientations of physicians and patients: the effect of doctor–patient congruence on satisfaction. Patient Educ Couns 2000;39:49–59.
[13] de Boer D, Delnoij D, Rademakers J. The importance of patient-centered care for various patient groups. Patient Educ Couns 2013;90:405–10. [14] Deledda G, Moretti F, Rimondini M, Zimmermann C. How patients want their doctor to communicate: a literature review on primary care patients’ perspective. Patient Educ Couns 2013;90:297–306. [15] Swenson SL, Buell S, Zettler P, White M, Ruston DC, Lo B, et al. Patient-centered communication: do patients really prefer it? J Gen Intern Med 2004;19: 1069–79. [16] Hall JA, Blanch-Hartigan D, Roter DL. Patients’ satisfaction with male versus female physicians: a meta-analysis. Med Care 2011;49:611–7. [17] Derose KP, Hays RD, McCaffrey DF, Baker DW. Does physician gender affect satisfaction of men and women visiting the emergency department? J Gen Intern Med 2001;16:218–26. [18] Hall JA, Roter DL, Katz NR. Meta-analysis of correlates of provider behavior in medical encounters. Med Care 1988;26:657–75. [19] Blanch-Hartigan D, Hall JA, Roter DL, Frankel RM. Gender bias in patients’ perceptions of patient-centered behaviors. Patient Educ Couns 2010;80:315–20. [20] Krupat E, Frankel R, Stein T, Irish J. The Four Habits Coding Scheme: validation of an instrument to assess clinicians’ communication behavior. Patient Educ Couns 2006;62:38–45. [21] Hall JA, Roter DL, Blanch-Hartigan D, Schmid Mast M, Pitegoff CA. How patientcentered do female physicians need to be?: patient satisfaction with male and female physicians’ identical behaviors; 2013 [submitted for publication]. [22] Hall JA, Gunnery SD. Gender differences in nonverbal communication. In: Hall JA, Knapp ML, editors. Handbook of nonverbal communication. Berlin: De Gruyter Mouton; 2013. p. 639–69. [23] Kring AM, Gordon AH. Sex differences in emotion: expression, experience, and physiology. J Pers Soc Psychol 1998;74:686–703. [24] Eagly AH, Johnson BT. Gender and leadership style: a meta-analysis. Psychol Bull 1990;108:233–56. [25] Su R, Rounds J, Armstrong PI. Men and things, women and people: a metaanalysis of sex differences in interests. Psychol Bull 2009;135:859–84. [26] Dindia K, Allen M. Sex differences in self-disclosure: a meta-analysis. Psychol Bull 1992;112:106–24. [27] O’Brien E, Konrath SH, Gru¨hn D, Hagen AL. Empathic concern and perspective taking: linear and quadratic effects of age across the adult life span. J Gerontol B Psychol Sci Soc Sci 2013;68B:168–75. [28] Briton NJ, Hall JA. Beliefs about female and male nonverbal communication. Sex Roles 1995;23:79–90. [29] Hall JA, Carter JD, Horgan TG. Gender differences in the nonverbal communication of emotion. In: Fischer AH, editor. Gender and emotion: social psychological perspectives. Paris: Cambridge University Press; 2000. p. 97– 117. [30] Mead N, Bower P. Patient-centredness: a conceptual framework and review of the empirical literature. Soc Sci Med 2000;51:1087–110. [31] Heilman ME, Okimoto TG. Why are women penalized for success at male tasks?: the implied communality deficit. J Appl Psychol 2007;92:81–92. [32] Heilman ME, Chen JJ. Same behavior, different consequences: reactions to men’s and women’s altruistic citizenship behavior. J Appl Psychol 2005;90:431–41. [33] Heilman ME. Description and prescription: how gender stereotypes prevent women’s ascent up the organizational ladder. J Soc Issues 2001;57: 657–74. [34] Blanch-Hartigan D, Hall JA, Krupat E, Irish JT. Can naı¨ve viewers put themselves in the patients’ shoes?: reliability and validity of the analogue patient methodology. Med Care 2013;51:16–21. [35] Fossli Jensen B, Gulbrandsen P, Dahl FA, Krupat E, Frankel RM, Finset A. Effectiveness of a short course in clinical communication skills for hospital doctors: results of a crossover randomized controlled trial (ISRCTN22153332). Patient Educ Couns 2011;84:163–9. [36] Gulbrandsen P, Benth JS, Dahl FA, Fossli Jensen B, Finset A, Hall JA. Specialist physicians’ sensitivity to patient affect and satisfaction. Med Care 2012;50:290–3. [37] Gulbrandsen P, Fossli Jensen B. Post-recruitment confirmation of consent by SMS. J Med Ethics 2010;36:126–8. [38] Roter DL, Larson S. The Roter Interaction Analysis System (RIAS): utility and flexibility for analysis of medical interactions. Patient Educ Couns 2002;46:243–51. [39] Fossli Jensen B, Gulbrandsen P, Dahl FA, Saltyte-Benth J, Krupat E, Finset A. The interrater reliability of the Four Habits Coding Scheme as part of a randomized controlled trial. Patient Educ Couns 2010;80:405–9. [40] Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive meta-analysis version 2. Englewood, NJ: Biostat; 2005. [41] Hall JA, Epstein AM, De Ciantis M, McNeil BJ. Physicians’ liking for their patients: further evidence for the role of affect in medical care. Health Psychol 1993;12:140–6. [42] Fiske ST, Cuddy AJC, Glick P, Xu J. A model of (often mixed) stereotype content: competence and warmth respectively follow from perceived status and competition. J Pers Soc Psychol 2002;82:878–902. [43] Taylor MC, Hall JA. Psychological androgyny: theories, methods, and conclusions. Psychol Bull 1982;92:347–66. [44] Blanch DC, Hall JA, Roter DL, Frankel RM. Medical student gender and issues of confidence. Patient Educ Couns 2008;72:374–81. [45] Rosenthal MB, Landon BE, Normand ST, Frank RG, Epstein AM. Pay for performance in commercial HMOs. N Engl J Med 2006;355:1895–902.
