Empathy and sympathy are two important aspects of the physician–patient relationship, and both are linked to patient care.1–4 Despite the important roles of empathy and sympathy in patient care, the academic medicine community has reached no consensus on their definitions or on their different effects on patient outcomes. As a result, researchers have often used the two concepts interchangeably. But operational definitions of these terms are important. Clear definitions of empathy and sympathy will enable a framework both for developing content-specific instruments for their measurement and for examining their relevant clinical outcomes.
The conceptual confusion and interchangeable use of “empathy” and “sympathy” may not cause a serious problem in social psychology, but separating the two in the context of patient care is important. In social psychology, both empathy and sympathy can lead to a similar outcome (e.g., prosocial behavior), albeit for different behavioral motivations. A prosocial behavior induced by empathic understanding is more likely to be elicited by a sense of altruism.5 A prosocial behavior prompted by sympathetic feelings, however, is more likely to be triggered by a self-serving egoistic motivation to reduce the observer's personal distress.5
In patient care, the two constructs must be distinguished because, in this context, they lead to different outcomes. For example, Nightingale and colleagues6 have shown that in simulated conditions empathic physicians, compared with their sympathetic counterparts, used resources appropriately by ordering fewer laboratory tests, had less preference for unwarranted patient intubation, and did not perform cardiopulmonary resuscitation for an excessively long time.
Generally, some researchers have described empathy as a cognitive attribute,7,8 which means it predominantly involves understanding another person's concerns. Others have described empathy as an emotional characteristic,9,10 which implies that it primarily involves feeling another person's pain and suffering. A third group views empathy as both cognitive and emotional.11,12 More detailed descriptions of those features are given elsewhere.5,13
Adding to the confusion is the fact that some researchers have proposed two types of empathy: “cognitive” and “emotional.”10,14 For example, Davis14 describes cognitive empathy as “attempts to entertain the perspective of others”(p17) and “the capacity for role taking.”(p29) He describes emotional empathy as “a tendency to react emotionally to the observed experiences of others.”14(p55) Others have also described emotional empathy in terms of vicarious empathy.15
To understand the operational definition of a concept, researchers must not only describe specific features of the concept but also take into consideration factors such as theoretical adequacy, linguistic consistency, logical distinction, and clinical relevance.16 On the basis of these factors and in consideration of the specific features of empathy and sympathy,5,13,17 we defined empathy in the context of patient care as a predominantly cognitive attribute that involves an understanding of patients' experiences, concerns, and perspectives combined with a capacity to communicate this understanding and an intention to help.5,13,17 We defined sympathy as a predominantly emotional attribute that involves feeling patients' pain and suffering, often emerging from an egoistic motivation to help the patient and thereby alleviate one's own personal distress.5,13 Our adapted definition of empathy is conceptually equivalent to Davis'14 definition of cognitive empathy, whereas our adapted definition of sympathy is more consistent with the idea of emotional14 or vicarious15 empathy.
To our knowledge, no psychometrically sound instruments are available for measuring empathic or sympathetic orientations in the context of the delivery of medical care. Thus, we designed this study to develop an instrument that measures empathic and sympathetic orientation in patient care and to provide evidence in support of their psychometrics.
Participants were 201 third-year students at Jefferson Medical College.
We used three research instruments in this study: the Jefferson Scale of Empathy (JSE), the Interpersonal Reactivity Index (IRI), and the Measure of Orientations Toward Empathic and Sympathetic Care.
1. The JSE. This is a psychometrically sound instrument developed specifically to measure empathy in the context of patient care. It contains 20 items that are answered on a seven-point Likert-like scale (7 = strongly agree; 1 = strongly disagree). Medical researchers in the United States have used this scale widely, as have researchers abroad, where it has been translated into 39 languages. Research has provided evidence in support of its construct validity,5,18,19 criterion-related validity,19,20 predictive validity,21 internal consistency reliability,18,19 and test–retest reliability.18 Three versions of the JSE exist: one for administration to medical students (S-Version), one for physicians and other health professionals (HP-Version), and one for health professions students other than medical students (HPS-Version). We used the S-Version in this study.13 A sample item (reverse scored) is “It is difficult for a physician to view things from a patient's perspective.” The possible range of scores is 20 to 140, and higher scores indicate a more empathic orientation.
2. The IRI. This instrument includes 28 items that measure both empathy (or cognitive empathy) and sympathy (or, as described by its author, emotional empathy).11,14 The IRI has been frequently used in social psychology and medical education research.14 The IRI contains the following four scales (each comprising seven items): (1) Perspective Taking, defined as adopting another's point of view (a sample item is “I sometimes try to understand my friends better by imagining how things look from their perspective”), (2) Empathic Concern, defined as experiencing sympathy (“I often have tender, concerned feelings for people less fortunate than me”), (3) Fantasy, defined as “the tendency to imaginatively transpose oneself into fictional situations”14(p57) (“When I am reading an interesting story or novel, I imagine how I would feel if the events in the story were happening to me”), and (4) Personal Distress, defined as self-oriented feelings of discomfort in response to distress in others (“I tend to lose control during an emergency”). Each item is answered on a five-point Likert-like scale (0 = Does not describe me well; 4 = Describes me very well).
The author of the index considers the Perspective Taking scale of the IRI to be the best among the four for measuring empathy (or, as he would describe it, cognitive empathy). He considers the other three scales to be more related to sympathy (or, in his terms, emotional empathy).11(p116) The possible range of scores in each scale is 7 to 28; similar to the JSE, higher scores indicate greater presence of the corresponding attribute.
3. The Measure of Orientations Toward Empathic and Sympathetic Care. We developed an instrument to measure students' empathic and sympathetic orientations toward patient care by assessing their responses to clinical vignettes. We adapted the idea from a procedure used by Nightingale and colleagues6 to examine patient outcomes of physicians' empathic and sympathetic orientation.
On the basis of a panel discussion, we developed several clinical vignettes. After a few iterations, we selected four vignettes involving a diabetic patient with complications, a rape victim, a patient who has undergone radical prostatectomy and developed complications, and a 16-year-old girl who is pregnant (Appendix 1).
After gaining the approval of the university's institutional review board (IRB), we distributed the three instruments to 258 third-year medical students at Jefferson Medical College in Philadelphia, Pennsylvania in 2010. We distributed the instruments during a class devoted in part to topics related to medical professionalism. We told the students that we designed the survey to better understand empathy and sympathy in patient care. We offered the students no incentive for participating in the study. Participation was voluntary and anonymous. We reminded students that individual data would be treated with strict confidentiality, that only aggregated group data would be reported, and that their responses would not become part of their academic records. The university IRB deemed the study exempt; thus, we did not need informed consent.
We instructed the students to review the four clinical vignettes and to assume that the patient in each of the scenarios was their own. We asked them to log their responses to each scenario on two given four-point, Likert-like scales (1 = very unlikely; 4 = very likely). One scale pertained to empathic responses for quantifying the student's inclination toward understanding the patient's pain and suffering. The other scale pertained to sympathetic involvement for quantifying the student's tendency to feel a patient's pain and suffering (Appendix 1). We calculated the sums of the Likert weights for the empathic and sympathetic responses separately, and we used these sums as measures of, respectively, empathic and sympathetic orientations. The possible range for each measure was 4 to 16, and higher scores indicated greater orientation toward the corresponding attribute. Thus, we obtained for each student an empathic orientation score and a sympathetic orientation score.
We used principal component factor extraction, followed by varimax rotation (to obtain independent factors and a simplified factor structure matrix) in order to examine the underlying construct of students' responses to the clinical vignettes. We noticed that the distributions of the empathic and sympathetic orientation scores were not normal. Because correlation coefficients cannot capture the essence of relationships when score distributions are not normal (i.e., in this study, multimodal), we used the method of contrasted groups instead of the correlation method which is often used in validity studies.22 For that purpose, on the basis of empathic orientation score distributions, we divided our sample into three groups of students with, respectively, high, moderate, and low empathic orientation. Likewise, on the basis of sympathetic orientation score distributions, we divided our sample into another three groups of students with, respectively, high, moderate, and low sympathetic orientation. This division allowed comparisons among the extremes.
We used two-way multivariate analysis of variance (MANOVA) to simultaneously examine group differences on both the JSE and IRI scores. The levels of empathic and sympathetic orientation were the two independent variables, and scores on the JSE and IRI were the dependent variables. When the multivariate test results were statistically significant, we conducted univariate analysis of variance. We used the Duncan multiple range test to detect the statistically significant pairwise group differences. We performed all statistical analyses with SAS (version 9.1 for Windows, Cary, North Carolina).
Of 258 total students in the class, 201 (78%) responded to all four vignettes and completed the JSE and IRI.
Construct validity, descriptive statistics, and reliability
Factor analysis of responses to clinical vignettes resulted in two reliable factors, each with an eigenvalue greater than 1 (indicating an important factor). Summary results are reported in Table 1.
As Table 1 shows, responses to the four vignettes that described students' inclination to share patients' feelings had the highest factor coefficients (accounting for 43% of the variance) under Factor 1, a factor we called sympathetic orientation construct. Responses related to the understanding of the patients' concerns showed the highest factor coefficients (accounting for 21% of the variance) under Factor 2, a factor we called empathic orientation construct.
The mean (± standard deviation [SD]) was 13.7 (±2.4) for empathic orientation and 9.2 (±3.2) for sympathetic orientation. The Cronbach coefficient alpha for the empathic orientation measure was 0.79; for sympathetic orientation, it was 0.84.
On the basis of the distribution of empathic orientation scores, we portioned 76 students into a high-empathic-orientation group, 62 into a moderate-empathic-orientation group, and 63 into a low-empathic-orientation group. Likewise, on the basis of the distribution of sympathetic orientation scores, we portioned 54 students into a high-sympathetic-orientation group, 94 into a moderate-sympathetic-orientation group, and 53 into a low-empathic-orientation group.
Table 2 reports the means and SDs of the total scores of the JSE and IRI, for the groups classified based on the levels of empathic and sympathetic orientations, and the summary results of our statistical analyses.
As Table 2 shows, statistically significant differences occurred on the JSE scores among the three groups who were classified based on their empathic orientation and those who were classified based on their sympathetic orientation; however, we observed no significant interaction effect between levels of empathic and sympathetic orientations. These and other findings (Table 2) indicate that scores of the JSE are significantly associated with empathic orientation scores, and total scores of the IRI (a composite measure of empathy and sympathy) were significantly associated with both empathic and sympathetic orientation scores.
We performed additional analyses to examine the relationships between empathic and sympathetic orientations and scores on the four scales of the IRI which specifically measure empathy (Perspective Taking) and sympathy (Empathic Concern, Fantasy, and Personal Distress). We report the summary results of our statistical analyses in Table 3. As Table 3 shows, the results of MANOVA indicate that no significant differences in the scores on the four scales of the IRI occurred in the high-, moderate-, and low-empathic-orientation groups; however, significant differences did occur among the high-, moderate-, and low-sympathetic-orientation groups.
These and other findings (Table 3) suggest that sympathetic orientation is significantly associated with scores of the three scales of the IRI which measure sympathy (emotional empathy).11,14
The operational definition of a concept is an integral part of any research that serves as a building block for either the formulation of a theory23 or the evaluations of pertinent outcomes. We adapted operational definitions for empathy and sympathy in this study. Then, according to key features of those definitions, we developed pertinent measures derived from responses to clinical vignettes to quantify empathic and sympathetic orientations in the context of patient care. Further, we examined aspects of the psychometrics of the measures of empathic and sympathetic orientations.
The results of the factor analysis clearly indicate that the two types of responses (understanding and feeling) represented two different constructs of empathic and sympathetic orientations, thus providing support for the construct validity of the two measures. The magnitudes of the reliability coefficient alpha for each of the two measures indicate that each is internally consistent.
Our findings that the empathic orientation is significantly associated with scores of a validated, conceptually relevant measure of empathy (as measured by the JSE) provide support for the criterion-related validity of the empathic orientation measure (convergent validity). Also, our findings that sympathetic orientation was not significantly associated with scores of the JSE (discriminant validity), but was significantly associated with the conceptually relevant measures of sympathy (as measured by the Empathic Concern, Fantasy, and Personal Distress scales of the IRI), provide support for the criterion-related validity of the sympathetic orientation measure. Notably, the total scores of the IRI (a mixed measure of empathy and sympathy) were significantly associated with both empathic and sympathetic orientation. Therefore, these findings provide evidence in support of the psychometrics of the two measures of empathic and sympathetic orientations.
The findings have important implications for both medical education and health care research. For example, others have speculated that, in the context of patient care, empathy (as defined in this study) almost always leads to positive clinical outcomes, whereas sympathy in excess can be detrimental to objectivity in clinical decision making.5 Others believe that cognitively defined empathy often leads to personal growth, career satisfaction, and optimal clinical outcomes, whereas emotionally defined sympathy can lead to career burnout, compassion fatigue,24 exhaustion, and vicarious traumatization.1 These speculations require empirical verification using such measures as those developed in this study.
We can assume that, up to a certain point, the relationship between empathy and positive clinical outcomes is linear; that is, the outcomes progressively become better as a function of an increase in empathy. We can also assume that the relationship between sympathy and clinical outcomes resembles an inverted U shape (similar to that between anxiety and performance); that is, sympathy—to a limited extent—is beneficial, but excessive sympathy is detrimental. The availability of a validated instrument for measuring empathic and sympathetic orientations toward patient care provides an opportunity for empirical research to verify these two assumptions and to examine other issues in medical education and patient care (e.g., relationship between empathy, sympathy, burnout,25 and well-being26).
This study is limited by the fact that, as a result of its accessible sampling design, it included only third-year medical students at a single institution, rather than a more heterogeneous population of medical students and practicing physicians. These limitations may jeopardize the generalizability of the findings. Other investigators should replicate our research in multiple institutions and/or with more representative samples of students and physicians. Although we reported the internal consistency of the measures of empathic and sympathetic orientation, examining their stability over time will also be important. Despite these limitations, the instrument we developed and tested in this study has the potential for use in empirical investigations of factors that contribute to physicians' competence, quality of care, and patient outcomes.
The authors would like to thank Dorissa Bolinski for her editorial help.
This study was approved by the institutional review board of Thomas Jefferson University.
1 Linley PA, Joseph S. Therapy work and therapists' positive and negative well-being. J Soc Clin Psychol. 2007;26:385–403.
2 Shanafelt TD, West C, Zhao X, et al. Relationship between increased personal well-being and enhanced empathy among internal medicine residents. J Gen Intern Med. 2005;20:559–564.
3 West CP, Huschka MM, Novotny PJ, et al. Association of perceived medical errors with resident distress and empathy. JAMA. 2006;296:1071–1078.
5 Hojat M. Empathy in Patient Care: Antecedents, Development, Measurement, and Outcomes. New York, NY: Springer; 2007.
6 Nightingale SD, Yarnold PR, Greenberg MS. Sympathy, empathy, and physician resource utilization. J Gen Intern Med. 1991;6:420–423.
7 Kohut H. The Analysis of the Self: A Systematic Approach to the Psychoanalytic Treatment of Narcissistic Personality Disorders. New York, NY: International Universities Press; 1971.
8 Basch MF. Empathic understanding: A review of the concept and some theoretical considerations. J Am Psychoanal Assoc. 1983;31:101–126.
9 Eisenberg N. Empathy and Related Emotional Response. San Francisco, Calif: Jossey-Bass; 1989.
10 Hoffman ML. The development of empathy. In: Rushton JP, Sorrentino RM, eds. Altruism and Helping Behavior: Social Personality Developmental Perspectives. Hillsdale, NJ: L. Erlbaum Associates; 1981:41–61.
11 Davis MH. Measuring individual differences in empathy: Evidence for a multidimensional approach. J Pers Soc Psychol. 1983;44:113–126.
12 Hodges SD, Wegner DM. Automatic and controlled empathy. In: Ickes WJ, ed. Empathic Accuracy. New York, NY: Guilford Press; 1997:311–339.
14 Davis MH. Empathy: A Social Psychological Approach. Boulder, Colo: Westview Press; 1996.
15 Mehrabian A, Epstein N. A measure of emotional empathy. J Pers. 1972;40:525–543.
16 Morse JM, Mitcham C. Compathy: The contagion of physical distress. J Adv Nurs. 1997;26:649–657.
17 Hojat M. Ten approaches for enhancing empathy in health and human services cultures. J Health Hum Serv Adm. 2009;31:412–450.
18 Hojat M, Gonnella JS, Nasca TJ, Mangione S, Vergare M, Magee M. Physician empathy: Definition, components, measurement and relationship to gender and specialty. Am J Psychiatry. 2002;159:1563–1569.
19 Hojat M, Mangione S, Nasca TJ, et al. The Jefferson Scale of Physician Empathy: Development and preliminary psychometric data. Educ Psychol Meas. 2001;61:349–365.
20 Hojat M, Gonnella JS, Mangione S, et al. Empathy in medical students as related to academic performance, clinical competence, and gender. Med Educ. 2002;36:522–527.
21 Hojat M, Mangione S, Nasca TJ, Gonnella JS, Magee M. Empathy scores in medical school and ratings of empathic behavior 3 years later. J Soc Psychol. 2005;14:663–672.
22 Anastasi A. Psychological Testing. 4th ed. New York, NY: Macmillan; 1976.
23 Morse JM, Mitcham C, Hupcey JE, Tason MC. Criteria for concept evaluation. J Adv Nurs. 1996;24:385–390.
24 Figley CR. Compassion Fatigue: Coping With Secondary Traumatic Stress Disorder in Those Who Treat the Traumatized. New York, NY: Brunner/Mazel; 1995.
26 Thomas MR, Dyrbye LN, Huntington JL, et al. How does distress and well-being relate to medical student empathy? A multicenter study. J Gen Intern Med. 2007;22:177–183.