Empathy is a critically important characteristic for physicians to possess. It is fundamental to the physician–patient relationship and likely has an effect on clinical outcomes.1 Although empathy can be defined in many ways, it is typically viewed as a multidimensional construct that encompasses both cognitive and affective components.2 The former is concerned with one's ability to perceive another's point of view and be aware of one's effect on others.3 The affective component is concerned with one's vicarious emotional responses to the perceived emotional experiences of others.4
Two recent studies have examined how empathy changes during a year of postgraduate training in internal medicine. In 2002, we reported a significant decrease in multiple domains of empathy over the course of an internal medicine internship at a university-based training program using the Interpersonal Reactivity Index.5 In contrast, a study of 98 internal medicine residents at a university-based training program using the Jefferson Scale of Physician Empathy revealed empathy was stable over the internship and that there were no significant differences among residents at different training levels.6 Similarly, a study with third-year nursing students demonstrated the stable nature of empathy; although the study's baseline empathy scores were quite high.2 To date, we could find no longitudinal reports of how the specific aspects of empathy vary over the course of residency training.
In addition to empathy, several other indicators of mood and temperament have also been studied with physicians-in-training, because of their importance for the trainee as well as implications for the physician–patient interaction. For example, many interns feel anxiety and depression at some point over the course of the year.7–11 Early studies of the rates of depression during residency were retrospective and revealed that depression developed in 30% of interns.12,13 A 1985, year-long prospective study of depression symptoms in medical houseofficers using the Center for Epidemiological Studies-Depression Scale demonstrated depressive responses in 28.7%, 21.5%, and 10.3% of first-, second-, and third-year residents, respectively.14 Later studies using the Beck Depression Inventory demonstrated between 7% and 8.6% of interns experienced depressive symptoms.7,8,10 We previously reported seasonal depressive symptoms as measured by the Profile of Mood States, with incidence of depression reaching 25% among interns in February and dropping to 19% by the end of the year.5 A prospective investigation of emotions and attitudes of 22 interns also revealed the peak incidence of depression, fatigue, and sense of defeat to be in the eighth month of internship.15
Although a fair amount of empirical work has been directed towards studies of empathy, depression, and other mood states during the internship year, the trajectory of the changes over the three-year residency has not been reported. Accordingly, this study examined the variation of empathy and mood disturbances over the course of internal medicine residency training in a cohort of residents for whom extensive data collected during the internship year was previously published.5 To our knowledge, this is the first study to examine the longitudinal changes in empathy over a three-year training period. We expected that, as residents progressed through residency, empathy would increase and mood would improve but neither would return to their preinternship baselines.
All 61 interns in the 2000 entering class of the Internal Medicine Residency Program at the Hospital of the University of Pennsylvania were enrolled in the study at the department's orientation session for interns. Interns were told that participation was voluntary and that only group data would be reported. The university's institutional review board approved the study.
Data were collected at six points in time:
▪ Time 1 (baseline) occurred during intern orientation in June 2000
▪ Time 2 in November 2000
▪ Time 3 in February 2001
▪ Time 4 in June 2001 (the end of internship)
▪ Time 5 in June 2002 (the end of the second year), and
▪ Time 6 in June 2003 (the end of the third year).
At Times 1, 2, 3, 4, and 6 participants completed the Profile of Mood States (POMS).16 At Times 1, 4, 5, and 6 they completed the Interpersonal Reactivity Index (IRI).17 The results of Times 1–4 have been published previously.5 For all administrations of the IRI, participants wrote their names on tear-off cover sheets. An administrative assistant coded identification numbers on the IRI so that subscale score relationships could be examined over time. Identification was not recorded on the POMS to preserve anonymity.
The POMS16 is a 65-item instrument that measures six mood states on a five-level adjective scale (not at all, a little, moderately, quite a bit, extremely). Five of the mood states are seen as negative: Tension–Anxiety, Depression–Dejection, Anger–Hostility, Fatigue–Inertia, and Confusion–Bewilderment. One state is considered positive: Vigor–Activity. The rationale for the use of POMS was reviewed previously.5
The IRI is a 28-item instrument consisting of four seven-item subscales, each tapping some aspect of the global concept of empathy.17 Two subscales were of primary interest; Perspective Taking and Empathetic Concern. Perspective Taking assesses the tendency to spontaneously adopt the psychological viewpoint of others, and Empathetic Concern assesses “other-oriented” feelings of sympathy and concerns for unfortunate others. The remaining subscales, Personal Distress and Fantasy, measure “self-oriented” feelings of anxiety and unease in tense interpersonal settings and the respondents’ tendencies to transpose themselves imaginatively into the feelings and actions of fictitious characters in books, movies, and plays. The IRI's subscales have shown predictable relationships among themselves as well as the related constructs of social functioning, self-esteem, emotionality, and sensitivity to others.17
Our analyses focused on two issues: (1) changes in scores for the POMS and IRI subscales over the three-year residency, and (2) distributions of scores at the end of residency compared with available adult and college students’ norms. Scores for the multiple administrations of the POMS were compared using analysis of variance (ANOVA). Because we did not identify interns, scores were treated as independent samples, thus providing a conservative estimate of the significance of differences among means. All available data points were used. Post hoc comparisons were made with the Duncan test and effect sizes were calculated comparing the final (Time 6) scores with the worst scores (typically middle-to-end of the internship) and the final scores to baseline scores. We analyzed scores for the multiple administrations of the IRI with a repeated-measures ANOVA with planned comparisons of means from each occasion, using data only for the 37 residents who completed all assessments. Means and standard deviations at the end of residency are presented for each subscale and compared with available data for adult and college students’ samples. Norms based on responses of other professional populations would have been preferable, but these data were not available. Also, the norms are typically published separately for each gender. Because we did not collect gender identifiers in the current study, the comparison data for males and females were weighted and combined. We compared observed scores for the interns with population norms using a t test.
The demographics of the interns at Time 1 have been previously reported.5 Of 61 interns, 45 were in the categorical program, eight in the primary care program, and 14 in the preliminary program. At Time 5 (July 2002), 47 of the residents completed their second year. From this group, four left the program to enter the American Board of Internal Medicine-sponsored research pathway. One month after Time 6 (June 2003), 41 of the residents finished their training. Two residents graduated in December 2003, due to maternity leave.
At Time 1, 60 (98%) of the interns completed the POMS and IRI. At Times 2 and 3, 44 (72%) and 48 (79%), respectively, completed the POMS. At Time 4, 48 (79%) interns completed the POMS and IRI. At Time 5, 44 (94%) residents completed the IRI, and at Time 6, 39 (95%) completed the IRI and POMS.
As shown in Table 1, four of the six POMS subscales changed significantly over time: Depression–Dejection (p = .0031), Anger–Hostility (p < .0001), Vigor–Activity (p < .0001), and Fatigue–Inertia (p < .0001). In each instance, Time 1 scores were significantly different than were the scores gathered at other times, especially Time 2 and Time 3. Levels of Depression–Dejection, Anger–Hostility, and Fatigue–Inertia started out low, peaked in midwinter of the first year, and improved by the end of internship. Additional improvement was minimal over the last two years of residency. Importantly, the residents’ Depression–Dejection scores in the second and third years improved to the point where they were not different from baseline. However, the changes for Anger–Hostility and Fatigue–Inertia—although improving—remained significantly different from baseline values. The pattern for Vigor–Activity showed the opposite trend —high (i.e., good) scores at the beginning of the year that decreased over the midyear and began to rise at the end of the year with little change over the last two years. In general, effect sizes comparing the final score with baseline scores were moderate, ranging in the .40s to .60s, reflecting sizable decrements in scores. When compared with the point at which scores were the worst, gains were consistent albeit small, with effect sizes generally in the −.20s.
Table 1 also shows significant changes over time in two of the IRI's subscales among the subsample who stayed in the program for all three years. For Empathic Concern, the worsening in scores that occurred during the internship remained low throughout residency. For Personal Distress, the peak of distress at the end of internship declined and Time 5 and Time 6 scores returned to baseline, less distressed levels. The effect sizes summarizing the magnitude of these changes were moderate (.41 and .51, respectively).
Table 2 shows Time 6 scores compared with those of national norms. (Similar comparisons to Time 1 scores were published earlier.5) At this phase of their careers, the graduating residents remained less tense, depressed, and confused than the norms. There was no difference in Vigor–Activity or Anger–Hostility. They also had less personal distress.
Internal medicine residency presents challenges in the form of long hours, acute and chronic partial sleep loss, and significant personal compromises.11,18 It makes sense that these conditions are related to a high incidence of depression and anxiety among interns and residents.7,15 Our initial finding that interns’ depressive symptoms peaked in the eighth month of training confirms earlier work.15 The sizable standard deviations at all observation points also confirms Reubens's work suggesting some residents exhibit depressive symptoms in all years of training.14 However, our current study suggests that residents’ depression improves after the midinternship peak. Even by the end of the first year, depression decreased and by the end of residency scores did not differ from baseline. As medical educators, it is comforting to know that one type of mood disturbance (i.e., depression) improves over the course of residency training. Personal distress scores also returned to baseline by the end of residency training. Given that this subscale is a measure of one's ability to remain calm in a crisis, its improvement indicates increasing confidence in facing emergencies: an expected and desired outcome of residency.
On the other hand, residents are completing the program sustaining the low levels of vigor and high levels of anger and fatigue that first became apparent in their internship year. They are no longer different from population norms. The vigor and fatigue issues are certainly explainable. Residency has always required long and intense work hours. It will be interesting to consider the impact of the new duty-hour standards from Accreditation Council of Graduate Medical Education on mood disturbances.19 Our immediate hypothesis is that with shorter and limited work hours, residents might exhibit less fatigue and more vigor. However, as a way to meet the new duty-hour standards, many programs have increased the amount of night work for residents.20,21 Unfortunately a recent study suggested significant disturbances of sleep and mood during night float rotations when compared with daytime rotations.22 In light of our work, this is very concerning and needs to be carefully monitored by future research. Regardless, graduating residents seem to be in better emotional shape than the general population.
This study has several limitations. It was conducted at one university-based internal medicine residency program. Programs with different structures may attract different types of residents and have different results. We did not administer the POMS at Time 5 (the end of the second year), so we might have missed peaks in the different subscales. We think the trends for this time can be inferred based on the scores at the end of residency. Most importantly, the norms for these instruments are based on responses from college-age students. Such students are at least five years younger than our cohort and represent a cross section of all possible career tracks, not the self-selected group that is pursuing internal medicine. We would like to think that the latter group has greater emotional health given their maturity and career choice. However, the lack of age-appropriate norms among graduate students limits the comparisons we can make.
The good news from this study is that the residents’ personal distress that peaked in internship declined toward baseline levels. Similar to third-year nursing students,2 interns arrive empathetic with little personal distress—they score quite high in the domains of Perspective Taking and Empathic Concern, two behaviors particularly important in medicine. Things get worse for them during the internship year.5 A similar trend is not apparent among third-year nursing students, but the results here show clear recovery during the latter part of residency. This is countered by the fact that the lowered Empathic Concern remains low throughout residency. A similar trend was observed for Perspective Taking. Our findings contrast with a cross-sectional study of 98 interns that suggested empathy is a relatively stable trait not easily changed in residency.6 Empathy is a critical trait for physicians. As medical educators, we are responsible for nurturing the trait that is so coveted by patients. We have shown that the early declines persist. Among the many challenges ahead for medical education, protecting the development of empathy during times of immense change will certainly be one of them.
1 Hojat M, Gonnella JS, Nasca TJ, et al. Physician empathy: definition, components, measurement, and relationship to gender and specialty. Am J Psychiatry. 2002;159:1563–9.
2 Becker H, Sands D. The relationship of empathy to clinical experience among male and female nursing students. J Nurs Educ. 1988;27:198–203.
3 Hogan R. Development of an empathy scale. J Consult Clin Psychol. 1969;33:307–16.
4 Mehrabian A, Epstein N. A measure of emotional empathy. J Pers. 1972;40:525–43.
5 Bellini LM, Baime M, Shea JA. Variation of mood and empathy during internship. JAMA. 2002;287:3143–6.
6 Mangione S, Kane GC, Caruso JW, et al. Assessment of empathy in different years of internal medicine training. Med Teach. 2002;24:370–3.
7 Kirsling RA, Kochar MS, Chan CH. An evaluation of mood states among first year residents. Psychol Rep. 1989;65:355–66.
8 Clark DC, Salazar-Grueso E, Grabler P, Fawcett J. Predictors of depression during the first 6 months on internship. Am J Psychiatry. 1984;141:1095–8.
9 Hainer BL, Palesch Y. Symptoms of depression in residents: a South Carolina Family Practice Research Consortium Study. Acad Med. 1998;73:1305–10.
10 Godenick MT, Musham C, Palesch Y, Hainer BL, Michels PJ. Physical and psychological health of family practice residents. Fam Med. 1995;27:646–51.
11 Committee of Interns and Residents (US). Petition requesting a limit on medical resident work hours through a standard issued under the authority of the occupational health and safety administration. Public Citizen, 2001.
12 Valko RJ, Clayton PJ. Depression in the internship. Dis Nerv Syst
13 Ford CV (ed). Emotional distress in resident and interns. Proceedings from the annual meeting of the American Psychiatric Association, New Orleans, LA, May 15, 1981.
14 Reuben DB. Depressive symptoms in medical house officers: effects of level of training and work rotation. Arch Intern Med
15 Girard DE, Elliot DL, Hickman DH, et al. The internship: a prospective investigation of emotions and attitudes. West J Med. 1986;144:93–8.
16 McNair DM, Lorr M, Droppelman LF. Profile of Mood States Manual. San Diego: Edits, 1992.
17 Davis M. Measuring individual differences in empathy: evidence for a multidimensional approach. J Pers Soc Psychol. 1983;44:113–26.
18 Shapiro SL, Shapiro DE, Schwartz GE. Stress management in medical education: a review of the literature. Acad Med. 2000;75:748–59.
19 Accreditation Council of Graduate Medical Education Statement on Duty Hours 〈http://www.acgme.org
〉.Accessed 15 October 2003. ACGME, Washington DC, 2003.
20 Morelock JA, Stern DT. Shifting patients: how residency programs respond to residency review committee requirements. Am J Med
21 University HealthSystem Consortium. Resident work hours benchmarking project summary. Oak Brook, IL: University HealthSystem Consortium, 2003.
22 Cavallo A, Jaskiewicz J, Ris MD. Impact of night float rotation on sleep, mood and alertness: the resident's perception. Cronobiol Int. 2002;19:893–902.