Trends of Change in Empathy Among Indian Medical Students: A Two-Year Follow-Up Study

Abstract

Background:

The development of an empathic approach is essential for doctor–patient relationships. Medical training is a challenging time that may affect empathy. This study aimed to assess the change in empathy in students during medical education.

Methods:

One hundred and fifty MBBS students were recruited at admission and assessed for empathy, interpersonal reactivity, and general health. They were followed for two years and assessed at three intervals.

Results:

A significant decline was seen in empathy for both male and female students. The decline was correlated with psychological stress. Gender, family structure, having siblings, and increasing General Health Questionnaire score predicted change in empathy.

Conclusion:

Empathy declines with advancing training, varying with constitutional and situational factors. The medical curriculum should include skills like empathic communication as well.

Key Messages:

• Development of an empathic approach starts during early medical education.

• Empathy shows a steady decline with advancing training.

• Empathy shows variation with both constitutional and situational factors.

• Changes in the medical curriculum and techniques of imparting training are needed to include empathic communication and personal stress-management training as part of medical education.

Developing empathic and meaningful interpersonal relationships between doctors and patients is of utmost importance to desirable clinical outcomes. ¹ Empathy is the ability to understand and share others’ feelings and emotions and convey this understanding during interactions. ² Empathy is widely viewed as a relatively stable, constitutional trait. In accordance, correlations of empathy with gender and personality traits, including sociability and agreeableness, have been studied. ³ An alternative view of empathy, that of a mutable state, also exists, in which cognitive and affective responses are considered sensitive to circumstances. ² Accordingly, life experiences and situational factors, such as quality of life along with one’s subjective perception of well-being and sense of personal accomplishment, have been reported to affect empathy. ⁴ However, no consensus exists about empathy being a state or a trait. ⁵

Medical institutions and professional bodies advocate a balance between clinical detachment and overinvolvement, defining empathy as correctly acknowledging the emotional state of another without experiencing that state oneself. ³ Importantly, the development of such an approach begins well before the commencement of professional practice, during undergraduate medical education, when empathy levels may begin to decrease. This may affect the consolidation of professional identity in the future. ⁶

Medical education is a period of intimate contact with disease, morbidity, and mortality. It often begins at a young age when students have had little personal experience with such profound issues and limited opportunities to reflect on them. ⁷ The Indian undergraduate medical education program consists of three phases: the first preclinical year, comprising the teaching of basic sciences; and clinical years (II to IV) where clinical exposure is provided under supervision and paraclinical and clinical subjects are taught, culminating in a year-long rotatory internship. Each of these situations is unique in terms of exposure, opportunities, and challenges, potentially shaping life experiences, psychological makeup, and fundamentals for future practice. ⁸

Studies and systematic reviews on empathy in medical students have reported mixed findings. Some studies report a significant decline in empathy,^{6, 9, 10} whereas others report no significant change ¹¹ or an increase with advancing medical training. ¹² Most of these studies are cross-sectional and ignore the effect of time spent in medical training and the various stressors associated with it. This study aimed to assess the changes that occur in empathy as students advance in their medical education and explore possible constitutional and situational correlates for the same.

Methods

This prospective observational study was conducted on a batch of 150 students who had recently joined MBBS (batch of 2017) at an urban private medical college in Western India. After approval was received from the Institutional Ethics Committee, both male and female students were recruited, excluding those who had a preexisting psychiatric illness. The students who developed a psychiatric illness or left the course during the study were dropped from the final analysis.

The students were approached before the commencement of the day’s academic schedule. They were provided with a detailed information sheet, and informed consent was obtained. Students less than 18 years of age were provided with assent forms, and parental consent was received by sending the forms home through the students.

After providing their sociodemographic details in the preliminary interview, students were required to answer a questionnaire that consisted of the following.

1. Interpersonal Reactivity Index (IRI): it consists of 28 items answered on a 5-point Likert scale and measures cognitive aspects of empathy, with higher scores indicating greater cognitive empathy. It has been validated for use among college students and young adults in India as well as European and Middle Eastern countries. ¹¹

2. Toronto Empathic Questionnaire (TEQ): it is a 16-item questionnaire with responses on a 5-point Likert scale. It has been widely used to measure emotional empathy among medical students in various countries, including India, China, Turkey, Italy, and the United States of America.^{12, 13} Higher scores indicate greater emotional empathy.

3. Standardized Assessment of Personality Abbreviated Scale (SAPAS): it is an 8-item yes/no response questionnaire for the screening of personality disorders. It has been used in several countries, including India. A score of 3 on the screening interview indicates further assessment of personality. The sensitivity and specificity are 0.94 and 0.85, respectively. ¹⁴

4. General Health Questionnaire-28 (GHQ-28): it is a 28-item screening tool for identifying psychological disturbances among adolescents and young adults. It has been validated for the Indian population, and higher scores indicate poorer psychological well- being.^{15, 16}

The questionnaire was administered to the students in their respective lecture theatres. The students were given brief verbal instructions on how to respond to the questions and were free to ask for clarification if required. They were also instructed to answer all questions. Completion was also ensured for each student by manual checking at the time of submission of the questionnaires back to the investigating team. Absent students were approached in their classes on the following days until all consenting students had responded to the questionnaire.

This exercise was conducted in person once at the time of joining the course in August 2017 (MBBS I year, semester-1), repeated at the beginning of the second year in August 2018 (MBBS II year, semester-3), and again at the end of six months of clinical training in February 2019 (MBBS II-year, semester-4).

Statistical Analysis

The responses were analyzed using the IBM Statistical Package for the Social Sciences (SPSS) Statistics for Windows, version 24.0 (IBM Corp., Armonk, NY, USA). Scores on different questionnaires were represented as mean (SD), and differences at each assessment were analyzed using a repeated-measures ANOVA. Differences between the means for male and female participants were analyzed using the independent sample t-test. Changes in personality screening scores were assessed using the chi-square test. Correlations between dependent and independent variables were drawn using Pearson’s correlation coefficient. A P-value of less than 0.05 was considered significant. Multiple linear regression was performed with change in IRI and TEQ scores as outcome variables, with gender, family structure, having siblings, and change in GHQ-28 scores as predictor variables.

Results

Out of 150 students included at the commencement of the study, 2 students discontinued their medical training for unknown reasons. They were dropped from the study, and a total of 148 students were considered for the final analysis, including 71 males and 77 females. Ages at admission ranged from 17 to 22 years, with a mean age of 18.40 (SD = 0.98) years. Of the 148 students, 111 belonged to nuclear families and 37 to joint families. Also, 116 students had one sibling, 14 had two siblings, and 17 had no siblings. The mean scores of the IRI, TEQ, and GHQ are presented in Table 1. Personality screening was done in each semester using the SAPAS, where the above cut-off scores were seen in four students at baseline and five and seven at first and second follow-ups, respectively. Personality showed no significant change (chi-square statistic = 0.907, p-value = 0.65) over the duration of the assessment. Both mean IRI and mean TEQ scores were seen to decrease in each follow-up. On the other hand, GHQ-28 scores were seen to increase in each follow-up. The decrease in scores on IRI and TEQ and the increase in scores of GHQ-28 in each follow-up were seen to be statistically significant using repeated-measures ANOVA (P < 0.001 each), indicating a decrease in both cognitive and emotional empathy as well as deteriorating psychological health with advancing medical training.

Table 1.

Differences in Mean IRI, TEQ, and GHQ-28 Based on Gender at Baseline and Each Follow-Up

–	IRI Scores [Mean (SD)]			t (df = 146); P-Value
	Total Score	Males	Females
Baseline	55.68 (17.39)	40.22 (6.90)	69.93 (10.61)	20.07**; <0.001
Follow-up 1	50.95 (18.43)	34.31 (7.89)	66.31 (10.17)	21.34**; <0.001
Follow-up 2	46.31 (19.38)	29.32 (9.00)	61.97 (11.57)	19.03**; <0.001
RM-ANOVA (df = 2)	9.57**	33.22**	10.15**	–
–	TEQ Scores [Mean (SD)]			t (df = 146); P-Value
	Total Score	Males	Females
Baseline	39.28 (15.65)	24.40 (4.55)	53.00 (7.57)	27.35**; <0.001
Follow-up 1	35.36 (16.78)	19.31 (5.03)	50.16 (7.64)	28.73**; <0.001
Follow-up 2	31.22 (19.38)	15.43 (5.54)	45.76 (7.64)	24.03**; <0.001
RM-ANOVA (df = 2)	7.99**	56.20**	21.00**	–
–	GHQ-28 Scores [Mean (SD)]			t (df = 146); P-Value
	Total Score	Males	Females
Baseline	18.45 (6.01)	19.46 (6.89)	17.52 (4.94)	1.93*; 0.04
Follow-up 1	24.94 (6.65)	26.86 (6.88)	23.18 (5.96)	3.48**; <0.001
Follow-up 2	35.40 (8.41)	36.18 (8.34)	34.69 (7.95)	0.266; 0.20
RM-ANOVA (df = 2)	211.97**	90.96**	143.62**	–

IRI, Interpersonal Reactivity Index; TEQ, Toronto Empathic Questionnaire; GHQ-28, General Health Questionnaire-28; df, degrees of freedom; SD, standard deviation.

* indicates P < 0.05; ** indicates P < 0.001.

Exploratory analyses of the relationship of various sociodemographic variables were done with baseline and follow-up scores of both IRI and TEQ. It was observed that females had significantly higher mean scores on IRI and TEQ at baseline and both follow-ups (P < 0.001 each, Table 1). Females had significantly lower GHQ-28 scores at baseline and first follow-up. However, at the second follow-up, this difference was not statistically significant. At comparable scores on GHQ-28, females had significantly higher scores on IRI and TEQ than males (Table 1). During the follow-up assessments, mean decreases in IRI and TEQ scores were significantly higher in males (IRI: 10.90 ± 7.98 and 8.97 ± 5.23 and TEQ: 7.96 ± 6.29 and 7.22 ± 6.65, Table 2) as compared to females.

Table 2.

Associations of Changes in IRI and TEQ Scores with Gender, Family Structure, and Siblings

Parameter	Change in IRI Score Mean (SD)	Association with Change in IRI Score t (df = 146); P-Value	Change in TEQ Score Mean (SD)	Association with Change in TEQ Score t (df = 146); P-Value
Gender Male (n = 71, 47.97%) Female (n = 77, 52.02%)	10.90 (7.89) 7.96 (6.29)	2.49*; 0.01	8.97 (5.23) 7.22 (6.65)	1.46*; 0.04
Family structure Nuclear (n = 111, 75%) Joint (n = 37, 25%)	11.21 (6.50) 3.84 (4.73)	5.89**; <0.001	9.47 (5.81) 3.84 (4.73)	5.33**; <0.001
Siblings Absent (n = 17, 11.48%) Present (n = 131, 88.51%)	11.13 (5.03) 8.01 (3.35)	2.48*; 0.01	9.75 (4.20) 4.88 (3.08)	4.87*; 0.002

IRI, Interpersonal Reactivity Index; TEQ, Toronto Empathic Questionnaire; GHQ-28, General Health Questionnaire-28; df, degrees of freedom; SD, standard deviation.

* indicates P < 0.05, ** indicates P < 0.001.

There were 111 students from nuclear families and 37 from joint families. Students from nuclear families had no difference in mean IRI [t (df) = 0.687 (146); P=0.5] and TEQ [t (df) = 0.733 (146); P = 0.47] scores as compared to those from joint families. However, students from nuclear families had a significantly greater mean decrease in IRI (P < 0.001) and TEQ (P < 0.001) scores as compared to those from joint families (Table 2).

There were no differences in mean IRI and TEQ scores at baseline among adolescents who had siblings compared to those who did not. However, the change in IRI and TEQ scores was significantly lower in the students who had siblings than in those who did not, indicating greater stability in IRI and TEQ scores (Table 2).

Four students had above cut-off scores in SAPAS screening. These students had lower mean IRI and TEQ scores at baseline as compared to those who scored below the cut-off. However, the difference was not statistically significant.

Age at joining MBBS had a weak but significant negative correlation with IRI (r = −0.351, P < 0.001) and TEQ (r = -0.288, P < 0.001) scores, indicating that IRI and TEQ scores decreased with increasing age at joining the academic training. However, the decrease in IRI and TEQ scores had no significant correlation with age at joining MBBS (P= 0.13 and 0.69, respectively).

GHQ-28 scores showed a significant and strong positive correlation with a decrease in IRI (r = 0.814; P < 0.001) and TEQ scores (r = 0.645; P < 0.001), indicating a greater decrease in IRI and TEQ scores with a worsening of psychological health. Thus, both IRI and TEQ scores showed variations with gender, family structure, presence or absence of siblings, and changes in GHQ-28 scores.

After the preliminary assumptions were met, multiple linear regression was performed to predict change in IRI and TEQ scores, using change in GHQ-28 scores, gender, family structure, and presence or absence of siblings as independent variables. Change in IRI scores had an R of 0.859, indicating a good level of prediction and coefficient of determination, and R² indicated that GHQ-28 scores, gender, family structure, and the presence or absence of siblings could predict 73.1% variability in IRI scores. F and P values indicated an extremely good fit for the data. Multiple linear regression predicting change in TEQ scores using change in GHQ-28 score, gender, family structure, and the presence or absence of siblings was also done. It had an R of 0.705, indicating a good level of prediction and coefficient of determination. R² indicated that GHQ-28 scores, gender, family structure, and the presence or absence of siblings could predict 48.3% variability in IRI scores. F and P values indicated an average fit for the data. Both of these models are shown in Table 3. It was seen that individually, gender, family structure, and change in GHQ-28 score were significant predictors of change in IRI scores, whereas gender and change in GHQ-28 score were significant predictors of change in TEQ scores.

Table 3.

Multiple Linear Regression Predicting IRI and TEQ Scores Based on Gender, Family Structure, Siblings, and Change in GHQ-28

Outcome Variables	Predictor Variables	Unstandardized Coefficient (Standard Error)	Standardized Coefficient	95% Confidence Interval	t	P-Value
Change in IRI scores	Constant Gender (F = 1, M = 0) Family structure (joint = 1, nuclear = 0) Siblings (present = 1, absent = 0) Change in GHQ-28 score	7.64 (3.53) 3.98 (2.07) 0.14 (1.16) 0.40 (0.98) 2.47 (1.67)	6.58 2.01 1.12 4.71	1.09–8.13 0.09–2.17 0.08–1.39 1.97–5.32	2.16* 1.91** 1.69* 0.32 1.41*	0.03 <0.001 0.004 0.13 0.03
Change in TEQ scores	Constant Gender (F = 1, M = 0) Family structure (joint = 1, nuclear = 0) Siblings (present = 1, absent = 0) Change in GHQ-28 score	6.46 (4.29) 3.17 (1.95) 1.11 (0.98) 0.74 (0.36) 1.98 (0.57)	6.41 2.83 2.06 4.37	2.14–6.87 1.77–3.15 0.31–2.19 1.52–4.77	1.99* 1.33* 0.89 0.67 0.32*	0.004 0.01 0.09 0.07 0.003

Model summary:

1. Change in IRI scores: R² = 0.731, F (4,143) = 100.668, P < 0.001

2. Change in TEQ scores: R² = 0.483, F (4,143) = 35.401, P < 0.001

IRI, Interpersonal Reactivity Index; TEQ, Toronto Empathic Questionnaire; GHQ-28, General Health Questionnaire-28; F, female; M, male.

* indicates P < 0.05, ** indicates P < 0.001.

Discussion

This longitudinal study assessed changes in empathy among undergraduate medical students. With the intent to assess both the emotional and cognitive components of empathy in detail, we used two different scales, the IRI and the TEQ, both of which have been used in studies on the Indian student population before. Our findings indicate a significant decline in both cognitive and emotional empathy as students advance in their medical training. This is consistent with earlier studies using similar instruments, some of which reported a decline in empathy during medical training, whereas others reported no such change and no added advantage of a behavioral science course with routine medical studies in terms of changing the students’ orientation toward a patient as a person.^{7–10, 13, 17} A probable reason for this observation could be the increasing cynicism resulting from exposure to agony, diseases, and death at a young age, when students may struggle in processing such deep philosophical processes. This situation may be made more difficult by the lack of positive role models and the absence of dedicated clinically oriented empathic training.^{4, 1}

In an exploratory analysis, this declining trend in empathy held true for both male and female students, despite females showing significantly higher empathy scores throughout the duration of the study, as reported in earlier studies.^{8, 9} This may also indicate that female students may provide a different type of health care because of a greater ability to empathize with the patient’s experiences and feelings. ¹³ Greater empathy may be because of the inherent warmth and caregiving attitude associated with the female gender and upbringing along with a similar image or in preparation for a role assigned by the society.^{7, 14}

Exploratory analysis of the relation of various sociodemographic variables with the change in empathy revealed some positive findings. Initially, there were no significant differences between students who belonged to nuclear or joint families and those who had siblings or not. However, as their medical training progressed, students who came from joint families or had siblings showed a significantly lesser decline in cognitive and emotional empathy than students who belonged to nuclear families and those who did not have any siblings, respectively. This difference may be because of greater emotional security and emotional need fulfillment in children who grow up in joint families. ¹¹ Similarly, siblings may be a source of warmth, closeness, and emotional support, and having siblings has been shown to prepare children better for adaptation to life in a larger society. ¹⁸ This difference in the support system may translate into a difference in emotional intelligence and resilience against stress, both directly and indirectly affecting change in empathy.

Because of the association between personality and empathy, it has been suggested that in medical education, individual personality traits need to be considered while designing programs and curricula that aim to enhance empathy. ¹⁹ However, we could not establish a statistically significant relationship between personality and empathy. The possible reasons for this could be a small sample size and examination of personality from a disorder perspective rather than individual traits.

Students showed increasing scores on GHQ-28 in each follow-up, indicating worsening psychological health with advancing medical training. This displayed a strong correlation with a decrease in empathy scores, indicating that empathic responses to others are affected by one’s personal well-being and perception of oneself. In the context of medical students, this highlights the importance of addressing individual tolerance for difficulties and stress and encouraging activities directed toward self-care and rejuvenation along with managing the heavy workload during training. ¹⁸ This finding is important not only for those in training but also for those who have advanced in their medical career and for whom working through times of personal distress has become the normal way of life, as this may lead to expecting others (patients, colleagues, and subordinates) to deal with distress in the same manner. This may have adverse outcomes on their relationship with their patients and their social and familial relationships.^{18, 19}

It is interesting to note that at comparable levels of personal distress, females had significantly greater cognitive as well as emotional empathic scores. This, along with other correlations and regression analysis, points out that empathy is affected by constitutional factors, like gender, family structure, and having siblings, and situational factors, like psychological health and general well-being.^{9, 10, 18}

These observations warrant a deeper understanding of declining empathy and concern over our medical education system in terms of inculcating in medical students humanistic qualities like empathic communication. It must not be assumed that these skills are acquired automatically during clinical training, and systematic training for good clinical practice, value education, and human behavior is strongly recommended. Constitutional factors that affect empathy are essentially nonmodifiable, whereas situational factors need to be actively addressed in this context. It is vital to encourage medical students to undertake some activities for stress relief, personal growth, and upliftment while striving to maintain a healthy balance between work and personal life. Steps need to be taken to create a nonthreatening environment in medical institutions, encouraging students to recognize signs of personal distress and psychological disturbances and seek help without fear of repercussions or discrimination.

Our findings gain recognition and validation in recent meta-analyses that find empathy training programs in medical education to be effective and recommend further experimental studies aimed at the development and utilization of standardized training programs.^{19, 20} In the same vein, it is encouraging to witness the attempts of the National Medical Council to inculcate training modules (attitude, ethics, and communication skills) aimed at enhancing these skills in the new competency-based medical education curriculum that is planned for inclusion in Indian medical institutions from the first year of MBBS itself.^{21, 22}

This study suffers from a few limitations, including data being from a single medical institution and a lack of a comparison arm. Some factors pertaining to the students’ current situation, like the current place of residence (home/hostel/rented accommodation) and scores in the first year’s professional examination, which are confounding variables, were not considered. Also, psychological disturbances were not evaluated systematically using a diagnostic tool. The tools used to assess empathy and interpersonal reactivity have not been validated for Indian medical students yet. Follow-up could be conducted for only 1½ years, and the empathy picture during the final years of the course could not be assessed.

Conclusion

Among medical students, empathy decreases with advancing training. Empathy is a dynamic attribute influenced by both constitutional factors and situational perception of well-being. There is a need to study this declining empathy and its correlates in greater detail as a potential dimension for improvement in the medical education curriculum. This study may also serve as a call for a change in the technique of imparting medical education to forthcoming generations of health care providers.