The main findings of this study are that empathy declined among men medical students' during their first extensive clinical experience, and that the decline was associated with pre-clinical humanities program. Among women students, there was no decline in empathy during the fourth-year of studies, regardless of type of humanities program.
In addition, women who participated in the extended humanities program had higher JSPE-S scores during the fourth-year as compared to women who participated the limited program.
In contrast to the finding in women, among men we found a significant decline in empathy during the fourth-year of studies in those who participated in the extended humanities program, but not in those who participated in the limited program.
Our study suggests, according to the large to medium effect sizes observed, that the differences in empathy scores in women who participated in the extended humanities program as compared to the limited program are not only statistically significant but are also likely to be substantial. Further study is required to determine the practical and educational implications of these findings. Such implications may include increased awareness of curriculum planners to potential sources for gender differences in educational interventions aimed to enhance empathy among medical students.
Previous studies [1,10,11,12] have suggested that an overall decline in JSPE-S scores during the third-year of medical studies exists. Our study suggests that an extensive program that included exposure to ethical issues, communication skills, and humanities studies, was associated with a decline in empathy among men students, while among female students a decline was not observed, regardless of medical humanities program.
Previous studies on gender effect on medical students’ empathy yielded inconsistent findings [1,11,15-17]. Our findings suggest that gender differences in empathy may exist, and go along with gender differences previously reported in clinical practice [32, 33]. It is possible that gender-specific impact of educational programs that we and others [22, 23] have observed were underestimated in previous studies because of small samples which did not allow to assess such effects. It is also possible that measuring empathy at a single time point would be less sensitive to detect gender differences in empathy as compared to longitudinal studies.
A secondary aim of the study was to contribute to the validity evidence of the Hebrew version of the JSPE-S by examining its relationship with an established measure of empathy. We observed a significant correlation between the JSPE-S and the IRI total score administered at the beginning of the study. The correlations between the JSPE-S and the IRI subscales were significant for Perspective Taking and for Empathic Concern subscales, while no correlation was observed with the Fantasy and Personal Distress subscales, similarly to the findings of Hojat in the original JSPE English version , providing validity evidence of the Hebrew version of the JSPE-S.
Because our study was not randomized it is possible that the observed differences in empathy change between cohorts are the result of baseline difference between cohorts. Although there was no difference between cohorts in gender, average age, marital status, it is possible that other variables that were not captured by sociodemographic measures that were collected contributed to the observed differences.
The decline in empathy during the first clinical year, while students are introduced to the clinical work in the wards, can have several potential explanations. These include de-idealization of students' perception of medicine , lack of proper role models , and students’ perception that, as compared to the power of technology and the intense clinical experience, empathy may not be a significant tool in the profession of medicine as students had believed it to be before they entered clinical life . Students can easily put aside the importance of interpersonal engagement in patient care when the majority of their studies are based on quantitative scientific outcomes. At the same time, the decline in empathy among medical students may reflect a protective mechanism that can help students to deal with emotionally difficult situations .
Counterintuitively, the larger decline in empathy was observed in male students from the cohort who participated in the extended medical humanities program, as compared to male students who participated in the limited program.
This finding may be explained by the lack of continuous medical humanities program during the fourth year of studies. This could have potentially resulted in the greater decline in empathy in students who were used to participate regularly in a medical humanities program during their first three years of studies, and had a greater level of empathy at the beginning of the fourth year, as compared to students who received only a limited program during their first year of studies (even though the difference at the beginning of the fourth year was not statistically significant). At the end of the fourth year, empathy in both cohorts was similar, but the decline was significantly greater in the extended medical humanities cohort, who had higher levels of empathy at the beginning of the fourth year of studies.
These findings do not support the possibility that extensive pre—clinical medical humanities programs have an "immunizing" effect on medical students' decline in empathy during the clinical studies, and may support the need for continuous medical humanities program through-out medical studies.
Additional research is required to investigate this possibility, and to determine why the decline was observed in men but not in women.
Our study has several limitations. The study included a single medical school, which may limit the generalization of the findings. Cultural differences and differences in the average age in which students start medical school, may affect students' previous life experiences and empathy levels. For example, the average starting age for medical school in Israel is higher than in USA  or Ethiopia . Such differences might have an impact on our results regarding students’ empathy. In addition, our study was based on a self-reported empathy measurement, the JSPE-S, and not on observed behaviors, that may only partially correlate [39, 40].
We followed students during the 4th year of studies, and not during all 3 clinical years. This has likely limited our conclusions regarding changes in empathy during medical studies. In addition, the aim of this study was to explore empathy changes that have been previously suggested to occur following the first students' clinical experiences during clerkships. Thus, we evaluated empathy at the beginning and following the fourth year of medical studies in three consecutive classes of medical school. However, since JSPE-S scores at entry to medical school are not available, we cannot exclude the possibility that differences in empathy between cohorts, prior to entrance to medical school, could have contributed to our findings. However, we examined the effect of medical humanities curriculum on empathy in two cohorts of students that were admitted to medical school using the same admission system (MMI), and this could contribute to reduce differences between cohorts in baseline empathy.
Due to the observational design of our study, students' randomization to the medical humanities programs was not possible, and comparisons were made between cohorts. Although randomization is considered a gold standard in clinical studies, it has been recognized that it is difficult to blind learners to their assigned group in educational studies . A clinical research model that has been suggested to be more applicable for educators is the "pragmatic trail" in which interventions are compared in real-world practice . Thus, we measured empathy in a prospective controlled study. We used a longitudinal design with repeated measurements to compare changes in empathy over time in the cohorts studied. In addition, a single humanities program was offered in each year (limited/extended), and students were obligated to participate in it, therefore students' preferences could not affect their participation in the limited/extended programs.