IntroductionSome studies have reported a decline in medical students’ clinical empathy during their first-year internships. Empathic skills in doctor–patient relationships have proven to have substantial benefit on the physiological as well as psychological parameters, and on the patients’ and physicians’ satisfaction. A range of experiential and interactive training modules have the potential to limit this decline and even foster the development of empathy, and yet French medical schools do not routinely implement such non-technical training. Moreover, there is no consensus about the concept of empathy, and previous studies rarely coordinate their definition with the method chosen to implement how it is measured. This study aims to evaluate the effectiveness of a training module that combines human simulations (simulated patient interviews and role-playing) and Balint groups in order to develop medical students’ empathy as a measured cognitive attribute, declarative knowledge, and their correlation. MethodsWe implemented a monocentric prospective study of fourth-year medical students from the Île-de-France area, which included pre/post-test assessment. First-year clinical interns initially completed five role-play sessions of one hour and a half each, with a group of 12 to 14 students led by a trained senior medical doctor. The sessions addressed successively issues of the doctor–patient relationship, breaking bad news scenarios, management of chronic conditions, violence and aggression, inappropriate treatment requests, or communication about medical errors. A pre-briefing session provided oral and written instruction and promoted a non-judgmental approach. Then two students role-played the doctor–patient interaction. Finally, a three-stage debriefing supported reflective learning. Subsequently, the students attended six Balint group sessions of one hour and a half each, within the same groups. One of the students was encouraged to bring up a challenging clinical situation. Then, the trained Balint supervisor invited all the students to share their observations, so as to explore the assumptions, beliefs and emotions raised by this account, and to collectively reframe the multiple perspectives on the case. At the end of the year, each student carried out two ten-minute consultations with a simulated patient, without an observer. The clinical scenarios were an HIV diagnosis disclosure and resisting a request for a prescription for antibiotics for a common cold. After each simulation, the simulated patient (SP) provided direct feedback to the student, with at times additional feedback given by their supervisor. The SP feedback was focused on empathy. The level of the medical students’ empathy was assessed using the self-reported Jefferson Scale of Empathy, which measures their declarative knowledge through multiple-choice questions. ResultsOf the 100 students who completed the empathy questionnaires, we found a statistically significant difference between the pre (108.8±11.9) and post-test (113.7±10.4; P<0.001). On the three JSPE-MS subscales, two students reported a statistically significant improvement (perspective taking; and the ability to put themselves in the patient's shoes). Of the 101 students who completed the knowledge questionnaires, we found a significant difference between the pre (2.3±1.1) and post-test (2.9±1.2; P<0.001). Moreover, a weak correlation was found (r=0.256) between improvements of empathy and knowledge assessments. DiscussionThese results in the French context are consistent with those reported in international literature on the benefits of implementing simulation training and Balint groups to limit the decline of empathy during the medical students’ first clinical internships. These results are further reinforced by the consistency between the definition of empathy and the implementation of a means to measure it as a cognitive dimension. The reliance on self-assessment of empathy might lessen the validity of these results and stresses the need for a wider multidimensional approach. The quality of the evidence is also limited by a monocentric, non-randomized and non-controlled design, which calls for further research. ConclusionThe use of human simulations combined with Balint groups seems to support the development of the cognitive empathy of medical students in the French context. Given the recent developments in remote learning worldwide to comply with social distancing measures, their online translation into role-played tele-consultations would appear to be a very topical opportunity, while also avoiding the inherent limitations of virtual reality to address non-technical skills.