The ability to vicariously share the emotional state of others is an important evolutionary phenomenon that governs social experience and interpersonal relationships [1]. This skill seems to be affected by sleep. Recent studies have shown that sleep deprivation affects emotional empathy [2], even when individuals experience natural occurring changes in sleep quality [3]. These findings are consistent with the extensive literature on sleep loss and its negative effects on general cognitive and emotional processes [4] and confirm that poor sleep quality affects the ability of healthy individuals to experience emotional empathy.

Sleep loss and sleep fragmentation are commonly seen in individuals involved in shift work, with paramedics and police officers among the most affected [5]. Emergency medical services (EMS) are indeed required to provide uninterrupted service (24/7, every day of the year), and are often required to work an eight-day rapid rotating schedule, consisting of two consecutive day shifts (12 h each), followed by two consecutive night shifts (12 h each) and four consecutive days off, a schedule that does not allow full recovery and adequate time to adapt between night- and day-time sleep.

Here, we hypothesized that chronic poor sleep quality would have a negative impact on the emotional empathy of experienced paramedics (as compared to non-paramedics and paramedic trainees). To test this hypothesis, we evaluated participants’ quality of sleep and asked them to perform a computerized emotional empathy task; we expected paramedics to show poor quality of sleep and a significant reduced ability to empathize with others as compared to controls.



We recruited a total of 41 participants. Twelve participants [six females; mean (SD) age = 33.2 (5.47) years; mean (SD) education = 16.1 (2.12) years] with at least 3 years of experience as paramedics and currently working a full-time rapid rotating schedule, formed the paramedics group. Thirteen paramedic trainees [five females; mean (SD) age = 25.8 (6) years; mean (SD) education = 14.7 (2.27) years] with no previous experience in emergency services or patient-care occupations, formed the trainees control group. Sixteen healthy undergraduate students [10 females; mean (SD) age = 23 (7.10) years; mean (SD) education = 14.3 (2.5) years] not pursuing education towards paramedicine and with no experience in emergency services, formed the non-paramedic control group. None of the participants reported taking psychoactive medication nor had a history of medical, neurological, or psychiatric disorders including concussion. The study was approved by the local research ethics board (REB 13–0143).


We asked participants to complete the Pittsburgh Sleep Quality Index (PSQI) [6], assessing participant’s sleep quality during the month preceding the study. Anxiety and depression symptoms were evaluated through the state-trait anxiety inventory (STAI) and the beck depression inventory (BDI). Further, we used the Davidson trauma scale to assess symptoms of post-traumatic stress disorder (PTSD) (Table 1).

Table 1 Descriptive statistics

Emotional Empathy task (Fig. 1, Panel a)

Fig. 1
figure 1

Panel a Example of scrambled, neutral and negative stimuli; Panel b and c group differences in PSQI and sensitivity to negative images; Mean ± SE reported

The task included a total of 180 trials. In each trial, participants were shown color images taken from the international affective picture system (IAPS), with either negative or neutral valence. In each trial, participants were presented with one stimulus for the duration of 5 s, and asked to rate on a figural likert scale their empathic concern answering the question: “How strong is your empathy for the character(s) in the scene?”; the figural scale consisited of a reduced version of the self-assessment manikin (SAM) valence scale displaying emotional expressions, ranging from 1 (not concerned) to 4 (very concerned). The images were presented in pseudorandomized blocks of three images of the same valence and participants responded with the keyboard.

Data analyses

We assessed between groups differences in sleep quality with a one-way analysis of covariance (ANCOVA) on PSQI scores while including age and total trauma score (as computed from the Davidson Trauma Scale) as covariates. The ANCOVA was then followed up with Bonferroni-corrected (α = 0.05/2 = 0.025) a priori comparisons (Paramedics vs. Trainee, and Paramedics vs. Non-Paramedic Controls). To evaluate participants’ emotional empathy we ran a linear regression analysis with participants’ average empathy scores to the negative images as the dependent variable, and participants’ average empathy scores to the neutral images as a predictor; this analysis allowed us to output the residuals which represent participants’ responses to negative images while taking into account individuals’ empathic baseline (expressed by their responses to neutral images), a measure used in previous studies and known as “sensitivity to negative images” [3, 7]. In this analysis, we tested the assumptions of independence of residuals (Durbin-Watson = 1.626) and multicollinearity (VIF = 1 and tolerance test = 1) and confirmed absence of any violations. We then performed a one-way analysis of variance (ANOVA) to test group differences on individual’s sensitivity to negative images and followed it up with two Bonferroni corrected comparisons to test our a priori hypothesis of between groups differences in empathic responses. To further explore this relationship, we ran a Pearson correlation analysis to confirm the relationship between sleep and emotional empathy to then add the individuals’ scores on the PSQI as a covariate in the model; we performed this analysis to verify that sleep quality, as measured by the PSQI, was predictive of differences in sensitivity to negative images between groups.


The ANCOVA on the PSQI scores while controlling for age and total trauma revealed statistically significant differences between the groups (F(4,36) = 8.401, p = 0.001, \(\eta _{p}^{2}\) = 0.318). Follow-up tests with the covariates included in the model, revealed that paramedics group [Mean (SD) = 9.58 (2.42)] reported significantly worse quality of sleep than the trainees [Mean (SD) = 5.38 (1.85), p = 0.000] and the non-paramedic control group [Mean (SD) = 6.25 (2.59), p = 0.000]; the two control groups did not differ from each other (Fig. 1, Panel b). The ANOVA on the sensitivity to negative images revealed statistically significant differences between groups (F(2,38) = 5.997, p = 0.005, \(\eta _{p}^{2}\) = 0.240). Follow-up tests revealed that individuals in the paramedic group [Mean (SD) = − 0.472 (0.162)] empathized to negative images significantly less than individuals in both the non-paramedic [Mean (SD) = 0.213(0.140), p = 0.008] and trainees [Mean (SD) = 0.173 (0.156), p = 0.018] control groups (Fig. 1, Panel c). Further, the Pearson correlation revealed the presence of a statistically significant correlation between the PSQI and the sensitivity to negative images (p = 0.004, r = − 0.440), which allowed us to use the PSQI scores as a covariate in the model. As predicted, when adding individuals’ scores on the PSQI as a covariate in the model, the previously found differences between groups in sensitivity to negative images disappeared (F(3,37) = 2.115, p = 0.135, \(\eta _{p}^{2}\) = 0.103), confirming that paramedics’ lower quality of sleep, was indeed accounting for the differences in sensitivity to negative images between groups. Further, the ANCOVA with the scores on the BDI as covariate (BDI and sensitivity to negative images are correlated, p = 0.021, r = − 0.360), resulted in persisted between groups differences (F(3,37) = 4.203, p = 0.023, \(\eta _{p}^{2}\) = 0.185), indicating that depression symptoms were not predictive of differences in empathic responses to negative images between groups. Similar results were found if adding the Total trauma score as a covariate (F(3,37) = 5.328, p = .009, \(\eta _{p}^{2}\) = 0.224), indicating that variance in the empathic responses to negative images between groups was not accounted by trauma.


In this study, we found a reduction in emotional empathy responses to negative images in experienced paramedics. A variety of studies on healthcare professionals have defined this reduction in empathy as compassion fatigue [8], the loss of the ability to care empathically for patients. Importantly, this is often associated with increased risk of mental illnesses such as depression and anxiety disorders [9]. Regehr and colleagues [10] reported that paramedics often consciously dissociate themselves from their emotional reactions to both increase their ability to focus on the treatment plan, and to try to prevent long-term emotional distress. This conscious strategy is adaptive for professionals whom encounter such situations frequently. The authors also report a gradual loss of control over this conscious adaptation which can lead to the inability to relate emotionally to loved ones and engage in an emotional relationship [10]. Our findings are consistent with these findings revealing that paramedics who have been involved in the job for at least 3 years already show signs of loss of empathic ability and emotional burnout. Importantly, our data extend these findings by showing that this effect is mostly due to the poor quality of sleep that paramedics experience due to their shift work, and not due to depression or previous traumatic experience. Future research could lead to improvements in work conditions, shifts schedules, and resiliency training for paramedics, which would ultimately benefit all users of emergency medical services.