Compassion in Healthcare Education

Empathy, in the strictest sense, is the sharing of another person's emotional state. When a clinician empathizes with a patient in pain, the clinician's own neural networks for processing pain activate, specifically the anterior insula and the anterior cingulate cortex (Singer et al., 2004; Singer & Klimecki, 2014). The clinician is not metaphorically feeling the patient's pain. The clinician is, at the level of brain circuitry, generating a version of it. Repeated activation of these networks, day after day, patient after patient, without protective skill, produces what Singer and Klimecki (2014) named empathic distress fatigue.

The healthcare literature has been calling it compassion fatigue since Figley (1995) introduced the term. The label was wrong. What fatigues caregivers is not their compassion. It is their unguarded empathy.

Compassion is something else. It is caring concern for another person's suffering combined with a motivation to help, but without the absorption of that suffering as one's own. The neural signature is categorically different. Compassion activates the ventral striatum, the medial orbitofrontal cortex, and other regions associated with affiliation, reward, and caregiving motivation (Klimecki et al., 2014; Singer & Klimecki, 2014).

Matthieu Ricard captured the difference: "Empathy is the resonance with another person's feelings. Compassion is a benevolent state of mind that wishes for others to be free from suffering" (Ricard, 2015).

Martingano and colleagues (2025) provided the most comprehensive empirical synthesis of this distinction across healthcare populations. Cognitive empathy and compassion were negatively associated with burnout. Emotional contagion (the involuntary absorption of others' emotional states) was positively associated with burnout. The pattern is robust. It is not the practitioners who care most who burn out fastest. It is the practitioners who care in the wrong mode.

More than two thirds of nurses report burnout on most days, and over half report wanting to resign. Physician burnout, which had recently dropped below 50% for the first time in four years, remains far above pre-pandemic levels (American Medical Association, 2024).

The downstream effects on patient care are equally well documented. Li and colleagues (2024), in a meta-analysis covering 85 studies and more than 288,000 nurses across 32 countries, found that nurse occupational distress was significantly associated with lower patient safety climate, increased nosocomial infections, more medication errors, more adverse events, more patient falls, and lower patient satisfaction. A Mayo Clinic study reported that surgeons with the highest depersonalization scores were three times more likely to commit major surgical errors. Compassionate care is not a luxury. It is, increasingly, a patient-safety variable.

Healthcare students are not insulated from this. They arrive into a workforce in crisis, often through training programs that mirror the very dynamics producing the crisis. Hojat and colleagues (2009), in a longitudinal study they titled "The Devil Is in the Third Year," documented that empathy scores in medical trainees decline significantly during clinical training, with the steepest erosion occurring during clinical rotations, exactly the point at which students first encounter real patient suffering in quantity.

The mechanism is consistent and predictable. A student arrives in a healthcare program with normal, intact human responsiveness to suffering. Then the curriculum begins. The hidden curriculum, in particular, begins. Students learn that emotion is unprofessional, that vulnerability is a liability, that not knowing is shameful, that asking for help signals weakness, and that what counts in evaluations is technical proficiency rather than the felt quality of presence with patients (Hafferty, 1998; Lempp & Seale, 2004).

They watch their senior role models cope by detaching, and they internalize detachment as the professional norm. They are exposed to suffering before they have any practiced way of meeting it, so they do what nervous systems do under repeated overload. They wall it off, or they burn out (Hojat et al., 2009).

Add to this the cultural water in which Western healthcare swims. Western culture, broadly, runs on independent self-construal: the idea that identity is built through individual achievement and personal autonomy rather than through relationships and community (Markus & Kitayama, 1991). It rewards a particular form of self-worth that is contingent on accomplishment and outperformance (Crocker et al., 2003). It tends to pathologize self-directed kindness as weakness or self-indulgence.

When the team at Stanford's Center for Compassion and Altruism Research and Education (CCARE) first piloted Compassion Cultivation Training with American undergraduates, they encountered something unexpected. American participants, asked to direct kind wishes toward themselves, reported discomfort, resistance, and outright aversive reactions (Jinpa, 2015). The traditional starting point of the practice was actively blocking the rest of it.

Paul Batalden's often-quoted observation applies: Every system is perfectly designed to get the results it gets. A training system that produces guarded, depleted, depersonalized graduates is a system that has been designed, perhaps unintentionally, to produce them.

A muscle has four properties that matter here. It is built, not given. It strengthens with use and weakens without. It can be actively inhibited by the surrounding system, even when the person owning the muscle is willing to use it. And the rate at which it gains or loses strength is much faster than most people assume.

The lesson, then, is to stop treating compassion as a feature of personality and start treating it as a feature of the body, trained or untrained, supported or suppressed, used or neglected. Once we make that shift, the depleted clinician is no longer a character we are reading. They are a body whose conditioning we can read, and whose conditioning we can change.

The evidence is at this point converging:

Weng and colleagues (2013) showed that two weeks of compassion-based meditation training produced measurable changes in brain regions associated with understanding others and emotional regulation, with changes that predicted real-world altruistic behavior.

Seppala and colleagues (2014) demonstrated that even a single ten-minute session of loving-kindness meditation increased other-focused positive affect and decreased self-focused rumination.

Jazaieri and colleagues (2013), in a randomized controlled trial of Stanford's Compassion Cultivation Training, found significant increases in self-compassion, compassion for others, mindfulness, and happiness alongside reductions in worry and emotional suppression.

Villalon and colleagues (2025), in a randomized trial of 474 Chilean physicians, reported large and sustained reductions in burnout (effect size d = -1.08 at six-month follow-up) following a culturally adapted mindfulness and compassion intervention, with downstream reductions in self-reported medical errors mediated by the burnout reduction itself.

The training is dose-responsive and brief enough to be feasible inside busy professional curricula. Two weeks of practice produces neural change. Ten minutes shifts affect. Three minutes of compassionate breathing activates self-soothing physiology (Germer & Neff, 2013). A four-week program with three short sessions per week sits comfortably within the schedule of any undergraduate or graduate healthcare program.

The Western context makes this training more important rather than less. Cultures with strong interdependent self-construal can lean on community and relationship as the natural cradle of compassion. Western individualistic cultures, on the whole, cannot. Self-compassion has to be deliberately taught, often before other-directed compassion can take hold.

First, the empathy decline begins in pre-clinical years and accelerates in clinical rotations (Hojat et al., 2009). Adding a compassion curriculum after that decline is already underway means trying to repair damage that did not need to occur. Adding it before clinical exposure equips students with skill before they need it.

Second, professional identity is forming, and what gets taught early gets taught deeply. A first-year student who learns explicitly that empathy and compassion are different, that the distinction has neural correlates, and that the protective form of caring is a practiced skill rather than an innate trait, builds a different professional identity than a student who learns implicitly that caring well means caring less.

Third, the cultural impediments to self-compassion in Western learners (Jinpa, 2015) need time to be worked through. A student who first encounters loving-kindness practice in a senior elective will probably hit the same wall the Stanford undergraduates hit, and may not have the time or repeated exposure to move past it. A first-year student who hits that wall has three or four years of curriculum ahead of them in which to work through it.

Fourth, the educational opportunity is uncommonly clean. In most areas of curriculum, content is hotly contested and time is scarce. Compassion training, by contrast, has been shown to require a small number of short sessions, can be embedded in existing professional formation or wellbeing modules, and produces measurable benefits on outcomes that programs already report on, including academic stress, retention, and professional identity.

Two findings that matter for educational design: First, the most effective student engagement was informal and relational rather than mass-mailed. Cohort-wide emails produced 1 to 2 percent attendance. Tutors and trusted faculty members talking to their students about Rounds produced markedly higher attendance. Second, the drivers of success behaved like a dimmer switch rather than a binary. The more strongly each driver was present, the more successfully Rounds were implemented.

Schwartz Rounds operate in over 600 organizations across the US, Canada, Australia, and New Zealand, with roughly 280 additional sites in the UK and Ireland.

The skills include:

These translate the inner stance into the patient's experience. The bridge between intention and effect is skill. The skills are teachable.