Burnout Syndrome in the Intensive Care Unit Before and During a Palliative Care Trial

One hundred ICU physicians, nurses and clinical psychologists drawn from 4 / 8 ICUs of the Greek group of the 28 European adult units enrolled in the EPIC stepped-wedge trial (ClinicalTrials.gov NCT06605079) will each complete an identical four-part survey twice-once during their unit's usual-care phase and again after crossover into the EPIC tele-palliative-care intervention. Administered in two stages (Parts A and B via secure email link; Parts C and D during a facilitated online session), the entire battery requires about forty minutes per administration and produces detailed profiles of clinician demographics, workplace stressors, core burnout dimensions and self-compassion. Within-subject repeated-measures analyses will then quantify how the combination of blended-learning palliative-care modules and standardized teleconsultations influences individual well-being over time.

Part A consists of fifteen fixed-response items capturing demographic and professional background. Respondents select from categorical options for gender (male/female), age bracket, family status and number of children, as well as highest educational attainment (secondary, vocational, undergraduate, postgraduate, PhD) and whether they have previously received palliative-care or ethics training (yes/no). Work-related variables include role (doctor, nurse, psychologist), years of ICU experience (0-10, 10-20, 20-30, >30), average evening hours per month (<3, 3-5, >5), typical nightly sleep duration (<5 hours, 5-7 hours, >7 hours) and annual leave days (<15, 15-30, >30). Participants also indicate the type of unit in which they work (mixed/multi-purpose, surgical, cardiology or other). These objective data provide essential covariates for exploring whether baseline characteristics moderate the EPIC intervention's effects on burnout and self-compassion trajectories.

Part B probes the frequency of twenty-two common ICU stressors and ethical dilemmas using a five-point Likert frequency scale (1 = Never, 2 = Rarely, 3 = Sometimes, 4 = Often, 5 = Always). Questions address emotional distance from colleagues, resource constraints that force care adjustments, pressure from patient families, internal moral conflict between life-support and quality-of-life decisions, perceived control over treatment choices, work-life balance and opportunities for professional development. For example, clinicians rate how often they must limit treatments due to limited resources, experience doubts about ethical issues in prognostic decision-making or feel that their performance does not reflect their skills. By capturing both external pressures and internal moral distress, Part B establishes a baseline against which EPIC's specialist palliative consultations can be assessed for stress-reduction efficacy.

Part C combines two respected burnout instruments. The first is a nineteen-item subset of the Maslach Burnout Inventory (MBI), administered on a seven-point frequency scale (1 = Never, 2 = A few times a year, 3 = Once a month, 4 = A few times a month, 5 = Once a week, 6 = A few times a week, 7 = Every day). Items form three subscales-emotional exhaustion (for instance, "I feel emotionally drained by my work"), depersonalization ("I treat some patients as if they were objects") and personal accomplishment ("I feel I have achieved many worthwhile things in this job"). Previous ICU studies consistently report Cronbach's alpha of approximately 0.90 for the MBI total and its subscales, indicating excellent internal consistency. The second is a sixteen-item subset of the Oldenburg Burnout Inventory (OLBI), using a four-point agreement scale (1 = Strongly agree, 2 = Probably agree, 3 = Probably disagree, 4 = Strongly disagree). The OLBI yields two subscales-exhaustion ("After work I often feel worn out") and disengagement ("I feel indifferent toward my work"). Cronbach's alpha for the combined OLBI scale typically exceeds 0.80, with subscale values around 0.87 for exhaustion and 0.81 for disengagement. By utilizing both instruments, Part C captures affective depletion and cognitive withdrawal components of burnout, providing a comprehensive measurement of clinicians' occupational distress.

Part D employs the full sixteen-item Self-Compassion Scale developed by Neff, rated on a five-point frequency scale (1 = Almost never, 2 = Rarely, 3 = Sometimes, 4 = Often, 5 = Almost always). Items evaluate three positive dimensions-self-kindness, common humanity and mindfulness-and their negative counterparts-self-judgment, isolation and over-identification. Sample statements include "I try to be understanding toward those aspects of my personality I don't like" and "When I'm feeling down I tend to obsess on everything that's wrong." In Greek ICU populations, both the positive and negative subscales consistently achieve Cronbach's alpha near 0.86, confirming reliable assessment of clinicians' self-compassion levels. Because higher self-compassion has been shown to buffer against burnout, Part D provides critical insight into personal resilience factors that EPIC may enhance.

To maximize data integrity, Parts A and B are distributed seven days before the scheduled online session via encrypted email links, allowing clinicians to complete demographic and stressor items thoughtfully at their convenience. During a facilitator-led videoconference, respondents then complete Parts C and D in real time, with on-screen guidance to clarify any item interpretations. Electronic responses are pseudonymized and stored in a secure database meeting GDPR requirements. Ethical approval covers both survey administrations, and informed consent materials emphasize voluntary participation, the right to withdraw at any point and assurances that individual responses will remain confidential and not influence professional evaluations.

The primary analytical objective is to test within-subject changes in total and subscale scores for Part B stressors, Part C burnout dimensions and Part D self-compassion between the control and EPIC phases. Mixed-effects models-with clinician as a random intercept and fixed effects for time period, unit and relevant Part A covariates such as years of experience or prior palliative training-will assess whether the blended-learning modules and teleconsultations yield statistically significant reductions in emotional exhaustion, depersonalization and disengagement, alongside increases in self-compassion. Secondary analyses will explore moderation by baseline characteristics, clarifying which clinician subgroups derive the greatest benefit from EPIC. Effect sizes will be interpreted against established benchmarks for within-subject interventions, with special attention to medium-sized shifts in Cronbach's alpha-validated scales.

By integrating rich background profiles, a granular inventory of ICU stressors, complementary burnout measures and a validated self-compassion assessment-all framed within a robust repeated-measures design-this extended four-part survey promises nuanced insights into how telemedicine-enabled palliative-care integration can bolster resilience, reduce moral distress and foster sustainable clinician engagement in the high-pressure ICU environment.

Sample Size Determination A priori power analysis using a one-group, two-measurement repeated-measures ANOVA (within-subjects) indicated that, to detect a medium effect size (f = 0.25) on the primary outcome (within-subject change in composite burnout score) with α = 0.01, 95% power, a correlation of 0.5 between pre- and during-intervention scores, and perfect sphericity (ε = 1), 75 clinicians are required (denominator df = 74, λ = 18.75, critical F = 6.99). To allow for up to 25% attrition and support subgroup and moderator analyses, the target enrollment is set at 100 ICU clinicians.

Multivariable mixed-effects logistic regression will test the odds of a clinically meaningful burnout change-defined by 95% minimal detectable change thresholds (≥7-point reduction on the Maslach Burnout Inventory or ≥8-point reduction on the Oldenburg Burnout Inventory). The binary responder outcome (achieved vs. not achieved) will include a random intercept for ICU site to account for clustering.

Fixed-effect predictors (up to 10) will include:

• Age (years)
• Gender
• Profession (nurse, physician, other)
• Years of ICU experience
• Unit type (medical, surgical, mixed)
• Shift pattern (day, night, mixed)
• Baseline burnout score
• Baseline self-compassion score
• Prior palliative-care training (yes/no)
• EPIC trial phase (intervention vs. control)