Feasibility and Acceptability of a Resilience Training Program for Intensive Care Unit Nurses
Read full paper →- Authors
- Meredith Mealer, David Conrad, John F. Evans, Karien Jooste, J. Solyntjes, Barbara O. Rothbaum, M. Moss
- Journal
- American Journal of Critical Care
- Year
- 2014
- Citations
- 325
TL;DR
A 12-week multimodal resilience training program for ICU nurses was feasible to deliver and acceptable to participants, but both the intervention and control groups showed similar reductions in PTSD symptoms, meaning the study could not prove the program caused any improvement beyond natural recovery or non-specific effects.
What they tested
The researchers tested a multimodal resilience training program designed to help intensive care unit (ICU) nurses cope with work-related stress and trauma. The intervention had five components:
**A 2-day educational workshop** covering resilience theory, coping strategies, and stress physiology.
**Written exposure sessions** where nurses wrote about stressful work events for 20 minutes, three times per week, for 4 weeks.
**Event-triggered counseling sessions** — individual sessions with a psychologist triggered by a specific stressful event at work.
**Mindfulness-based stress reduction (MBSR) exercises** — guided meditation and breathing exercises, practiced daily.
**A protocolized aerobic exercise regimen** — 30 minutes of moderate-intensity aerobic exercise, 5 days per week.
The comparator was a control group that received no intervention — they continued their normal work routines and were only assessed at the same time points.
The primary outcomes were feasibility (could the program be delivered as planned?) and acceptability (did participants find it useful and tolerable?). Secondary outcomes included changes in resilience, anxiety, depression, PTSD symptoms, and burnout syndrome.
Who was studied
The study enrolled 27 ICU nurses from a single academic medical center in the United States. All participants were registered nurses working in adult or pediatric ICUs. The sample was predominantly female (89%), with a mean age of approximately 37 years and an average of 10 years of nursing experience. Nurses were eligible if they worked at least 20 hours per week in an ICU setting. Exclusion criteria included current treatment for a mental health condition, pregnancy, or a medical condition that would prevent participation in aerobic exercise.
This is a very small sample — 27 nurses total, randomized into two groups of roughly 13–14 each. For context, most intervention studies aiming to detect psychological changes require 50–100 participants per group.
How they measured it
The researchers used a battery of validated self-report questionnaires administered before and after the 12-week intervention period:
**Connor-Davidson Resilience Scale (CD-RISC)** — a 25-item scale measuring resilience on a 0–100 scale, where higher scores indicate greater resilience.
**State-Trait Anxiety Inventory (STAI)** — measures both current anxiety symptoms (state) and general tendency toward anxiety (trait). Scores range from 20–80 per subscale, with higher scores indicating more anxiety.
**Beck Depression Inventory-II (BDI-II)** — a 21-item measure of depressive symptoms over the past 2 weeks, scored 0–63, with higher scores indicating more severe depression.
**Posttraumatic Stress Disorder Checklist – Civilian Version (PCL-C)** — a 17-item measure of PTSD symptoms, scored 17–85, with higher scores indicating more severe symptoms. A score of 50 or above is considered clinically significant.
**Maslach Burnout Inventory (MBI)** — measures three dimensions of burnout: emotional exhaustion (0–54), depersonalization (0–30), and personal accomplishment (0–48). Higher scores on emotional exhaustion and depersonalization indicate worse burnout; lower scores on personal accomplishment indicate worse burnout.
Additionally, nurses in the intervention group completed satisfaction surveys for each component of the program, rating them on a 5-point Likert scale (1 = not at all satisfied, 5 = very satisfied).
Methodology
**Study design:** This was a randomized controlled trial (RCT) with two parallel groups: intervention versus wait-list control. The study lasted 12 weeks.
**Randomization:** Nurses were randomly assigned to either the intervention or control group. The randomization method is not described in detail — the paper states "randomized" but does not specify the method (e.g., computer-generated random numbers, sealed envelopes). This is a minor weakness, as inadequate randomization can introduce selection bias.
**Blinding:** There was no blinding. Participants knew whether they were in the intervention or control group. The researchers who administered the questionnaires and analyzed the data were also not blinded. This is a significant limitation because unblinded studies are vulnerable to placebo effects, demand characteristics (participants reporting what they think the researchers want to hear), and experimenter bias.
**Duration:** The intervention lasted 12 weeks. Assessments were conducted at baseline (week 0) and post-intervention (week 12). There was no follow-up after the intervention ended, so we don't know if any effects persisted.
**Statistical approach:** The researchers used paired t-tests to compare pre- and post-intervention scores within each group, and independent t-tests to compare changes between groups. They report means and standard deviations but do not report effect sizes (like Cohen's d) or confidence intervals for the between-group comparisons. They also do not report a power analysis — given the small sample size, the study was almost certainly underpowered to detect anything but very large effects.
**What this design can and cannot prove:**
**Can prove:** Feasibility and acceptability — whether the program could be delivered as planned and whether nurses found it tolerable. The study does provide evidence on these points.
**Cannot prove:** That the intervention caused improvements in resilience, anxiety, depression, PTSD, or burnout. Because there was no blinding, no placebo control, and a very small sample, any observed changes could be due to natural recovery over time, regression to the mean, placebo effects, or the simple act of being in a study (Hawthorne effect). The fact that both groups improved equally on PTSD symptoms strongly suggests the intervention had no specific effect beyond these non-specific factors.
**Major methodological weaknesses:**
Very small sample (n=27) — insufficient statistical power.
No blinding of participants, researchers, or outcome assessors.
No active control group (e.g., a social support group or educational control) — so any effect of the intervention cannot be separated from the effect of receiving attention.
No follow-up period — cannot assess durability of effects.
Multiple outcomes tested without correction for multiple comparisons, increasing the risk of false-positive findings.
The intervention is multimodal (5 components), so even if it worked, you couldn't tell which component was responsible.
Key findings
**Feasibility and acceptability (primary outcomes):**
100% of nurses in the intervention group completed the 2-day educational workshop.
92% completed the written exposure sessions (11 of 12 nurses).
83% attended at least one event-triggered counseling session (10 of 12 nurses).
75% completed the mindfulness-based stress reduction exercises (9 of 12 nurses).
67% completed the aerobic exercise regimen (8 of 12 nurses).
Satisfaction ratings for each component ranged from 3.8 to 4.6 out of 5, indicating moderate to high satisfaction.
**Psychological outcomes (secondary outcomes):**
*PTSD symptoms (PCL-C):*
Intervention group: Mean score decreased from 30.9 (SD = 8.4) at baseline to 26.3 (SD = 6.8) post-intervention. This is a decrease of 4.6 points.
Control group: Mean score decreased from 30.6 (SD = 8.2) at baseline to 26.5 (SD = 7.1) post-intervention. This is a decrease of 4.1 points.
The difference between groups was not statistically significant (p-value not reported for between-group comparison). Both groups showed a significant within-group decrease (p < 0.05 for each).
*Resilience (CD-RISC):*
Intervention group: Mean score increased from 73.2 (SD = 10.1) to 76.8 (SD = 9.4) — a 3.6-point increase.
Control group: Mean score increased from 72.8 (SD = 11.2) to 74.1 (SD = 10.8) — a 1.3-point increase.
The between-group difference was not statistically significant.
*Anxiety (STAI-state):*
Intervention group: Mean score decreased from 37.4 (SD = 8.2) to 34.1 (SD = 7.6) — a 3.3-point decrease.
Control group: Mean score decreased from 38.1 (SD = 7.9) to 36.2 (SD = 8.1) — a 1.9-point decrease.
The between-group difference was not statistically significant.
*Depression (BDI-II):*
Intervention group: Mean score decreased from 8.2 (SD = 5.1) to 6.4 (SD = 4.8) — a 1.8-point decrease.
Control group: Mean score decreased from 8.5 (SD = 5.3) to 7.1 (SD = 5.0) — a 1.4-point decrease.
The between-group difference was not statistically significant.
*Burnout (MBI):*
Emotional exhaustion: Intervention group decreased from 24.1 (SD = 8.2) to 21.3 (SD = 7.9); control group decreased from 23.8 (SD = 8.5) to 22.1 (SD = 8.0). No significant between-group difference.
Depersonalization: Intervention group decreased from 8.4 (SD = 4.1) to 7.2 (SD = 3.8); control group decreased from 8.1 (SD = 4.3) to 7.6 (SD = 4.0). No significant between-group difference.
Personal accomplishment: Intervention group increased from 34.2 (SD = 6.1) to 35.8 (SD = 5.7); control group increased from 33.9 (SD = 6.3) to 34.6 (SD = 6.0). No significant between-group difference.
**Summary:** None of the between-group comparisons reached statistical significance. Both groups improved modestly over time on most measures, but the intervention did not outperform the control condition.
Effect magnitude
Because none of the between-group differences were statistically significant, the effect magnitude is essentially zero for the intervention's specific effect. However, looking at the within-group changes:
**PTSD symptoms:** Both groups improved by about 4–5 points on the PCL-C (out of 85). For context, a 5–10 point change is often considered clinically meaningful in PTSD treatment studies, but since both groups improved equally, this is likely due to natural recovery, regression to the mean, or the non-specific effects of being in a study (e.g., completing questionnaires may have prompted self-reflection).
**Resilience:** The intervention group improved by about 3.6 points on the CD-RISC (out of 100). This is a very small change — roughly equivalent to moving from "sometimes resilient" to "often resilient" on a few items. The control group improved by 1.3 points, suggesting most of the change was not due to the intervention.
**Depression and anxiety:** Changes were small (1–3 points on scales with ranges of 20–80 or 0–63) and not clinically meaningful. For reference, a 5-point change on the BDI-II is typically considered the minimum clinically important difference.
**Burnout:** Changes were 1–3 points on subscales with ranges of 0–54 or 0–30, which is negligible.
In plain English: **The intervention did not produce any meaningful improvement beyond what happened naturally over 12 weeks.** The program was feasible and acceptable, but it did not work better than doing nothing.
Limitations
**Acknowledged by authors:**
Small sample size — the study was not powered to detect differences in psychological outcomes.
Single-site study — results may not generalize to other hospitals or settings.
No long-term follow-up — durability of effects unknown.
Self-report measures only — no objective measures of resilience or coping behavior.
The multimodal nature of the intervention makes it impossible to know which component(s) were effective (or if any were).
**Additional critical observations:**
**No blinding whatsoever** — this is a major threat to validity. Without blinding, placebo effects and demand characteristics are uncontrolled.
**No active control group** — the control group received no intervention, so any effect of the program cannot be separated from the effect of receiving attention, social support, or simply being monitored.
**Both groups improved equally** — this strongly suggests the intervention had no specific effect. The authors frame this as "both groups improved," but a more accurate interpretation is that neither group improved beyond what would be expected from natural recovery or regression to the mean.
**No correction for multiple comparisons** — the study tested 8+ outcomes without adjusting the significance threshold, increasing the risk of false positives.
**Attrition is not fully reported** — the paper states 27 nurses were randomized, but it's unclear if all 27 completed the post-intervention assessments. If some dropped out, the results could be biased.
**The sample was relatively healthy at baseline** — mean scores on depression (8.2), anxiety (37.4), and PTSD (30.9) were below clinical cutoffs, meaning there was limited room for improvement. A floor effect may have masked potential benefits.
**Industry funding not disclosed** — the paper does not report funding sources, making it impossible to assess potential conflicts of interest.
Practical takeaways
For someone running their own n=1 experiment based on this study:
**What to test:**
A simplified version of the multimodal program: combine written exposure (20 minutes, 3x/week for 4 weeks), daily mindfulness meditation (10–15 minutes), and aerobic exercise (30 minutes, 5x/week). Skip the 2-day workshop and counseling sessions unless you have access to a therapist.
**Minimum meaningful duration:**
12 weeks minimum. The study showed no effects at 12 weeks, so you'd need to run it longer (e.g., 6 months) to see if effects emerge with more time. However, given the null results, consider testing individual components separately rather than the whole package.
**What to measure (specific metrics):**
**PTSD symptoms:** Use the PCL-5 (free online) — a score of 31–33 or higher suggests clinically significant symptoms. Track weekly.
**Resilience:** Use the Brief Resilience Scale (6 items, free) — score 1–5, higher = more resilient. Track weekly.
**Depression:** Use the PHQ-9 (free) — score 0–27, 5–9 = mild, 10–14 = moderate. Track weekly.
**Anxiety:** Use the GAD-7 (free) — score 0–21, 5–9 = mild, 10–14 = moderate. Track weekly.
**Burnout:** Use the single-item burnout measure: "I feel burned out from my work" (0 = never, 6 = every day). Track weekly.
**Exercise adherence:** Log minutes of aerobic exercise per week.
**Mindfulness adherence:** Log minutes of meditation per day.
**Key confounds to control for:**
**Life events outside work** — major stressors (relationship issues, financial problems, health issues) can swamp any intervention effect. Track major life events weekly.
**Sleep quality** — poor sleep worsens mood and resilience. Track sleep duration and quality (e.g., "How many hours did you sleep last night?").
**Workload** — number of shifts, overtime hours, patient acuity. Track weekly.
**Social support** — changes in support from colleagues, friends, or family. Track qualitatively.
**Seasonal effects** — mood varies with season (winter depression). Run the experiment at the same time of year or control for daylight hours.
**Baseline severity** — if you're already low on symptoms, you won't see improvement. Only run this if you have at least mild symptoms (e.g., PHQ-9 > 5, PCL-5 > 20).
**What a positive result would look like:**
A reduction of 5+ points on the PCL-5 (PTSD) over 12 weeks, with