R2D2-ICU — Restrictive vs Liberal Physical Restraint Strategies

1. Publication

• Title: Restrictive vs Liberal Physical Restraint Strategies in Critically Ill Patients: The R2D2-ICU Randomized Clinical Trial
• Acronym: R2D2-ICU
• Year & Journal: JAMA, epublished March 17, 2026 (2026;335(14):1232-1242)
• Citation: Sonneville R, Couffignal C, Sigaud F, et al; R2D2-ICU Investigator Study Group. Restrictive vs Liberal Physical Restraint Strategies in Critically Ill Patients: The R2D2-ICU Randomized Clinical Trial. JAMA. 2026;335(14):1232-1242. doi:10.1001/jama.2026.2897

2. Context & Rationale

Background: ~41.6% of ICU patients are exposed to physical restraint (high between-study heterogeneity), used to prevent self-extubation and device removal, but the rationale has always been largely inferential. Adverse-event literature links restraint with skin injury, subsequent delirium, neurofunctional impairment, and PTSD symptoms. Until 2026, restraint use had never been examined in a randomized trial.

Research Question/Hypothesis: In mechanically ventilated ICU adults, does a restrictive (low-use) wrist-strap restraint strategy increase days alive without delirium or coma over 14 days, compared with a liberal (high-use) strategy?

Why This Matters: A rigorous trial was needed before routine prophylactic restraint could be defended as standard care; restraint restricts autonomy and may worsen distress despite being normalized in practice.

3. Design & Methods

• Study Type: Institutionally sponsored, multicenter, open-label, parallel-group, randomized superiority trial
• Setting & Centers: 10 ICUs, France
• Population:
• Inclusion: Adults ventilated <6h at screening, expected to need ventilation ≥48h, eligible for restraint prescription (not already restrained, no pre-existing prescription)
• Exclusions: Delirium before admission, dementia, anticipated alcohol withdrawal, primary neurological admission, major visual/auditory impairment, pregnancy/lactation, SAPS II >65, DNR order
• Intervention: Restrictive strategy — 2-point wrist straps only for severe agitation (RASS ≥3)
• Comparator: Liberal strategy — wrist straps applied systematically from day 0, reassessed daily
• Blinding: Open-label; primary outcome depended on structured RASS/CAM-ICU assessments (partly standardized, not fully blinded)
• Statistical Power & Follow-Up: 422 patients needed to detect 1-day increase in delirium/coma-free days (SD 3 days, 90% power). Primary endpoint at 14 days; additional outcomes to day 90.

4. Key Results

405 randomized; 396 in primary analysis (196 low-use, 200 high-use).

Sedative use: No compensatory increase in chemical sedation with restrictive restraint (sufentanil, propofol, midazolam, antipsychotic, dexmedetomidine doses all similar between arms).

5. Internal Validity Assessment

• Randomization & Allocation: Central, computer-generated, variable block sizes, stratified by site/age/coma status — sound concealment.
• Protocol Adherence & Separation: Substantial nonadherence in both directions: 33.7% of low-use patients restrained ≥4 consecutive hours without severe agitation; 28.5% of high-use patients had restraints removed outside protocol — yet actual restraint-time separation remained large (0h vs 16.8h median).
• Missing Data: 9/405 excluded post-randomization; 14.1% had ≥1 missing primary-outcome assessment; only 180/396 (45%) completed formal day-90 follow-up.
• Blinding & Detection Bias: Open-label; bedside nurses knew allocation, a genuine concern since the primary outcome depends on RASS/CAM-ICU assessment rather than a hard endpoint.
• Overall Internal Validity Conclusion: Moderate — good randomization and real treatment separation, but open-label assessment, sample size below target (396 vs 422 planned), substantial nonadherence, and incomplete long-term follow-up limit confidence in ruling out smaller true effects. Weighted per-protocol analysis (1.07d, 95% CI −0.12 to 2.19, P=0.055) trended toward benefit while standard per-protocol trended oppositely, illustrating how contamination complicated causal inference.

6. External Validity Assessment

• Population Representativeness: General adult medical ICU population, respiratory failure predominant, 87-88% comatose at randomization — less transferable to more awake, behaviorally active agitation populations where restraint decisions are most debated.
• Practice Context: Participating centers were relatively restraint-intensive at baseline (75-100% restraint at ventilation initiation); routine delirium-scale use existed in only 5/10 centers.
• Overall External Validity Conclusion: Moderate — highly relevant to general ICUs with similar restraint practices and staffing, much less generalizable to units already using restraint sparingly or to high-risk agitation populations (neurological, alcohol withdrawal, dementia) that were excluded.

7. Strengths & Limitations

Strengths:

• First multicenter RCT directly comparing restraint strategies in ventilated ICU patients
• Real, substantial restraint-time separation despite protocol nonadherence
• Protocolized co-interventions (sedation, delirium screening, weaning, mobilization) in both arms
• No compensatory chemical-sedation escalation with restrictive strategy

Limitations:

• Open-label, unblinded primary-outcome assessment
• Sample size below target (396 vs 422)
• Substantial bidirectional nonadherence
• ~87% comatose at randomization, potentially diluting any delirium-specific effect
• Incomplete day-90 follow-up (45%)
• High-risk agitation populations excluded

8. Interpretation & Practice Impact

• Clinical Implications: Does not support routine prophylactic wrist restraint to improve delirium/coma outcomes; reducing restraint use produced no penalty in self-extubation, device removal, pressure injury, or mortality.
• Mechanistic Coherence: The safety-neutral result (no more self-extubation with less restraint) is the trial's most clinically useful and actionable signal.
• Systems-Level Takeaway: Clinicians in similar ICUs can reasonably move away from universal prophylactic restraint toward reserving it for clearly justified, short-duration severe agitation.

9. Controversies & Subsequent Evidence

• Editorial Commentary: Accompanying JAMA editorial (Haines, Nørregaard Kjær, Deane) notes the trial tested a predominantly deeply-comatose population, making it less informative for the agitation-prone patients in whom restraint is most often defended, and flags the unblinded primary outcome as a key limitation.
• Guideline Integration: No formal guideline update yet incorporates R2D2-ICU; the PADIS framework remains dominant. The result contradicts a 2025 meta-analysis of guide-based restraint-reduction interventions that had suggested lower delirium and unplanned extubation with restriction — R2D2-ICU, as a more direct head-to-head test, tempers that optimism.

10. Summary & Executive Takeaway

Summary: R2D2-ICU randomized 405 mechanically ventilated French ICU patients to restrictive (low-use) or liberal (high-use) wrist restraint. Days alive without delirium/coma at 14 days were not significantly different (6.67 vs 6.30 days, P=0.51), and self-extubation, device removal, pressure ulcers, and mortality were all similar between arms.

Overall Takeaway: Routine liberal wrist restraint confers no measurable delirium, coma, or safety advantage over a restrictive strategy in mechanically ventilated ICU patients — a landmark negative trial pushing practice toward restraint minimization, though its applicability to the most agitation-prone patients (largely excluded here) remains untested.

11. Bibliography

• Berger S, Grzonka P, Amacher SA, et al. Adverse events related to physical restraint use in ICUs. J Intensive Med. 2023;4(3):318-325.
• Haines KJ, Nørregaard Kjær MB, Deane AM. Physical restraint use in critical care: balancing safety and harm [editorial]. JAMA. 2026.