Editorial 

Physical Restraint Use in Critical Care: Balancing Safety and Harm

Kimberley J. Haines, Maj-Brit Nørregaard Kjær, Adam M. Deane

JAMA Published Online: March 17, 2026

doi: 10.1001/jama.2026.2785

I was already fighting so much, not only for my life, but just like the need for a drop of water and any basic human need was being taken away anyway and then not to have the use of your bodily parts, your hands to hug somebody, to wipe away tears, to help you talk.

Perez et al¹

Critical care is a place where safe patient care is constantly balanced against the risks of harm. Some patients are sufficiently agitated that providing life-sustaining care requires a form of restraint to deliver that care and often to protect the patient from harming themselves or those who are delivering that care. Restraints may be chemical, physical, or a combination of both. As the epigraph highlights, being physically restrained can be a highly distressing experience for patients.¹ Although hospitals have a duty of care to provide a safe workplace—and there may be specific scenarios where physical restraints are necessary to limit or prevent aggression and violence toward health care workers^2,3—indiscriminate application of physical restraints has the potential for harm.⁴Damage may occur to (1) the patient, including pressure, musculoskeletal, and vascular injury; delirium; posttraumatic stress disorder (PTSD)⁵; anxiety; depression; and reduced health-related quality of life; (2) family members, including distress and traumatic memories; and (3) health care workers, including the moral distress from conflict with the use of physical restraints and the ethical principles of beneficence and nonmaleficence.^6,7

Despite the potential for harm if used indiscriminately, there are sparse data to inform the use of physical restraints on patients in the intensive care unit (ICU).⁸ Inadequate data may explain why the use of physical restraints varies between ICUs, from rare occurrences to all patients receiving mechanical ventilation.^9,10

It is within this context that Sonneville and colleagues conducted the Restrictive Use of Restraints and Delirium Duration in ICU (R2D2-ICU) trial, reported in this issue of JAMA.¹¹ R2D2-ICU was an open-label, multicenter randomized clinical trial conducted in 10 ICUs in France to evaluate the effect of a restrictive wrist-strap physical restraint strategy in patients receiving mechanical ventilation in the ICU. A total of 405 participants were randomized in a 1:1 ratio to the low-use intervention, which avoided wrist-strap use unless necessary because of severe agitation, or a comparator of usual care or high use, with systematic application of wrist straps and daily reassessment. Restraints were permitted to be discontinued in participants in the comparator group who were awake or extubated in the absence of delirium.

The authors report no statistically significant between-group differences across multiple outcomes. The primary outcome of the mean number of days alive without coma or delirium was 6.67 (95% CI, 5.69-7.65) in the low-use group and 6.30 (95% CI, 5.35-7.24) in the high-use group (adjusted mean difference, 0.37 days [95% CI, −0.71 to 1.46]). The secondary outcome of delirium occurred in 122 of 196 patients (62.2%) and 135 of 200 (67.5%) in the low- and high-use groups, respectively (adjusted risk ratio, 0.92 [95% CI, 0.80-1.07]). Self-extubation occurred in 9.2% of patients in the low-use group and 8.5% in the high-use group, and pressure ulcers were reported in 15.3% of patients in the low-use group and 17.0% in the high-use group. Hospital mortality was 37.2% in the low-use group and 41.0% in the high-use group. At 90-day follow-up, cognition, physical function, and PTSD outcomes were similar between groups.

The trial makes important advancements in a field where rigorous data are sparse and clinical practice is guided by dogma. Despite no difference in the measured outcomes between low use of physical restraints and usual care for the measured outcomes over the course of the study, there is potential for long-term harm from physical restraints that must be considered. These harms include PTSD, depression, anxiety, and reduced health-related quality of life, but it may be difficult for clinicians to detect such long-term harms at the bedside.

This trial was carefully designed and well conducted. However, there are several caveats and limitations that warrant consideration. This trial excluded patients from certain diagnostic groups or because of coexisting conditions (eg, neurological conditions, alcohol withdrawal, delirium on admission). It was unclear how patients with substantial coexisting illicit drug use or withdrawal were handled. These high-risk groups often experience psychomotor agitation and hyperactive delirium that can be challenging to manage and risk harm to health care workers, especially nursing and clinical assistants who spend more time at the bedside. Excluding these patients from this trial limits its generalizability to some of the most challenging clinical encounters. Furthermore, at randomization, almost 90% of the participants were deeply unconscious, with a Richmond Agitation-Sedation Scale score of −4 or −5. Thus the results of the R2D2-ICU trial may not generalize to jurisdictions where physical restraints are only used in those at greatest risk of hyperactive delirium who are emerging from sedation. The trial does provide insights into self-extubation risk, which has been commonly cited as a key reason for using physical restraints to enhance patient safety.⁸ The trial reported no between-group differences in self-removal of the endotracheal tube, which indicates that it may be safe to use less physical restraint, although the trial was not powered to detect a statistically significant difference in this outcome.

There were challenges in implementing the intervention. Although there was wide treatment separation (restraints were applied in 100% of participants in the high-use group for a median time of 14.4 hours per day), restraints were still applied to 36% of participants in the low-use group. In some countries, the use of physical restraints in 1 in 3 patients would be more representative of usual care than a reduction in the use of restraints. We speculate that countries with better nurse to patient ratios afford greater surveillance with immediate, but gentle, intervention when it appears that a tube or catheter is about to be removed, rather than mandating a preventive strategy of physical restraints. The nurse to patient ratios for patients receiving mechanical ventilation at trial sites were 1:2.5 and 1:3. Accordingly, the results from the R2D2-ICU trial may not generalize to regions with better nurse to patient ratios and where physical restraints are not routinely used.^12,13 In regions where a nurse is routinely responsible for more than 1 patient, the R2D2-ICU trial data results provide considerable confidence that routine restraint use in all patients can be discontinued.

The authors should be commended for the trial conduct. First, the reporting of the coadministered medications was rigorous, with no between-group differences in the use of chemical restraints (opioids, sedatives, or antipsychotics). Second, the authors comprehensively reported relevant outcomes, with extended follow-up to 90 days, including valuable outcomes, such as physical function, cognition, PTSD, pressure injuries, and other injuries. However, there was a high loss to follow-up rate for the 90-day assessment—only 180 of the 405 participants (94 of 196 in the low-use group; 86 of 200 in the high-use group) completed the 90-day assessment. It was unclear in the methods and trial processes why blinding of the outcomes was not possible or whether the trial statistician(s) were blinded to group allocation, which does risk bias.

Currently, clinicians are left to make difficult choices at the bedside, faced with choosing between undesirable options. The R2D2-ICU trial prompts us to examine our practices and consider that new and innovative approaches need to be developed and tested that better align with community expectations regarding safe, compassionate, and person-centered care.¹⁴

In summary, the results of this multicenter randomized clinical trial indicate that physical restraints should not be routinely used in all patients receiving mechanical ventilation during intensive care. If ICUs reduce their use of physical restraints, we recommend that the use of sedative drugs is monitored. Although the use of chemical restraints may be more palatable to families and health care workers, we cannot assume that an increase in chemical restraint will not cause unintended physical or psychological harm to the patient that is worse than physical restraints.¹⁵ Clinicians will always need to weigh safety with the risk of harm when considering use of physical restraints, and the application of the R2D2-ICU trial results to clinical practice should be carefully considered based on local practice and nurse to patient ratios. Finally, greater advocacy is needed for research to identify and test better approaches to restraint management in patients who are critically ill.