Using the evidence to decision frameworks to formulate the direction and strength of recommendations for adapted guidelines of physical restraints in critical care: A Delphi study

It has been shown that the use of physical restraints has negative physiological and psychological impacts on critically ill patients. The development of clinical practice guidelines is necessary for minimizing the use of physical restraints.

As part of a guideline adaptation project, this study aimed to formulate the direction and strength of recommendations for adapted guidelines of physical restraints in critical care.

Literature retrieval of guidelines and systematic reviews was performed and guideline steering committee discussions were conducted to develop an inquiry questionnaire. From February to April 2022, a two-round study, including 27 multidisciplinary experts, was conducted using the evidence to decision frameworks and Delphi methods.

The 27 experts were from 22 of the 33 provinces/municipalities/autonomous regions of mainland China. For the first and second questionnaires, the recovery rates were 85.7% and 100%, respectively. In the first and second rounds, the average authoritative coefficients were 0.93 and 0.94, respectively. Regarding Kendall W values, the values were 0.555 and 0.120, respectively. A consensus was reached on 15 recommendations, including one strong recommendation, seven conditional recommendations and seven good practice statements.

In the two rounds of consultations, a consensus was reached on 15 recommendations for the adapted physical restraint guidelines in critical care through the evidence to decision frameworks and Delphi methods. Recommendations in the adapted physical restraint guidelines are related to institutional and educational interventions, risk assessment, the management of patients' unsafe behaviours, pain, mechanical ventilation, sleep, family engagement, mobility, sedation, and delirium.

Based on our recommendations, we suggest nurses develop a physical restraint education bundle.