Because we are interested in comparing the effectiveness, rather than the efficacy, of the two training methods, we are using a pragmatic randomized controlled trial (RCT) design. The study setting is a large mental health hospital in Ontario, Canada, which delivers inpatient, outpatient, and emergency health care across a wide range of patient populations.

Newly hired staff that are classified as clinical direct service staff must undergo 2 weeks of onboarding, which includes the trauma-informed de-escalation education for safety and self-protection (TIDES) program. Within those 2 weeks, physical skills training is a full-day, in-person session scheduled on the last day. Because the onboarding is tightly scheduled, there is limited time for the typical processes of providing information about the study, gaining agreement and informed consent, and randomization. Consequently, the following procedures have been introduced.

Prior to the 2-week onboarding training, all newly hired clinical direct-service staff will be randomly assigned to their physical skills training session, and the physical skills training sessions will be randomly assigned to either BST or TAU. On the day before the physical skills session, the study will be introduced by a research team member to all attendees at the end of the last session on the previous day; during that introduction, consent will be obtained via a WebEx poll to send interested attendees a copy of the informed consent, and all attendees will be informed that a question-and-answer session along with obtaining informed consent will be carried out before the start of the physical skills session the next day, and then the informed consent form will be emailed to those who provided permission.

Just before the physical skills session, questions about the study or informed consent will be answered, and one-on-one meetings will be conducted with each attendee in a separate room to further review the informed consent and to sign indicating whether they agree or decline to participate in the study. This procedure is designed to protect against attendees identifying who is involved in the study (hereafter “study participant”) and who is not (hereafter “trainee”).

This study was internally funded and has been approved by the CAMH Research Ethics Board (#101-2020).

The study processes are designed to ensure that all attendees receive the same attention from the research team and facilitators and the same assessments during the training sessions. The goals are to provide equivalent training experiences, regardless of whether the person is a study participant or a trainee and regardless of whether they are in a TAU or BST session, as well as to minimize the possibility that attendees can identify who among them are or are not study participants. For example, in gaining informed consent, all attendees meet one-on-one privately with a research team member, and all sign the consent form by indicating whether they agree or decline to be study participants. In addition, skill assessments for all attendees are done individually in a separate room, but only the study participants (ie, those providing consent) are videotaped.

Potential study participants include all newly hired clinical staff attending the mandatory training. There are no exclusion criteria (except the attendee’s desire not to be a study participant).

There is no published information on the expected effect size of BST versus other training methods. Indeed, most of the published BST studies have focused on efficacy and consequently have involved relatively small sample sizes. Consequently, we have opted for a sample size of 80 participants total (ie, 40 each in the BST and TAU groups) consistent with sample sizes providing 80% power for an expected medium to large effect size [25].

Recruitment began in January 2021, after REB approval was received. We reached our recruitment goal in September 2021.

Both the TAU and BST methods teach the same skills for protecting a person against aggression (eg, someone attempting to punch or choke the staff member) as well as to help a team of staff members physically restrain a patient who is becoming an increased risk of harm to self or others. There are 11 target skills (6 for self-protection and 5 for team-control) that are mandatory for all newly hired staff (Textbox 1). Each skill has a number of defined components, and the same sequence of steps is followed to teach each component. However, the 2 training methods differ in how these steps are administered and monitored (Table 1). The BST protocol involves attendees actively performing the targeted skills through a loop involving instruction, modeling, practice, and feedback using competency checklists [20]. This is continued until the competency of the target skill is demonstrated based on the observation of the trainer [20]. While the common practice as described in the literature is for the attendee to demonstrate successful performance of a skill 2 to 3 times before moving to the next skill on the checklist [26], we have chosen a more stringent threshold of requiring correct execution 5 times consecutively with the expectation that this will further consolidate skill acquisition and possibly retention. In contrast, TAU does not assess competency or require trainees to reach a specific level of competence for a skill before moving on to the next skill. It does, however, include modeling, practice, and feedback.

Each training session is run by 2 facilitators, each training half of the attendees to make the most efficient use of the allotted time. To be compliant with COVID-19 restrictions at the time of the study on the numbers of in-person attendees (maximum of 10 individuals including the trainer), 1 trainer is present at the session, while the other delivers the same material virtually.

The variables to be assessed include skill competence, mastery, and retention; participant confidence in using each skill; the frequency of skill use in the previous month; and overall satisfaction with the training. Skill competence, mastery, and retention will be evaluated using checklists which were created guided by the considerations proposed by Stufflebeam [27] as well as through consultations with 2 coauthors who are board-certified behavior analysts (KB and LB). These are designed to evaluate the 11 mandatory skills being taught in the TIDES training program. The checklist creation process involved the following steps: (1) observing the training experts performing the skills and consulting with them; (2) having the organizational experts and trainers define and agree to the written descriptions in the training curriculum of the mandatory 11 target skills; (3) presenting the checklist drafts to key organizational stakeholders (professional practice office, union representatives, staff coaches, and executive leadership); (4) field-testing the resulting checklists in the mandatory staff training and making appropriate revisions; (5) and finally, making final edits based on the results of the interobserver agreement (IOA) assessment (please see further description of IOA below).

The questions assessing participant confidence, the frequency of skill use in the previous month, and overall satisfaction with the training are either adapted from existing questions used in the research team’s department or developed specifically for this study. Where applicable, 10-point Likert scales were used to improve accuracy.

The checklists (see self-protection checklist example, Multimedia Appendix 1) will be used during training for the BST sessions and for the baseline, posttraining, and 1-month study assessments for all study participants. On the training day, attendees will be asked individually to go to a separate room. They will be asked once to demonstrate each of the 11 skills prior to the start of training for baseline and just after the training for posttraining. Only those attendees who have consented to be study participants will have their baseline and posttraining assessments videotaped. For the 1-month follow-up, study participants will be asked to come in to have their skills assessed and videotaped. Videotaping will be done through WebEx (Cisco) and will be rated by a research team member using the predefined competency checklists. On these checklists, each skill component will be judged to be performed “correctly” or “incorrectly.” The final competency rating for each skill will be the percent of its component skills that were rated as correctly performed.

The reliability and validity of these ratings will be assessed using IOA calculations [28]. Prior to the study, raters had been trained on a test set of videotapes drawn from a previous study and had reached agreement levels of at least 90% for self-protection and team-control skills. For this study, all videotapes are rated by a primary rater. At the same time, a 30% random sample of videotapes from training months 1, 3, 4, 6, and 9 will be independently rated by a designated secondary (reliability) rater. IOA with the ratings from the primary rater will be calculated after every 2 videotapes rated by the secondary rater and, if the IOA is less than 90%, a recalibration discussion is held and recorded between the 2 raters with a third neutral rater available if needed to resolve any remaining differences. Recalibration decisions are then applied to all subsequent videotape ratings.

A predetermined threshold of 80%, consistent with clinical practice [29,30], is then applied to each competency rating to define skill mastery. Retention between posttraining and the 1-month follow up is measured in terms of whether the competency ratings changed as well as whether the percentage of study participants meeting the mastery threshold has changed.

Study participants complete self-reported evaluation forms covering descriptive information, confidence, training satisfaction, and frequency of skill use. The participants also provide information regarding their professional role and service department at CAMH. For confidentiality reasons, personal characteristics such as age and gender or sex were not provided to the research team, and, for the same reasons, we chose not to collect this information. All participants are asked to evaluate their confidence on a 0-10–point scale (0=not at all confident to 10=extremely confident) in using the 11 self-protection and team-control skills at baseline, immediately after the training, and at 1-month follow-up (Multimedia Appendix 2). All participants were asked to rate for satisfaction with the training on a 4-point scale immediately after the training and 1 month later. The number of times in the past month that self-protection or team-control skills were used is asked at baseline and 1-month posttraining. Finally, an open-ended item is provided in the posttraining assessment form for any comments that trainees want to communicate to the facilitators.

Data from the competency checklists, completed by raters on the research team, and data from the self-reported evaluation completed by participants are collected via REDCap, which is CAMH’s designated web-based application for research data capture and then exported to Excel (Microsoft Corp) spreadsheets [31]. After data cleaning, these spreadsheets will be imported to a statistical software package (eg, SPSS, R) for analysis.

Testing of the 3 hypotheses will be done using repeated measures ANOVA. In addition, the study participants will be compared to all attendees in terms of their role and department to evaluate whether there are any striking differences despite our randomization methods. We also plan to assess the association between the frequency of physical skills used in the past month and our measures of skill competence, mastery, retention, confidence, and satisfaction.