Case No.: 2017-02
Brink’s Canada Ltd., Appellant
Michael Childs and Unifor, Respondents
Case No.: 2017-12
Kevin Barber, Appellant
Brink’s Canada Ltd., Respondent
Indexed as: Brink’s Canada Ltd. v. Childs and Unifor
Matters: Appeal under subsection 146(1) of a direction issued by an official delegated by the Minister of Labour (2017-02);
Appeal under subsection 129(7) of a decision issued by an official delegated by the Minister of Labour (2017-12)
Decision: The appeal in File 2017-02 is upheld and the direction is rescinded
The appeal in File 2017-12 is dismissed and the decision of absence of danger is upheld
Decision rendered by: Mr. Pierre Hamel, Appeals Officer
Language of the decision: English
For the appellants (2017-02)/respondents (2017-12): Mr. James D. Henderson, Counsel, Grosman Gale Fletcher Hopkins LLP
Mr. Gregory J. Heywood, Roper Greyell LLP
For the respondents (2017-02)/appellants (2017-12): Ms. Niki Lundquist, Associate Counsel, Unifor Legal Department
Citation: 2018 OHSTC
 These reasons concern two appeals relating to the alleged unsafe nature of the All Off (A/O) delivery model used by Brink’s Canada Ltd. (“Brink’s” or “the employer”). The first appeal (the Childs appeal) is an appeal by Brink’s against a direction issued by Mr. Lewis Jenkins, in his capacity as official delegated by the Minister of Labour (ministerial delegate) on December 29, 2016, in Ottawa, further to a work refusal made by Mr. Childs, an employee of Brink’s. The direction was issued against Brink’s pursuant to subsection 145(2) of the Canada Labour Code (Code) further to Mr. Jenkins’ finding of danger.
 The second appeal (the Barber appeal) is an appeal filed by Mr. Kevin Barber against a decision of absence of danger rendered by Mr. Jenkins on April 28, 2017.
 Both appeals arise out of work refusals made by the two employees concerned. The refusals were related to the implementation of the 2-person crew delivery model referred to as the “A/O” model. As the appeals raised similar issues, they were joined for a common hearing. The evidence adduced at the hearing related to circumstances that were specific to each appeal as well as to generic information common to both appeals. The totality of that evidence was made part of the record for both appeals. The two refusing employees were represented by Unifor (or “the Union”) throughout the proceedings.
 It should be mentioned that, while these appeals were pending before the Tribunal, I was seized of an appeal previously filed with the Tribunal, related to a direction issued following a work refusal by [text redacted], an employee of Brink’s working at its Edmonton Branch, on essentially the same grounds, i.e. the unsafe nature of Brink’s A/O model. I issued my decision in Brink’s Canada Ltd. v. Dendura, 2017 OHSTC 9 (Dendura), on June 16, 2017, and rescinded the direction on the basis that the hazards and risks presented by Brink’s A/O model were normal conditions of employment. The Dendura decision is obviously central to the disposition of the issues raised in the present cases.
 In fact, the employer sought to obtain a summary decision on the present appeals, without a hearing, based on my findings and conclusions in Dendura and on the fact that Mr. Childs had, in the meantime, withdrawn his work refusal and now considered the A/O model to be safe. I dismissed the application and the reasons for my decision are set out in Brink’s Canada Ltd v. Childs and Unifor, 2017 OHSTC 18.
 At the parties’ request, the hearing was held in camera, in light of the nature of the matter at issue and the sensitivity of the information discussed at the hearing. The parties jointly signed a Confidentiality Agreement relating to the conduct of the present proceedings and the protection of the Tribunal’s record from disclosure to the public. By order dated January 9, 2018, I endorsed the parties’ agreement. The grounds on which the Confidentiality Order was sought and obtained are identical to the grounds set out in my reasons in Dendura, at paragraphs 10 to 13, and form the basis of my decision to endorse the Confidentiality Agreement of the parties in the present cases. The Confidentiality Order is appended to the present reasons.
 I will deal with each appeal consecutively.
 The issues in the present appeals may be described as follows:
- Were employees Childs and Barber exposed to a danger as defined under the Code when they exercised their right to refuse unsafe work; and
- If a danger existed, was the danger a normal condition of employment, under paragraph 128(2)(b) of the Code?
The Childs appeal (2017-02)
 On November 10, 2016, Mr. Jenkins conducted an investigation into a work refusal initiated by Mr. Michael Childs. Mr. Childs is an employee of Brink’s’ armoured car services with the Ottawa Branch of the employer. The refusal took place at the CIBC Hawkesbury stop of the run on which Mr. Childs was assigned (Run 78). Mr. Childs had bid to work this run.
 After inquiring into the circumstances of the refusal, Mr. Jenkins informed the employer, on December 29, 2016, of his finding of danger and of his direction, issued on the same day. The direction reads as follows:
In the matter of the Canada Labour Code
Part II – Occupational Health and Safety
Direction to the employer under the subsection 145(2)(a)
On November 10, 2016, the undersigned official delegated by the Minister of Labour conducted an investigation following a refusal to work made by Michael Childs in the work place operated by Brink's Canada Limited, being an employer subject to the Canada Labour Code, Part II, at 2755 Lancaster Road, Ottawa, Ontario, K1B 4V8, the said work place being sometimes known as Brink’s Canada Limited.
The said official delegated by the Minister of Labour considers that the performance of an activity constitutes a danger to an employee while at work:
The “All Off” model that is currently being used (in which the driver/guard exits the armoured vehicle and escorts the messenger carrying the valuables, into customer locations for drop-offs and pick-ups) does not sufficiently mitigate against the danger of employees being assaulted during a robbery attempt. The model does not provide the employees with any information of suspicious persons or activities occurring outside while they are inside the customer’s location. As a result, the employees have a diminished ability to avoid potential ambush upon returning to the armoured vehicle.
Although the direction issued and the measures identified in the appeals officer’s stay were in regards to the Edmonton location, the All Off model is being implemented throughout its organization. At the Ottawa location, the employer has failed to apply the five measures identified in the appeals officer’s stay decision (2016-34) of October 26, 2016 in which Appeals Officer Olivier Bellavigna-Ladoux agreed that those measures would serve to protect the employees’ health and safety while performing the “All Off” model that is currently being used pending the decision of the appeal for the direction issued at the Edmonton location.
Therefore, you are hereby directed, pursuant to paragraph 145(2)(a) of the Canada Labour Code, Part II, to alter the activity that constitutes the danger immediately.
Issued at Ottawa, this 29th day of December, 2016.
Official Delegated by the Minister of Labour
 Described succinctly at this point, the A/O delivery model consists of a 2-person crew where both crew members exit the armoured car and enter customer locations to drop-off or pick-up valuables. After the completion of the work, both crew members return to the vehicle together, following specific protocols developed for that delivery model.
 On March 3, 2017, the appellant’s application to obtain a stay of the direction until the final decision on the appeal was granted, for the reasons set out in Brink’s Canada Ltd v. Childs and Unifor, 2017 OHSTC 4. I will refer to the stay in more detail later in the present reasons because of its particular relevance to the circumstances of the Barber appeal.
Summary of the evidence presented at the hearing
 The employer called a number of witnesses who testified on the A/O model generally, without necessarily having been directly involved in the events leading to the work refusal.
 Mr. Pierre Brien testified as an expert witness for the employer. He is a security consultant with the TRAK Group, who was retained by Brink’s to assess the A/O model in terms of its safety for employees. The TRAK Group is a security consulting firm specialized in corporate security, risk assessment and risk management. Mr. Brien had a 30-year career in law enforcement during which he held strategic operational and administrative functions at the municipal and provincial levels. During his career, Mr. Brien was involved with national issues on crime prevention, intelligence and police management with Criminal Intelligence Service Canada (CISC) as Quebec’s bureau director. Since retiring in 2007, he has been retained by national and international clients to address security concerns in the fields of risk mitigation, emergency planning, airport security (Erbil International Airport in Iraq), and crisis management, community policing and training. From 2012 to 2014, he acted as project director for the Haitian National Police Academy, on behalf of the Government of Canada. He performed threat and risk assessments for 60 sites for the benefit of Montreal’s Transit Authority and proposed and implemented security concepts and procedures. He was a professor at the Canadian Critical Infrastructure Institute (2008). He recently completed a guide to community policing for Francopol, the international French-speaking police training organization. This guide is destined to OIF (Organisation internationale de la francophonie) member countries involved in democratic reinforcement or post-conflict reconstruction.
 Based on these credentials and his combination of work experience and education, I accepted Mr. Brien’s qualification as an expert in policing and risk assessment and mitigation.
 Mr. Brien introduced and spoke to the two TRAK reports (TRAK 2015 and TRAK 2017) prepared as a result of a Brink’s request for their opinion and recommendations regarding its A/O model. The need to consider the A/O model by Brink’s’ management was driven by the increasing competitive business environment for the cash-in-transit (CIT) industry. This new crew configuration raised health and safety concerns as well as a competitive edge for Brink’s’ competitors. The objectives of the research were to: assess the specific risks of injuries resulting from robbery or attempted robbery when armoured vehicle crews leave the vehicle to provide service and return to the vehicle; assess whether the A/O model constitutes an acceptable risk under the Code; and formulate recommendations regarding all facets of the job hazard analysis of the A/O model.
 Mr. Brien testified that he applied the Institute’s critical infrastructure protection model in his analysis of the risk, which calls on an examination of historical scenarios (past incidents and indicators, replication of offenders’ thinking), statistical base (collection of incidents by Brink’s corporate security) and intelligence base (police and other sources). The risk is assessed by reviewing the probability and impact of events taking place in reality, which in turn allows the identification of the right course of action to mitigate. He also reviewed international literature relating to risks in the CIT industry, which he accessed through a professor of criminology at Georgetown University utilizing his network and research tools for the CIT industry. The objective is to get an understanding of how criminals make their decision and decide on in which situations to act, in order to maximize the gain and minimize the risk of being arrested or killed.
 Noting that the CIT industry involves a somewhat high level of inherent danger particular to the very nature of the business, Mr. Brien believes that the A/O model would likely present an unacceptable danger to employees, unless a number of measures were to be implemented for the purpose of mitigating such danger. Mr. Brien examined decisions rendered under the Code, in particular the Brazeau et al. v. CAW-Canada and Securicor Canada Ltd, Decision No. 04-049 (December 16, 2004) (Securicor) and suggested that the employer must address the following areas with a view to minimizing the risk associated with the A/O model: condition of the location, time of day, visual contact, communications system, distance covered by the crew, operational procedure, training and supervision of employees and consultation with health and safety committees. Mr. Brien referred to the notion of there being several layers of protective measures that criminals would have to circumvent in order to be successful, so as to make it less and less appealing for criminals to launch an attack.
 In Mr. Brien’s view, the overview of the literature does not provide any evidence that the number of armed guards or their specific role have a significant impact in explaining the frequency, severity and patterns of attacks.
 Mr. Brien referred to the statistics utilized for his analysis. The reference period was 2000 to 2015. The data over that period reveals that 99 robbery attempts were made across the industry in the country. Of those, 69 (70%) were made against 2-crew member configurations and two A/O crews. Seventeen of those 99 attempts (17%) were made against Brink’s crews during that period, although the company had not deployed the A/O crew configuration. This might indicate a preference on the part of criminals for the A/O configuration when planning an attack. Turning to the severity of injury, Mr. Brien referred to the number of deaths and serious injuries reported during that period: three fatalities and three major injuries. However, given that the three fatalities and one major injury were sustained by armoured vehicle employees at the hand of a co‑worker in a single incident Edmonton in 2012, Mr. Brien concluded that the present safety situation in Canada could be considered as an acceptable risk, in comparison with other types of occupations.
 The risk of injury or fatality resulting from robbery or attempted robbery when an armoured vehicle crew leaves the vehicle and then returns to the vehicle remains difficult to determine with mathematical precision. Based on his analysis of available data and the industry’s track record, and applying his knowledge of risk management theories, Mr. Brien concluded that this risk is low. Mr. Brien considers that, by applying the mitigation measures it developed and continuing to move forward on its security, operational and health and safety initiatives, including the implementation of the recommendations set out in his report, Brink’s could demonstrate that the remaining risk is inherent to the work of employees.
 The 15 recommendations are as follows:
- Increasing efficient supervision of staff to ensure compliance with safety protocols
- Developing an Employee Operations Manual including a section on A/O crew configuration
- Sharing street inspections with crew members
- Maintaining and updating a database of all incidents and using that information to analyze trends and determine risks
- Making employees aware of the level of risks at serviced locations (site risks assessments (SRAs), intelligence reports) and training them to act according to the operational procedures
- [Text redacted]
- Lighting fixtures on armoured cars
- Efficient communications system between crew members
- Specific training for A/O crew configuration
- Formal training program for new employees: to ensure that the Learning in Field Experience (LIFE) program be more consistently implemented in all regions, a functional responsibility should be assigned to the Training and Development Manager
- [Text redacted]
- Development of formal SRAs for all locations
- Assessors for the SRA program to be properly trained to perform assessments efficiently and reliably
- Development of a formal crime prevention program (intelligence gathering, information bulletins to employees, street inspections, observations, supervision)
- Review of incidents protocols to ensure that roles and responsibilities of the National Control Centre (NCC), management and employees are clearly defined and adapted to the new A/O crew operational procedures
 Mr. Brien prepared a PowerPoint presentation and delivered the presentation to both National Health and Safety Policy Committees (NHSPC) (Teamsters and Unifor). Mr. Brien testified with respect to his second report (TRAK 2017). He was tasked with reviewing the implementation of the 2015 report and formulating any recommendations as appropriate. He noted that out of his 15 recommendations, Brink’s had not implemented # 7 (the light fixtures on trucks was deferred at the request of the Teamsters Policy Committee) and # 10 (LIFE training responsibility matrix) and that two others (supervision and incident protocol management) were deemed in progress. The other recommendations were implemented and he confirmed the accuracy of his estimation of the threat levels. He noted that there had been no attacks for two years on Brink’s A/O crews, and that there had been only six attacks against other companies, all against 2-person crews, which did not surprise him since the majority of other companies’ runs are operated with 2-person crews. He specified however that a 5-year period would be preferable in determining the efficiency of the risk reduction measures.
 Mr. Brien stated that TRAK is an independent firm that is used to dealing with many different clients and industries. While they are open to comments and suggestions from clients, they do not let themselves be influenced by clients’ suggestions and would never change the substance of a conclusion to make it more acceptable or favourable to the client. In the present case, [text redacted] clearly wanted an independent view. Mr. Brien added that it was not his mandate to compare the A/O model with the 3-person crew and was adamant that he found no data showing that the number of crew members may be a determining factor in a criminal’s decision to launch an attack.
 He agrees that the frequency of malfunctions of the equipment is an important piece of information that the employer should consider, and act upon. He does not consider this situation to be critical, except if the employer does nothing about it, which, in his view, is not the case here. Brink’s does take corrective measures when equipment failures have been reported. He also agrees that SRAs should be reviewed and updated, and that if something changed materially in the location, a reassessment would be mandated.
 Mr. Brien did not disagree that [text redacted] or call for assistance, depending on the circumstances. However, a third person in the truck could not intercede if there was an attack.
 [Text redacted].
 The employer called Mr. Derek Doiron. Mr. Doiron is secretary treasurer for the Teamsters in the Maritimes and their assistant director, Armoured Car Industry. He was co-chair of the National Health and Safety Policy Committee for the Teamsters, non-union and the Newfoundland and Labrador Association of Public and Private Employees (NAPE) dealing with the implementation of the A/O model (the “Teamsters Policy Committee”). Mr. Doiron testified that the Teamsters membership is approximately 3,000 members in the armoured car industry, while Unifor counts approximately 1,200 members. Mr. Doiron was summoned to testify at the hearing.
 He testified that he was involved in an appeal dealing with a “One Off” model that Brink’s had introduced in the Maritimes, and that during the course of those proceedings, an agreement was reached with Brink’s that the company would develop a better A/O model than the one competitors currently operated in the industry. He testified that he had personal knowledge of the A/O model utilized by Garda, as he had worked under it between 2004-2009 as a driver/guard and messenger/crew chief. He acknowledged that Brink’s needed to make changes in order to compete and that Securicor had been utilizing an A/O model since approximately 2004.
 Mr. Doiron stated that the Teamsters Policy Committee fully participated in the implementation of the A/O model. He agreed with Brink’s to have two separate policy committees to review the A/O model. As he explained in a letter to a Brink’s executive, Unifor and Teamsters represent over 90 % of the industry in Canada. Both unions have members running an A/O crewing model. Unifor vocally opposes this model, while the Teamsters recognizes that with working through the national policy committee, this model’s risk can be minimized. He stated that the Teamsters also recognize that the industry is out of balance with respect to crewing levels and has a responsibility to ensure there is a level playing field. He felt that one single committee would not work, as Unifor would “never accept” the A/O model.
 Through their policy committee, the Teamsters provided input and reached consensus on the selection of an outside firm (TRAK) to provide independent advice on the model. Mr. Doiron recalls having been provided a copy of the TRAK (2015) report and was present for Mr. Brien’s PowerPoint presentation. [Text redacted]. The committee was involved in reviewing the A/O training package for employees and its implementation across the system. He considered the consultation process to be productive and that discussions always focussed on safety.
 Mr. Doiron testified on the [text redacted].
 Mr. Doiron considers SRAs to be an important tool to mitigate the risk, as the risk is largely determined by the site. Employees should check the area as they come into a site and they should have looked at the SRA for that site. SRAs should be updated periodically, or as needed if something changes at the site. He also considers street inspections to be highly important, to ensure that employees who have been trained in the A/O are following procedures. He was personally involved in street inspections.
 The Teamsters Policy Committee was given the new section on the A/O model in the Employee Handbook to review, comment on and object to as the case may be. The committee arrived at a consensus on its content. [Text redacted]. He also reviewed the technology put in place by Brink’s for the A/O crews, such as the upgraded [text redacted] and the GPS capability. Overall, he considers that the 2-person crew configuration results in the two employees being more alert, more engaged and working better as a team than the 3-person crew. To his knowledge, the A/O is the main operating model for CIT in the industry and has been for some time.
 He stated that crews are always at risk, whether on a 2-person or 3-person crew. In his view, a driver is not pivotal to avoid robbery. Most Garda employees do not even know what a 3-person crew is. He was aware of complaints regarding technical failure of the equipment at times and believes that Brink’s addresses those problems as they come up and would simply stop performing the All Off if the electronics stopped working. When asked about dead zones for the [text redacted], he indicated that could also happen to a 3-person crew. Technology is not foolproof. [Text redacted].
 In his view, the two models (A/O and 3-person crew) are equivalent from a crew safety perspective. He testified that occupational health and safety of its members is very important for the Teamsters Union and that they will not hesitate to strike over these issues. If he and his members felt that the A/O was unsafe and presented an unacceptable risk, he would not allow it to be done. He noted that Brink’s follows and continues to monitor crews and has made safety enhancements to the model since its implementation two years ago. He stated that Brink’s A/O model is the best of its kind in Canada.
 The employer called [text redacted] to testify. [Text redacted] is the senior director, Corporate Security with Brink’s, a position he has held for seven years. He has 23 years of experience in the CIT sector. He is responsible for prevention, detection, education, monitoring, and risk evaluation and investigation activities. The NCC falls under his responsibility. He collects information through context and public information with respect to criminality trends in the industry. Concerning attacks on the A/O model, he testified that they have been largely made against Garda which is a different model than Brink’s’. He noted the differences such as the types of vehicles, on board security equipment, communication equipment, standard operating procedure (SOP) [text redacted].
 Brink’s services approximately two million stops per year and owns approximately 46% of the market. He stated that the majority of CIT attacks occur at night. Since implementing the A/O model 26 months ago, Brink’s had recorded 0 incident and 0 attempted robbery. There were no attacks at all in the industry in 2016. He provided the statistical information to Mr. Brien for the purpose of Mr. Brien’s risk management analysis of the A/O model. He felt that it was important to obtain an independent risk analysis and retained TRAK, which was chosen as a result of the credibility and expertise of Mr. Serge Barbeau, a well-known former police commissioner in Quebec, who heads that consulting firm.
 [Text redacted] first described the new technical features of the A/O model. [Text redacted].
 These features are described as a layered defence. [Text redacted] explained that when you try to defend something you build multiple layers in a circle. To enter an A/O vehicle now requires more tools, knowledge, time and a greater skill set, all of which creates a delay allowing employees to react. The configuration of the A/O truck eliminates the option of opening the front door and the possibility of someone overpowering the driver and driving away with the vehicle.
 Having the two employees inside the front of the cab gives the crew the ability to better communicate with each other and an increased capability of visual monitoring and observing. [Text redacted].
 The operation of the NCC is subject to continuous improvement, in terms of statistics collection, performance measures to respond to calls and a scorecard for NCC operators’ performance. The truck “mode change” function [text redacted], as the crew leaves and returns to the Brink’s compound, [text redacted] and represents 80% of the calls to the NCC; [text redacted]. [Text redacted] Brink’s has in place a [text redacted] if anything happens to a vehicle outside a particular zone, as an alert.
 SRAs are now centralized at the NCC in one database as suggested in the TRAK (2015) report. Consequently, all SRAs become a tool available to employees through the National Control Centre as their resource. When a crew is approaching a site, they can call the NCC to obtain information if they have not reviewed the SRA at the branch. Site inspections with respect to a 3-person crew and a 2-person crew have remained very similar to what a crew member looks for once inside a facility. [Text redacted] is the major differentiator. [Text redacted] noted that with a 3-person crew you are [text redacted]. With the 2-person crew and in [text redacted], the employees have a better sense of what is happening and can see areas the driver cannot see, as well as listen and smell where the driver cannot.
 [Text redacted] further stated that the distance between the driver/guard and messenger when outside the truck in the A/O model is [text redacted], thereby allowing an increased reaction capacity. The [text redacted], which is situated next to the [text redacted] on the truck, mitigates against exposure with valuables in hand. It allows the messenger to put the liability into it from inside the truck and once they have exited, [text redacted].
 The street inspections that security does are totally independent of the branch. In 2017, the security department did 146 street inspections out of a total 522 operators. For 2018, street inspections will be a key performance indicator for managers. [Text redacted] is requesting an increased number of street inspections during that year. The purpose of the street audit is to measure the performance of the crews on the street and is a means for ensuring ongoing supervision and oversight of staff as recommended in the TRAK (2015) report. The inspectors try to measure the level of compliance with Brink’s policies, i.e., is the driver wearing a seatbelt, are the guard and messenger remaining at a distance, [text redacted] Etc. Street audits have been standardized and data has been centralized. Brink’s Security can extract data and review results that can result in additional coaching for employees.
 [Text redacted].
 He stated that, based on his experience, a 3-person crew configuration does raise the concern that the driver [text redacted]. He confirmed that he does not recall that there were attacks on 3-person crews in the past five years; however, there had been attacks on 4 and 5-person crews in the past.
 [Text redacted] testified for the employer. [Text redacted] is the senior manager, Operational Compliance, for Brink’s. He has 25 years of experience in the armoured car industry. He initially started on the road as an armoured guard in the ATM division and with G4S and moved up through the ranks to his current position. He is responsible for the implementation of the Brink’s A/O model. I will summarize the salient points of his testimony.
 He had a role in the development of the training program, in consultation with the policy committees. He testified that the final language for the SRA training came out of feedback from both National Health and Safety Policy Committees. He made reference to the risk assessment matrix that he used to assess the hazard. The matrix is designed to determine severity, frequency and probability and it is an effort to come up with significance. Based on history, it came from a program that was used by Securicor and carried on by G4S via Australia's program 4390. He testified that it was standard 1300 and that Brink’s was adopting a standard that was already tested and it was recognized by HRDC and ESDC as being an acceptable approach. He testified that the risk assessment priority form is [text redacted] to bring risk down to acceptable levels. Referring to the SRA training materials, [text redacted] highlighted the fact that it is a “mocked up” version of an actual site used for training purposes and specifically highlights severity levels such as very high risk and medium risk, which stimulates questions about how a site is assessed. SRAs are developed jointly, one worker and one manager representative going into the field to assess the site and [text redacted]. Layout may be documented by pictures or sketches. Hazards are identified and an appropriate control measure is identified to mitigate it. Branch managers have the final sign-off on the SRAs.
 [Text redacted] also described the initial training provided to employees, including [text redacted], and of training given for use of force and pistol, up to three days, depending on the size of the class.
 A typical training session for the A/O model would be 9 to 10 hours. Employees who require more information are accommodated.
 With respect to security bulletins, [text redacted] testified that there are often team talks concerning them if they are posted at the branch and gave Christmas/Halloween as examples. Those bulletins are meant to warn employees of particular circumstances affecting the risk of being attacked and the safety of crew members.
 Concerning street inspections, [text redacted] testified that they were prepared by the corporate security department and that they took place at the branches, using the [text redacted] and, in some instances, paper methods to gather data. He also testified that there was a new initiative in place to provide the National Health and Safety Policy Committee with quarterly updates dealing with such things as frequently failed items, indication of items and corrective actions.
 [Text redacted] referred to an employee notice that all employees are required to wear the [text redacted], which was a commitment made to the National Health and Safety Policy Committee. Another notice was advising employees trained in the A/O model that, if they have any questions in regards to the operation or corporate policy related to the use of the A/O vehicle, they were to approach their supervisors or manager for assistance. There was the odd request made by employees for additional training or, if local management had identified that additional training is required for certain employees, it is given.
 [Text redacted].
 [Text redacted].
 He also spoke to the [text redacted] introduced to improve communications and about how a call is made to the NCC when an employee is under duress [Text redacted]. The phone [text redacted] allowing crew members to check in with one another and not depend solely on cell communications. The battery life for [text redacted] is 1000 hours of standby power and 40 hours of talk time and there are replacement batteries in the truck.
 Concerning the specifics of the operational procedures relating to the A/O model, [text redacted] testified that, for Brink’s, the starting point was the A/O procedure that G4S was using and then from there they moved forward, developing Brink’s specific standard operating procedures, which [text redacted] referred to in his testimony. I will set out a detailed description of the A/O entry and re-entry procedures given their importance to the issues raised by the appeal.
 [Text redacted] testified that the strength of the A/O model was the mobility of the driver/guard and his ability to explore areas outside a driver's normal view.
 [Text redacted].
 [Text redacted].
 [Text redacted].
 [Text redacted] indicated that they had an equal or superior service provider with [text redacted] which improves the transmission of the phone, reducing the delays to almost no delay, and that this was implemented mid-year 2016. Fundamentally, there was more control and it ensured that potential false alerts did not occur.
 [Text redacted] testified that the consultation process at the Teamsters Policy Committee was cooperative and collaborative, unlike the Unifor Policy Committee. Unifor would not support an A/O model and agenda items dealing with A/O were actually removed from the Committee’s agenda by Unifor, to be dealt with in collective bargaining that was taking place at the time. Negotiations eventually resulted in a settlement with respect to A/O and premium compensation was agreed to by the parties for employees working on an A/O crew.
 Turning to the events of October 25, 2016 and Mr. Childs’ work refusal, [text redacted] testified that Mr. Childs held a position in the Workplace Health and Safety Committee (WHSC). [Text redacted] stated that the A/O model was being rolled out in Ottawa and Mr. Childs was on the first run (Run 78). He was trained on the A/O model. He was trained on SRAs and there were SRAs prepared for every site on Run 78, including the CIBC Hawkesbury Branch location. [Text redacted] anticipated a refusal in light of the Union’s perspective on the A/O model and a certain level of resistance by employees, and he asked corporate security to do surveillance for [Text redacted]. When he was shadowing Mr. Childs and [text redacted] (driver/guard) in the days prior to the refusal, he felt that they were asking good questions and provided good feedback. They would stop them at every 2-3 calls to talk about where they could improve. On the second day of shadowing, [text redacted] testified that Mr. Childs was positive and even indicated that he enjoyed the A/O model.
 On the night of October 25, Mr. Childs stated that he was not comfortable, and that he was concerned about the lighting around the call and more concerned with exiting the bank. His partner had no concern and was not planning on refusing. [Text redacted] arrived at the scene about 20 minutes after being informed of the refusal. He watched the surveillance video taken by [text redacted] that night at the CIBC Hawkesbury location, where Mr. Childs made his refusal. Mr. Childs was the first one out of the truck, he was smoking and talking on his cell phone while travelling up and down, remaining outside for a number minutes. [Text redacted] was the driver/guard and he should have been the first person out of the truck in this case. He testified that Mr. Childs' actions were more like those of someone taking a break. Mr. Childs had advised him that he felt the A/O model was unsafe. When asked if he perceived any threats, [text redacted] testified he did not see anything, other than observing two ladies in the bank’s vestibule, who were asked to leave without a problem.
 The video was entered as an exhibit and played at the hearing.
 The employer called the refusing employee, Mr. Childs, to testify. Mr. Childs was summoned by the employer to testify at the hearing. He has been employed by Brink’s since 2007. Mr. Childs described his routine when he checks in at the branch. He ensures that he has [text redacted]. Everything was working correctly the night of his refusal. Mr. Childs was the messenger/crew chief on the night of his refusal. He described the A/O procedure that the crew must follow, once they arrive at a customer location, much along the lines of [text redacted] description.
 Mr. Childs confirmed that he had received training on the A/O model and procedures which lasted approximately nine hours, in addition to [text redacted] training and [text redacted] training. He received the [text redacted] training, which was in class for 2-4 days, and went back for more training as he was the ATM representative on the WHSC. He stated that the training was excellent in his opinion, that [text redacted] was a good trainer, going into a lot of details about the A/O crew configuration. Mr. Childs described his understanding of the purpose of the [text redacted], which essentially mirrored [text redacted] explanations set out above.
 Turning to his work refusal of October 25, 2016, he explained that he understood the position of his union, Unifor, to be against the A/O. He was ATM representative at the time and felt a sense of obligation. He felt some pressure from the floor (some colleagues were OK with the A/O, others were not) to make a refusal. He stated clearly that he was not coerced in doing so and it was his choice. He felt pressure from both ends, as he knew the A/O roll-out was important to management. The grounds for his refusal were not site specific, but a refusal about the A/O model generally. Run 78 was the first A/O run in Ottawa and some of his colleagues, including Mr. Barber, were upset that he did not do the refusal on his first shift. He further explained the reasons for his refusal in a letter to the ministerial delegate and sought input from other union colleagues in drafting the letter, but the letter is his.
 Mr. Childs then spoke to his letter of July 6, 2017 by which he wished to withdraw his refusal. He explained that the whole situation had caused him a lot of stress. Also, he had become used to the A/O model by then, and felt safe operating it. He enjoyed the technology and felt safe especially when teaming up with [text redacted] as driver/guard. He does not miss the third person/driver as he feels he is more in control and there is better communication with his partner. He has had situations where he had to [text redacted] for re-entering the truck after a service with valuables, although he states that in general, drivers were alert. He mentions that the new [text redacted] is efficient and is an improvement. He also appreciated the premium ($1.35) negotiated by Unifor.
 Mr. Childs also mentioned that he had been advised of potential threats by the driver on a 3-person crew configuration, and that, as a driver, he had notified crews several times in his career. He agreed that the most critical area is the distance between the truck and the location, [Text redacted]. He reported it. On another occasion, the [text redacted] on his phone went on. He shook the phone and it stopped. He recalls that an NCC operator called him “pretty quickly”. He states that he is not concerned with [text redacted] if it is done correctly and the crew remains alert all the time.
 The employer called [text redacted]. [Text redacted] has been employed with Brink’s for 11 years and held numerous positions including Guard, ATM Technician, Crew Chief, Cash Logistics Supervisor and now Branch Trainer for Ottawa. He testified that he had received all of the Brink’s training, including A/O and Site Risk Assessment, as well as Health and Safety modules. He had received his Site Risk Assessment training from [text redacted] and Mike Childs, and [text redacted] were also in the class. The training lasted for 8 hours. Concerning the Risk Assessment training that [text redacted] gave, he remembers risk examples that were used in the room such as a cable that was across the floor and a number of other scenarios. He also testified that [Text redacted] had a mock-up site risk assessment in the presentation, to show the class what it looked like when completed. [Text redacted] then presented a PowerPoint which resulted in questions being raised and issues being discussed. [Text redacted] stated that there was quite a bit of time spent going through the matrix.
 [Text redacted] described the procedure he followed to prepare the site risk assessments for Run 78, which includes the Hawkesbury site. He did the site assessments along with [text redacted], the employee co-chair of the Health and Safety Committee. That process took a few days to complete. He made reference to a discussion with [text redacted] concerning the poor lighting at the Hawkesbury location and advised that management would escalate the issue to the bank, forwarding a deficiency report to its head office.
 [Text redacted] described the site risk assessment process in further detail. He testified that he and [text redacted] arrived before the crew, took pictures and made notes. They followed them inside and assessed the interior and went back to the branch where they wrote up the site risk assessment. There were a number of points of concern raised by [text redacted], which led to a disagreement over some of the control measures. [Text redacted].
 [Text redacted] testified as to how an A/O crew should operate when they arrive at a site and described the procedure along the same lines as [text redacted] and Mr. Childs had described it. [Text redacted].
 [Text redacted] testified that Mr. Childs' work refusal took place on the second day that he and [text redacted] were shadowing his crew (on October 25, 2016). He testified that the shadowing was done to see the crew performing their job during the first two days of working an A/O run. He considered that Mr. Childs and this driver/guard, [text redacted], were showing improvements at applying the procedure as they serviced more locations. [Text redacted] was questioned about the comments he wrote on the inspection sheets regarding the site inspections by the driver/guard, as being “overdone” or “not consistent in efficiency”. [Text redacted].
 Prior to arriving at the Hawkesbury location on the night of the refusal, [text redacted] received a phone call from [text redacted], supervisor, that Mr. Childs was doing a work refusal at CIBC Hawkesbury. [Text redacted] confirmed that he did an investigation and took notes. Mr. Childs’ reasons for refusing were that the A/O model was unsafe. His refusal did not refer to an imminent threat or danger, or to any particular problem with the site.
 Finally, [text redacted] outlined the security measures of the truck which in his view provide a safe environment that mitigates the risks for employees, along the same lines as [text redacted] in their testimony.
 During the hearing, I took a view of the truck at the Ottawa Branch in the presence of all parties. [Text redacted] was responsible for the technical demonstration, [text redacted]. When he resumed his testimony at a later point in the hearing, he noted that he had already raised the problem with his boss and [text redacted]. He also stated that he had not had any issues [text redacted] before. Had this situation occurred on a real run, the truck would simply not have been used, as other witnesses and evidence have established.
 The Union called [text redacted] to testify. [Text redacted] has been employed with Brink’s since August 2006 and is currently a messenger/crew chief on an A/O run. He has been working on A/O since September 2017. He stated that as a driver in a 3-person crew, he had the opportunity to monitor the risks while the crew was inside the premises and to alert them of any unusual situation before they would leave. As a messenger, he was contacted by the driver on occasions, to give him a heads up.
 [Text redacted] testified that he took the A/O training given by [text redacted]. He thought the training was adequate and complete, and stated that he was impressed with the technology and felt that Brink’s was genuinely trying to make the A/O as safe as possible. He then recounted some of the problems he encountered with the technology: [Text redacted]. He feels that the A/O model is not as safe as the 3-person crew because of the absence of the driver, who is there to monitor the surroundings and whose presence is a deterrent to criminals. In his opinion, [text redacted] is not satisfactory as service calls make take up to 45 minutes and a lot may happen during that time.
 [Text redacted] added that he is now comfortable working on the A/O model. He testified that he had bid to work on an A/O crew, and it was his choice to do so. He understands the risks and agrees that a 3-person crew also entails risks. He stated that he has never been called upon to do [Text redacted], working as a messenger; he and his driver partner received a verbal warning for not following procedures, namely omitting [text redacted]. He agrees that [text redacted] were significant safety enhancements. [Text redacted].
 The Union called Mr. Maurice Mills. Mr. Mills was the co-chair of the Teamsters Policy Committee from June 2015 to June 2017. Mr. Mills stated that Unifor would not accept the A/O crew configuration. He stated that the training, procedures and technology were all discussed at the committee but that, overall, his committee members took the position that A/O was not safe because, when the crew comes out of a call, [text redacted]. The only acceptable control measure is to put the [text redacted].
 Mr. Mills explained that the Union eventually asked that the items related to the A/O be removed from the committee’s agenda, so as to avoid interfering with collective bargaining in Ontario and British Columbia at the time, where the A/O issue was on the table for discussion. Mr. Mills confirmed that Unifor had negotiated collective agreements with Garda in British Columbia which included wording on the implementation of A/O crews and introduced copies of the relevant articles of the collective agreements. That wording had been in place since 2005.
 Mr. Mike Armstrong testified on behalf of the Union. Mr. Armstrong has been a national representative of Unifor for 27 years and was responsible for the Armoured Car Industry sector for the past 10 years.
 He stated that he was involved in the negotiation of the most recent collective agreement reached in October 2017 with Brink’s and that the contentious issues were money and A/O. It was Unifor's position that it was firmly against the A/O model and it continues to be their position today. The A/O premium was put on the table by Brink’s and they maintain their position that they were going to implement an A/O, whether Unifor agreed or not, as Brink’s required a level playing field to compete. The majority of the competitors had been doing [text redacted]. Mr. Armstrong stated that, ultimately, Unifor accepted the A/O as people's livelihoods were at stake and it was up to the membership to decide. He is aware that the A/O model raised a great deal of anxiety with the membership. Some branches voted against the proposed agreement and ratification was considered weak at 57.6%.
 When asked if the national union considered the A/O model to be safe, further to this ratification, Mr. Armstrong stated that Unifor's position was that a [text redacted] was not safe. Unifor’s research department has prepared policy papers on the subject and called it a race to the bottom as there are no real regulations in the industry. He stated that Nova Scotia is the only jurisdiction who passed regulations in that area. The regulations set the minimum crewing to 2 persons.
 Mr. Armstrong indicated that he had not encouraged Mr. Childs or other employees to do a work refusal. His approach was that employees should try the model for a couple of weeks and not rush to judgement, and make up their own mind if it is safe or not. He testified that Canadian Auto Workers /Unifor in British Columbia has had A/O crews for over a decade and that Unifor has negotiated collective agreements which contemplate use of the A/O model. He pointed out that locals have a great deal of autonomy in their collective bargaining. He stated that Unifor had represented employees of G4S in Ottawa, which had an A/O model, based on what the membership wanted. Referring to a statement in a press release from Unifor’s President issued during negotiations, that Unifor would not “sign an agreement that puts lives in jeopardy”, he stated that he did not believe that the agreement reached with Brink’s would put lives in jeopardy. Finally, he stated that he was not aware of any attacks on Brink’s A/O crews over the past two years.
 [Text redacted] was called by the Union. [Text redacted] has nine years of experience as an employee of Brink’s. He has acted as ATM technician, messenger and driver. He is currently working the A/O configuration, which he started 6-7 months ago. He received the training prior to starting working on the A/O. [Text redacted] referred to issues that he has had with the equipment described by other witnesses: [text redacted] does not always work well (although he agrees that the recent change in software was an improvement) and [text redacted] do not always work correctly on the new trucks, [Text redacted]. He also testified that he had weekly [text redacted], and that sometimes the door would not close properly, which caused him to contact the NCC. [Text redacted]. The truck on which the problem occurred is no longer used. Although the equipment is checked to make sure everything is in order before leaving the branch, he stated that he had equipment malfunction when on the road, but agreed that the problem had been addressed.
 [Text redacted] testified that he has been doing [text redacted] when working on models other than A/O. [Text redacted].
 [Text redacted] explained that the element he viewed as the most dangerous with the A/O model is when the messenger goes to and from the truck. He explained the role of the driver in a 3-person crew configuration, much along the same lines as previous testimony, whose presence added an additional layer of security, in his view.
The Barber appeal (2017-12)
 Mr. Kevin Barber’s appeal concerns a decision of absence of danger given verbally by the Ministerial Delegate, Mr. Jenkins, on April 28, 2017 and confirmed in writing on May 11, 2017. The decision was rendered further to a work refusal made by the appellant on April 3, 2017. Mr. Barber works in the Ottawa Branch of the employer, like Mr. Childs. He had completed his first shift working the A/O model on March 26, 2017. The evening of April 3, 2017 was his third shift working an A/O configuration, on Run 72. He was in the driver/guard position that evening. Mr. Barber’s reasons for refusing to work essentially related to the A/O model as a whole being inherently unsafe, primarily because of the absence of a third person remaining in the truck, the driver, while the other 2 crew members carry the cash and valuables at each point of delivery and return to the truck.
 His text message setting out the reasons for his refusal on April 3, 2017, reads as follows:
I’m conducting a work refusal under the Canada Labour Code because I feel the All Off Model is too dangerous to work without having a driver as a third person to watch out for us while we are working inside the banks. The All Off Model is unsafe to work.
[Text redacted] has informed me that by refusing the Model that the ESDC deemed unsafe but also has given a stay to Brinks, with guidelines to follow, that my refusal could be deemed frivolous.
I then told [text redacted] the measures that ESDC put in place don’t make me feel safe enough. He replied that it is not up to me make those decisions.
I feel as though the model in its entirety is unsafe do to the fact that removing the driver as the third person takes away our ability to prevent or potentially stop robbers from ambushing us (especially the technician). At the end of the day I fear for my life and the life of my crew members.
The driver as the third person can survey the surrounding areas of the truck and bank while the crew is in a call. [Text redacted].
The driver as a third person can also return fire with the pistol or shotgun from the safety of the truck. The driver can also make sure that the truck isn’t being tampered with. [Text redacted].
Most importantly. The third person can warn the crew of suspicious behaviour or patterns that individuals develop around the truck. This increases the safety and security of the crew. The driver can notice a suspicious vehicle that appears often at the same call when the crew is in a call. The driver can warn the crew of people peering through the windows to look at the crews working in the bank.
 Seized with the matter at the initial stage of the process as contemplated in section 128 of the Code, the Workplace Health and Safety Committee reached an impasse regarding the alleged unsafe nature of the A/O model.
 The employer’s position regarding Mr. Barber’s refusal was that there was no particular hazard that presented an imminent or serious danger to the employee at the time of the refusal. Furthermore, the employer had implemented all temporary preventive measures directed by the appeals officer as conditions for staying the direction issued in the Childs refusal (Brink’s Canada Ltd. v. Childs, 2017 OHSTC 4). As there were no special circumstances invoked by Mr. Barber on the night of his refusal, the employer considered Mr. Barber’s refusal to be abusive and inappropriate. Mr. Barber disagreed that the additional measures resulted in improving the safety of the A/O model, which in his view remained unsafe.
 The ministerial delegate conducted his investigation into the refusal and concluded that a danger did not exist based on the grounds set out in his report as follows:
In his refusal the matter that had to be met first was whether his refusal was substantially different than that of the previous refusals in Edmonton and Ottawa that concluded a danger existed. After a review of the file a review of the decisions by the two Appeals officers on the application for a stay I found that this refusal was substantially the same as the Ottawa refusal in that the refusal was based on the All Off Model and not specifically on a set location. As such, since the employer has implemented all four conditions set out in the Appeals’ Officer’s stay decision on the Ottawa direction this officer is of the opinion that section 129(3.1)(a) of the Code can be used and that I am able to render a decision of no danger by relying on the OHSTC stay decision in so much as the Appeals Officer’s decision allows for the two All Off model to be used under certain conditions thereby alleviating the danger until a decision on the appeal occurs. It is for these reasons that I find that a danger does not exist under the Canada Labour Code Part II.
 It is useful to quote the conditions imposed along with the stay of the direction in the Childs appeal:
 As mentioned in the letter to the parties dated March 3, 2017, the granting of the stay is conditional on the employer abiding by the following four (4) conditions, which are essentially along the same lines as the conditions ordered in Brink’s Canada Ltd. v. Robert Dendura:
As undertaken by the employer, and with respect to its Ottawa location, the employer shall:
- Review with each regularly assigned “All Off” crew member on a monthly basis their comfort with the “All Off” protocols and procedures and consolidate any constructive feedback to be assessed and considered for any revision of “All Off” standard operating procedures;
- Continuous updating and revision as necessary of Site Risk Assessments;
- Continuous updating and revision of “All Off” Specific Operating Procedures as necessary;
- Ensure that over the course of each month, each “All Off” crew will be shadowed a minimum of two times by an additional Brink’s personnel for all or a portion of their route to ensure adherence to procedures, avoid complacency, assess and abate risk and to provide additional eyes to alert the crew to any risk.
 As the ministerial delegate’s investigation progressed and in his testimony at the hearing, Mr. Barber provided additional details on his concerns with the A/O model. In the employee’s description of the events, the ministerial delegate transcribed Mr. Barber’s concerns as follows:
I feel as though the model in its entirety is unsafe do to the fact that removing the driver as the third person takes away our ability to prevent or potentially stop robbers from ambushing us (especially the technician).
At the end of the day, I fear for my life and the life of my crew members.
The driver as the third person can survey the surrounding area of the truck and bank while the crew is in a call. [Text redacted].
The driver as a third person can also return fire with the pistol or shotgun from the safety of the truck.
[Text redacted]. This new All Off Model does not allow that measure of extra safety and security. [Text redacted] to prevent the crew from getting into the truck and waste enough time to make an attack.
Most importantly. The third person driver can warn the crew of' suspicious behaviour or patterns that individuals develop around the truck. This increases the safety and security of the crew. The driver can notice a suspicious vehicle that appears often from call to call when the crew is in a call. The driver can warn the crew of people peering through the windows to look at the crews working in the banks. The driver can definitely help mitigate any ambush attempt on the crew coming out of any call.
The driver allows the crew to always have a perspective of the outside of the bank. They can keep in constant communication to be notified of any possible issues or suspicious events that can arise.
[Text redacted]. The driver as a third person could give warnings to the whole crew.
The All Off model crew is not even provided with a copy of the training. We were only trained by a PowerPoint presentation and an hour of hands on. You're expected to learn on the job a much more dangerous model.
ln the training, l had asked [text redacted] if' we could be provided with [text redacted] and he declined. [Text redacted].
The company has informed us that the All Off is still on the job training.
I only work the All Off once a week as it is. There isn't enough time to learn to be comfortable with the model and the extra hazards that could arise.
We have constant issues with the new phones to add to the growing list of issues. [Text redacted]. Has this been a robbery, it could jeopardized the life of my crew and myself. [Text redacted].
Robbery scenarios were never explained and almost avoided at the training seminar as though they were not a possibility.
Bank checks are conducted only on "the footpath" to the ATM room as per the request of [text redacted]. (Run 72 has encountered a emergency door at 1930 St, Laurent (RBC) open on Jan 31 and that was with a 3man crew. Had that crew not checked the door regardless of [text redacted] request to stay on the footpath, robbers or other unauthorized persons could have entered the location.)
No safe loading area for the shotgun, Never tested for ricochet.
lf' something is forgotten in the truck. Both guard/driver and crew chief must go back to the truck increasing vulnerability to the crew.
This happens more often than none due to the fact that we may need extra supplies and or forget a parcel.
The truck is the only safe location at any stop and once we return to the truck, there is no absolute guarantee that the truck is still safe.
When I started at Brink’s we had to have the driver radio the turret guard to let them know everything was OK on the hour. [Text redacted].
No more base radios to communicate with other crews in case of emergency.
 Mr. Barber’s testimony at the hearing may be summarized as follows.
 He has been employed with [text redacted] for eight years, in all positions. Mr. Barber indicated that at the time he bid for an A/O position, he had not been trained but received A/O training one week before. He explained that the reason he bid for a position on an A/O run is that he wanted to be open-minded and try something new and hopefully there would be a monetary incentive to work the A/O model, which there was not at that time.
 Mr. Barber described the procedures followed by the crew with respect to a pre-trip inspection for the truck and the equipment, noting that it was a new truck and that all equipment was functioning before they left the branch on the evening of his refusal, as well as on the two previous days. The first two nights, he was followed by management and on the third night, he and [Text redacted] were asked to do an extra cash exchange. This was the first time that he was doing the A/O without the assistance of “extra eyes”. [Text redacted].
 He indicated that he did [Text redacted] and that his [Text redacted] takes him away from the line of sight. Once in the truck, he testified that he finished his paperwork, secured the old load and fixed the new load. The crew then took a coffee break. He began to think about what he did and spoke to [Text redacted] about a work refusal. Once arrived at the Hazeldean location, he made the call to his supervisor, [Text redacted], and advised him that he was doing a work refusal. He testified that he had advised his supervisor that he did not think the A/O model was safe enough and that he needed an extra set of eyes, "as you don't know what you are coming out to”. When [text redacted] arrived, he stated he was told that what he was doing was frivolous and vexatious because there were stay conditions in place.
 Mr. Barber went over a number of technical issues that were occurring with Truck 16208, in use the night of his refusal: [Text redacted] were wearing down and you needed to move them around to hear the grinding noise. He now drives a new truck on the A/O and [text redacted]. He also noted that the driver has “more power” than the messenger/crew chief because when the messenger tries to move the truck, [text redacted] and alarms: the truck can only be [text redacted], which has proven to be embarrassing in one situation where police had asked him to move the truck and he couldn’t.
 Mr. Barber further testified that, when he was shadowed by [text redacted] and [text redacted], the feedback he received was that he had to perform his checks of the premises faster than he did: [Text redacted]. He explained that there was nothing in the stay conditions that made him feel safe and only shadowing made him feel safer.
 Mr. Barber testified that the next morning after his work refusal, he advised management of the problems at the Almonte BMO call. He indicated that management did listen to his concerns and changed the standard operating procedures for that location, [text redacted].
 Mr. Barber acknowledged that, when he exercised his work refusal on April 3, all of the equipment was functioning properly at the Hazledean site. The only reason he stated for his refusal was that he didn't feel safe using the A/O, as with the absence of a driver, there was not an extra set of eyes. He stated that he does not always follow procedure and that Brink’s policies are not to keep the crew safer, but to protect the valuables and be efficient (“time is money”). He added that employees need to cut procedures to get time back and that he couldn't afford a suspension and loss of salary.
 Overall and when considered in light of the evidence of all other witnesses, I found that Mr. Barber’s testimony should be considered with caution. I found him to be argumentative and somewhat prone to exaggeration in his answers. His concerns about the A/O model, that I have chosen to reproduce in their entirety above, raise a number of considerations that were not raised by other witnesses called by the Union, or are in contradiction with their evidence [text redacted], and not entirely reliable. This does not mean that I reject his testimony altogether, but simply that I will accept it with an appropriate level of caution. I accept that the fundamental basis of his refusal on the evening of April 3, 2017 is that he did not consider the A/O crew configuration to be safe, primarily because of the absence of a driver to keep watch and notify the crew as needed, while the crew is providing service to the location and when it is about to walk back to the truck. That is indeed the heart of the issue in the present cases.
Submissions of the parties
 Counsel for the parties agreed to present their written submissions-in-chief simultaneously, within one month of the last hearing day. An additional two weeks was provided to allow counsel to respond to each other’s submissions in reply. The final submissions were received at the Tribunal on March 26, 2018.
Brink’s Canada Ltd.
 Mr. Henderson presented the written submissions on behalf of the employer. Counsel for the employer first summarized the testimony of each witness who testified at the hearing. It is not necessary to go into this summary, as I have outlined the evidence which I consider most relevant to the present appeals earlier in these reasons.
 Counsel for the employer then refers to the definition of danger in the Code and cites my decision in Dendura, as well as the two leading cases setting out the test to determine whether a danger exists, under the new definition of “danger” adopted by Parliament and which came into force on October 31, 2014: Ketcheson and Keith Hall & Sons.
 Counsel submits that the alleged “hazard, condition or activity” in the present cases is the risk of potential robbery or attack when the crew returns to the armoured vehicle from the premises without a driver having remained in the vehicle as a second set of eyes to keep watch on the surroundings and provide information to the crew when exiting the premises.
 On the first element of the definition as to whether the employees were facing a threat to their life or health on the night of their refusal, the employer submits that there was no evidence of an imminent threat as those words were interpreted in Ketcheson.
 Turning to the second element and citing Pogue v. Brink’s Canada Ltd., 2017 OHSTC 27, and Dendura, counsel for the employer submits that in order to conclude that the refusing employees will be exposed to a serious threat, the evidence must necessarily show that there is a reasonable expectation that they will be faced, in days, weeks, months ahead, with a situation that could cause them harm as a result of performing ATM services using the A/O model.
 The probability that an A/O crew member will be ambushed after exiting the customer location when returning to the vehicle is sufficiently low, based on the statistics over 26 months, that it does not constitute a threat. Counsel refers to the testimony of [text redacted]. He also referred to the testimony of Mr. Doiron, who expressed the view that, when you compare the 3-person crew to the 2-person A/O crew, they are equivalent because of the steps recommended and taken at the National Health and Safety Policy Committee. He went on to testify that, when using [text redacted] and SRAs for a site, the crew has 100% knowledge of the lay of the ground on every site. In conclusion, he testified that, if he did not feel the A/O could be done safely, the Teamsters would not be doing it.
 Counsel for the employer also referred to the evidence of management’s witnesses that the number of crew members on board at vehicle has little to no effect on a criminal’s decision making process, in fact, he indicated that there had been attacks on 4 and 5-person crews. He stressed the training received by crew members and the importance of [text redacted] made the risk low.
 Counsel stresses the fact that the previous statistics regarding assault on A/O crews involve different models than the best practices model developed and employed by Brinks, which has not had a robbery attempt since implementation. In implementing a best practices approach, Brink’s has improved its communications technology, personal protective equipment, theft mitigation devices, armored truck design, specialized training for A/O crews and site risk assessments to determine hazards at each location, as well as standard operating procedures that require [text redacted] also stressed the importance of being efficient at sites to further reduce time exposed to possible threats.
 Counsel for the employer submits that Brink’s has applied the conditions of the stay that was granted, which provided for additional mitigating measures pending the outcome of the present appeals, and that the reasonable possibility of employees being subject to a robbery attempt and suffering serious injury is low. The Tribunal in Ketcheson was clear that a very low risk, either because of low probability or because of low severity, is not a threat. Consequently, the probability that an A/O crew member will be ambushed when exiting a customer location and returning to the vehicle is low and, therefore, a serious threat does not exist.
 Counsel acknowledged my conclusion in Dendura to the effect that the A/O model exposed the employees to a serious threat. However, he submits that the evidence adduced at the hearing of the present appeals provides additional facts that should lead me to reconsider my conclusion and find that the employees were not facing a serious threat on the days of their refusals and that no danger should be found. Counsel referred in particular to the second TRAK report (2017) and the testimony of Mr. Brien that the risk of serious harm to employees operating the A/O model was low, in light of the most recent statistical information and the longer period of time during which no robbery attacks were reported. Counsel also pointed to the testimony of Mr. Doiron, who had not testified at the Dendura hearing.
 For all of the above reasons, the employer is asking that I rescind the direction issued in the Childs case and that I confirm the decision rendered in the Barber case.
 Ms. Lundquist, on behalf of Unifor, first submitted that the issues to be determined in the present appeals were also before me in the Dendura case. However, the employee in that case was not represented by counsel or his union. The appeals officer in Dendura was therefore not apprised of the statistical evidence that unequivocally shows the A/O model is a danger. That evidence is on record in the present case and shows that the A/O model creates an increased risk to the health and safety of employees.
 Counsel summarized her submissions that the A/O model can reasonably be expected to pose serious threat to the life or health of the employees exposed to it. The change to the traditional 3-person crew eliminates the driver and the tactical vision that previously protected crews from threats to their safety while exiting a service location. The increased danger resulting from the change is not a normal condition of employment. The technology and procedures implemented by Brink’s do not adequately mitigate the serious risk of robbery and assault that arises as a result of the adoption of the A/O model.
 In addition, counsel points out that the stay conditions granted in the Childs appeal do not adequately address the inherent dangers of the A/O model. That danger is an increased threat of robbery and assault and those measures simply cannot protect A/O crews from this serious and increased threat.
 Counsel for Unifor submits that the employer did not have meaningful consultations with the Union prior to adopting the A/O and the Union continues to believe that the A/O model is not safe.
 Counsel reviewed the testimony of Mr. Childs and the events of October 25, 2016 when he exercised his right to refuse on the basis that he considered the A/O model to be unsafe. Regarding his eventual withdrawal of his refusal, counsel stresses the importance of highlighting the context of his withdrawal: it is clear that money and job security were significant considerations in his decision to withdraw.
 Counsel for Unifor referred to the testimony of [text redacted], who testified that going to and from the truck as a guard/driver was the most dangerous aspect of the job and the A/O model increased that risk. That testimony is consistent with Brink’s training materials, as well as the expert report which states that the footpaths to and from the vehicle are where most of the robberies occur.
 Counsel then reviews the testimony of Mr. Barber and the events surrounding his work refusal of April 3, 2017.
 Regarding the fact that the union membership ratified a collective agreement, including an A/O premium, counsel submits that the context of such agreement at the bargaining table should be considered. Brink’s had stated in no uncertain terms that they would implement the A/O model without Unifor’s consent. That agreement should not be construed as the Union’s explicit or implicit consent to the A/O model. Unifor was cognizant of the present work refusals and the safety of the model would be adjudicated and ultimately decided by the Tribunal.
 Counsel for the Union refers to section 128 of the Code and the definition of danger, as well as the need to first identify the “hazard, condition or activity” in question here. She submits that the threat in the present appeals is two-fold: first, there is the increased risk of a guard being ambushed while returning to the vehicle from the service location without a driver being able to warn them of potential threats, or to be seen in the truck as a visual deterrent to criminals. The crew then becomes a “soft target” for criminals. Second, there is an added threat of delayed emergency response: [text redacted]. The technology introduced by Brink’s does not adequately address that added risk.
 On the first element of the definition, as to whether the employees were facing a threat to their life or health on the night of their refusal, counsel points out that neither Mr. Childs nor Mr. Barber asserted that they were facing an imminent threat.
 On the second element of the definition however, counsel submits that employees are facing a serious threat when working on the A/O model, as those words were interpreted and applied in Ketcheson and Keith Hall & Sons. Statistics indicate that the threat of robbery and resulting injury or death is a reality in the armoured car industry. It is not merely speculative or hypothetical (6.6 attacks per year on average, between 2000 and 2015). Although the likelihood of a robbery or assault remains relatively low, the severity of an injury, should a robbery occur, is extremely high (Securicor). It is evident that an attack could occur before the activity could be altered.
 Counsel further argues that the condition of danger is established in the present cases and that the danger is not a normal condition of employment. There is a level of inherent risk of robbery in the armoured truck industry, but the analysis is to determine whether the introduction of a policy - in this case the A/O model – increases that risk to an unacceptable level (Verville v. Canada (Correctional Services), 2004 FC 767)(Verville). Counsel submits that the expert report commissioned by the employer indicates that, between 2000 and 2015, 70% of the attacks were made against A/O crews. Brink’s had yet to implement their A/O model during this same time period. Only 17% of these attacks were made against their 3-person crews. Yet, Brink’s controls 46% of the market share, which might indicate, as the report notes, a preference on the part of criminals for the A/O configuration when planning an attack.
 Counsel for Unifor also cites Brink’s own statistics developed in a report by their security department in 2014, which concludes that, based on data extracted from past attacks, Brink’s traditional 3-person crew model results in a service that is materially safer for employees, customers and the general public (Canadian Armoured Truck Industry Statistics (Revised) - Exhibit U-2).
 Counsel argues further that the measures put forward by Brink’s to mitigate the risk do not adequately protect the employees from the danger. First, [text redacted] is a technique used with other models of delivery, and is not unique to the A/O model. The testimony of [text redacted] is clear that [text redacted] exposed the guard/driver to unacceptable risk; because you do not see the other side of the truck, it is easy for robbers to attack the guard/driver while he is performing [text redacted], and it was not uncommon for guards to forget to [text redacted].
 Likewise, the [text redacted] is not unique to the A/O model and counsel suggests that it is at best superficial, that efficiency in conducting [text redacted] is the main factor and that safety is only an auxiliary concern. Accordingly, there is no evidence on the record that this element of the A/O model reduces risk.
 Brink’s placed considerable emphasis on the new technology developed for the A/O model. Counsel argues that technology is reactive and not proactive, contrary to the requirements of section 122.2 of the Code. The potential to eliminate or greatly reduce the risk of attack by having a driver present should be prioritized over an emphasis on personal protective equipment. None of the technology described in the evidence protects A/O crews from the principal threat: robbery and assault, and is meant to be used after a robbery has already occurred or in progress. Furthermore, the evidence establishes frequent malfunction of the equipment, [text redacted] instantaneously as it should, [text redacted], etc. Counsel also submits that [text redacted] has proven to be an unreliable source of help; [text redacted] could be tasked with monitoring over 70 crews in a night across Canada, in addition to doing other administrative tasks.
 Counsel for Unifor submits that the TRAK report was commissioned with the objective of withstanding the scrutiny of the Tribunal and should be considered with caution. Further, the authors have no prior experience in the CIT industry, and the literature relied upon by the authors in making their assessment is international, and does not readily apply to the Canadian context. Some of the sources, such as Forbes magazine, are tenuous at best. The report states unequivocally that the A/O configuration without additional training and mitigation measures would likely represent an unacceptable danger under the Code; the mitigation recommendations are purely aesthetic and do little to increase the safety of employees.
 Finally, counsel submits that, in the Barber appeal, the stay conditions are similarly inadequate for mitigating the risk of the A/O model and Mr. Barber’s refusal is well-founded. They do not reduce proactively the risks of robbery and assault.
 Counsel concludes by stating that the evidence has established that a serious threat exists, that the A/O model increases that risk and that the mitigation measures are insufficient to reduce that risk to an acceptable level. It follows that the danger is not a normal condition of employment. Counsel seeks an order confirming the direction issued in the Childs case and rescinding the decision in the Barber case, a direction requiring Brink’s to cease using the A/O model. In the alternative, the Union seeks a direction requiring Brink’s to add a third person to runs deemed to be “high risk”, a determination to be made by Brink’s in consultation with the Unifor Policy Committee. In the further alternative, Unifor seeks a direction requiring Brink’s to have blended ATM/Armoured 3-person crews (armoured messenger, ATM technician and a driver from either division) on certain runs, to be determined in consultation with the Union.
Brink’s Canada Ltd. (Reply submissions)
 In response to Unifor’s submissions, the salient and more relevant points raised by counsel are as follows:
 The financial considerations that led Brink’s to adopt the A/O model may equally apply to the Union’s motivation to be against the model, i.e. loss of dues, and they signed collective agreements covering employees working A/O crews and getting a premium for doing so. Although Unifor states that they are committed to making the model as safe as possible, they removed that item from the consultation table at the NHSPC to be dealt with in negotiations.
 Counsel further commented on the Union’s approach during consultations and stressed the fact that, at the end of the day, the Union signed collective agreements (with Brink’s as well as with other competing companies) incorporating a premium for work on A/O crews, which the employer considers to be a recognition that the model is not unsafe, since a responsible union would never let its members work in unsafe conditions.
 Counsel for the employer disputes the Union’s statement that its A/O model is similar to the competitor’s models and that the statistics relating to robbery attempts should be considered against Brink’s model as well. He reiterated the considerable number of measures put in place by the employer to mitigate the risk: specific procedures, including SRAs, training of employees, protective equipment, telecommunications tools, and truck.
 Finally, the technical problems referred to in some of the witnesses’ testimony are at best sporadic and have been corrected when brought to the attention of the employer. For example, the talk delay issue with [text redacted] was resolved with the latest software update, [text redacted] are operational and, when they are reported to malfunction, they are replaced.
 Finally, counsel stressed that the only expert evidence introduced at the hearing was that of Mr. Brien and his two TRAK reports prepared in 2015 and 2017 and that considerable weight should be given to his conclusions on the nature of the risk associated with the A/O model.
Unifor (Reply submissions)
 In reply to the employer’s submissions, counsel for Unifor essentially reiterated her main arguments-in-chief. She submits that, while the TRAK report briefly mentions that the statistics may demonstrate a criminal preference for targeting A/O crews, the authors fail to properly assess this most important consideration in their conclusions. Counsel further points to Mr. Brien’s lack of independence and impartiality required of an expert witness, as evidenced in his “evasive answers” to questions and suggestions which were damaging to Brink’s.
 Counsel for Unifor reiterates that the statistics adduced in evidence, which were collected and analyzed by Brink’s unequivocally lead to the conclusion that criminals are more likely to target A/O crews. Nothing in the evidence establishes that Brink’s has a superior model to other A/O models in the industry and Brink’s claim to that effect is self-serving.
 Regarding increased risk of the driver/guard being attacked while doing the [Text redacted], counsel states that there is no evidence on the record suggesting that criminals will not attack a guard just because they do not have liability on them. Criminals may attack the driver/guard to take him/her as hostage to eventually get to the liability.
 Finally, counsel for Unifor reiterates that the presence of a driver in the truck is a deterrent to criminals. Communication with the driver immediately prior to the crew exiting the service location allows the driver to survey the surroundings and protects the crew from the hazard of robbery or assault pre-emptively. The risk of [text redacted], as argued by the employer, can affect any delivery model and the consequences of [text redacted] in the A/O model are more serious.
 The employees involved in the present appeals engaged in a refusal pursuant to subsection 128(1) of the Code, which reads as follows:
128. (1) Subject to this section, an employee may refuse to use or operate a machine or thing, to work in a place or to perform an activity, if the employee while at work has reasonable cause to believe that
(a) the use or operation of the machine or thing constitutes a danger to the employee or to another employee;
(b) a condition exists in the place that constitutes a danger to the employee; or
(c) the performance of the activity constitutes a danger to the employee or to another employee.
 “Danger” is the key concept in the exercise of the employee’s right to refuse to work and in the exercise of the Minister’s power (through a ministerial delegate) to issue a direction to the employer under paragraph 145(2)(a) of the Code. Section 122 defines “danger” in the following manner:
122. (1) In this Part,
“danger” means any hazard, condition or activity that could reasonably be expected to be an imminent or serious threat to the life or health of a person exposed to it before the hazard or condition can be corrected or the activity altered;
 The Childs appeal relates to a direction and is filed pursuant to subsection 146(1) of the Code:
146. (1) An employer, employee or trade union that feels aggrieved by a direction issued by the Minister under this Part may appeal the direction in writing to an appeals officer within 30 days after the date of the direction being issued or confirmed in writing.
 The Barber appeal is filed by the employee against a decision of absence of danger rendered by the ministerial delegate, pursuant to subsection 129(7) of the Code:
129. (7) If the Minister makes a decision referred to in paragraph
128(13) (b) or (c), the employee is not entitled under section 128 or this section to continue to refuse to use or operate the machine or thing, work in that place or perform that activity, but the employee, or a person designated by the employee for the purpose, may appeal the decision, in writing, to an appeals officer within 10 days after receiving notice of the decision.
 Subsection 146.1(1) of the Code sets out the authority of an appeals officer when a direction or decision concerning a “danger” is appealed. An appeals officer may vary, rescind or confirm the direction or decision:
146.1 (1) If an appeal is brought under subsection 129(7) or section 146, the appeals officer shall, in a summary way and without delay, inquire into the circumstances of the decision or direction, as the case may be, and the reasons for it and may
(a) vary, rescind or confirm the decision or direction […]
 The definition of danger cited above was introduced with amendments brought to the Code by the Economic Action Plan 2013 Act, No.2, S.C. 2013, c. 40, and came into effect on October 31, 2014. The circumstances that gave rise to the refusal and the present appeal occurred subsequently to that coming into force. This new definition of danger must therefore be applied to determine whether the situation described in the evidence presented a danger to the employees.
Were employees exposed to a danger?
 I will follow the same analytical path as in my Dendura decision.
 In two recent appeal decisions, appeals officers have had the opportunity to provide an interpretation on the meaning of the new definition: Ketcheson and Keith Hall & Sons.
 In Ketcheson, the appeals officer conducted an extensive review of the parties’ submissions on the meaning of the new definition. His conclusion was that the current definition of “danger” is different in nature from its predecessors and states, as follows, at paragraph 186:
 In summary, the legislative evolution of the definition of “danger” suggests that, in spite of some similarities in terminology, the 2014 definition is different in nature from its predecessors – both of them. It is neither a reversion to a pre-2014 “imminent danger”, nor is it merely a simplification of the 2000-2014 definition. There are two types of “danger”. They are both high risk, but for different reasons. The new definition adds a time frame for assessing probability. It adds the concept of severity of harm. In the context of the rest of the Code, a “danger” is a direct cause of harm rather than a root cause.
 The appeals officer states further, at paragraph 193:
 The caselaw during the period 2000-2014 contained many expressions for probability: “more likely than not”; “likely”; “reasonable possibility”; and “mere possibility”. What was often left unstated was the time period in which the probability was to be assessed: the day of the work refusal; the foreseeable future on the day of the work refusal; a year from the refusal? Is something likely? It may be almost certain to occur in the next five years, reasonably foreseeable to occur in the next year, but merely possible in the next five minutes. It is meaningless to talk about probability without specifying the time period. Unlike the 2000-2014 definition of “danger”, the 2014 definition, by distinguishing between “imminent threat” and “serious threat”, is adding a time frame for probability.
 And then moves on to define “threat”, as follows, at paragraph 198:
 In the New Shorter Oxford English Dictionary (1993) the word “threat” is defined as: “a person or thing regarded as a likely cause of harm”. Thus, it can be said that based on that definition, a threat entails the probability of a certain level of harm. Some risks are threats and some are not. A very low risk, either because of low probability or because of low severity, is not a threat. Both probability and severity each have to reach a minimum threshold before the risk can be called a threat. It is clear that a low risk hazard is not a danger. A high risk hazard is a danger.
 Likewise, the appeals officer in Keith Hall & Sons stated as follows:
 It also warrants noting that the concept of reasonable expectation remains included in the amended definition. While the former definition required consideration of the circumstances under which the hazard, condition, or activity could be reasonably expected to cause injury or illness, the new definition requires consideration of whether the hazard, condition, or activity could reasonably be expected to be an imminent or serious threat to the life or health of the person exposed to it. In my view, to conclude that a danger exists, there must therefore be more than a hypothetical threat. A threat is not hypothetical where it can reasonably be expected to result in harm, that is, in the context of Part II of the Code, to cause injury or illness to employees.
 For a danger to exist, there must therefore be a reasonable possibility that the alleged threat could materialize, i.e., that the hazard, condition or activity will cause injury or illness soon (in a matter of minutes or hours) in the case of an imminent threat; or that it will cause severe injury or illness at some point in the future (in the coming days, weeks, months or perhaps even years) in the case of a serious threat. It warrants emphasizing that, in the case of a serious threat, one must assess not only the probability that the threat will cause harm, but also the seriousness of the possible harmful consequences from the threat. Only those threats that can reasonably be expected to cause severe or substantial injury or illness may constitute serious threats to the life or health of employees.
 I stated in Dendura that I endorsed the analysis and conclusions reached by the appeals officers in those cases. I remain of that view. The excerpts quoted above correctly summarize the legal concepts that are relevant to the present case. Thus, the legal test to be applied to the facts in order to determine whether the refusing employees were in the presence of a danger (as defined in the Code) may be set out as follows:
- What is the alleged hazard, condition or activity?
- Could this hazard, condition or activity reasonably be expected to be an imminent threat OR a serious threat, to the life or health of a person exposed to it?
- Will the threat to life or health exist before the hazard or condition can be corrected or the activity altered?
 The first question then is to identify the “hazard, condition or activity” that is alleged to be a threat to the life or health of the employee. In the present case, the threat may be described by framing the issue as follows: the risk of suffering serious or fatal injuries as a result of a robbery attack when a crew returns to the armoured vehicle from a client’s location without a driver having remained in the vehicle to watch the surroundings and provide “fresh” intelligence to the crew upon exit.
 I will spend little time on the first element of the threat analysis, i.e. whether the activity constitutes an imminent threat. The appeals officer in Ketcheson aptly described what would be required to establish that an employee is facing an imminent threat, at paragraphs 205 and 206:
 An imminent threat is established when there is a reasonable expectation that the hazard, condition or activity will cause injury or illness soon (within minutes or hours). The degree of harm can range from minor (but not trivial) to severe. A reasonable expectation includes a consideration of: the probability the hazard condition or activity will be in the presence of a person; the probability the hazard will cause an event or exposure; and the probability the event or exposure will cause harm to a person.
 There is no doubt the level of harm from inmate violence can range from minor to severe, but that is not the issue. There was nothing in the evidence put before me to indicate that there was a reasonable expectation that the respondent would be exposed to violence from an inmate on the day of the work refusal and that he would be harmed through inmate violence. The testimony of the respondent was that he was not exposed to an imminent or serious threat on the day of his work refusal. I have given some weight to this statement of the respondent, however, I do not believe that it is dispositive of the matter as was argued by the appellant since it is not clear to me that the respondent understood the meaning of imminent or serious threat as intended by the Code.
 The parties are on the same page on this question in their submissions. Both submit that there was no evidence of an imminent threat being present on the days of the refusals. The employees did not point to any situation which would expose them to a threat to their life or health within minutes or hours on that day. The ministerial delegate never came to that conclusion in either case. I agree with the parties that the activity described above in the circumstances of the present refusals cannot reasonably be expected to be an imminent threat to the life or health of the respondent on the day of the refusals, in the present appeals.
 I turn to the next question, whether the “hazard, condition or activity” could reasonably be expected to be a serious threat to the life or health of the respondent. The combination of the concepts of “reasonable expectation” and “threat” in the statutory definition of “danger” evokes the notion that there must be a reasonable possibility that the hazard will materialize and cause harm to the life or health of the employees. The appeals officer in Ketcheson stated as follows, at paragraph 212:
 In order to conclude that the respondent was exposed to a serious threat to his health or life, the evidence has to show that there was a reasonable expectation that the respondent would be faced in the days, weeks or month ahead with a situation that could cause him serious harm as a result of not being able to carry OC Spray and handcuffs on his person.
 As I stated in Dendura, I agree with such formulation of the question that must be answered to satisfy the requirements of the definition of “danger”. I am also of the view that in order to find that an activity may “reasonably be expected to be a (…) serious threat to the life or health to a person exposed to it”, there must be more than a purely hypothetical threat. A threat is not hypothetical where it can reasonably be expected to occur and result in harm, that is, in the context of Part II of the Code, to cause injury or illness to employees.
 I found in Dendura that the refusing employee was exposed to a serious threat. The employer invites me in the present cases to move away from that conclusion and find that the employees were not exposed to a serious threat on the night of their refusals, since the evidence establishes that the 2-person model does not present an added risk and that there was nothing particular at the sites where the refusals took place to justify the refusals. The employer refers to evidence which I did not have before me in Dendura, namely the most current statistical information showing no attacks on A/O crews for the past 26 months, and the testimony of Mr. Doiron establishing that the 3-person crew and the A/O are equivalent from a safety perspective. Therefore, the employer submits that no serious threat can be found.
 The Union disagrees and argues that employees working on an A/O crew are exposed to a serious threat, essentially because of their increased vulnerability to an attack as a result of the removal of the driver, and that statistics show that attacks occur more frequently on A/O crews than on other crew configurations.
 I am not persuaded by either party’s submissions on that question. I remain of the view that the crew exiting the truck to provide a service to a customer location, and re-entering the truck, is always exposed to a serious threat within the meaning of the statutory definition. However, it is my view that this conclusion is true regardless of whether they work as a 2-person A/O crew, or a 3-person “traditional” crew. In both cases, the threat - the source of the hazard - is completely independent of the employer and is the result of unpredictable and criminal human behaviour. In other words, the execution of CIT activities inherently presents serious threats to employees, and that threat cannot be eliminated. The real question in my view is the extent to which the features of the model and the mitigation measures introduced by the employer make the danger a normal condition of employment. It is in that context that the Union’s argument of increased vulnerability should be examined, in my view. I will come back later on this point.
 Going back to the legal test to establish whether a serious threat exists and as I discussed in Dendura, the determination of whether a threat is a real possibility as opposed to a remote or hypothetical possibility is not always an easy task. It is a matter of fact in each case and will depend on the nature of the activity and the context within which it is executed. Statistical information is relevant to make an informed factual finding on that question, although in the final analysis, it involves a question of appreciation of facts and judgement on the likelihood of occurrence of a future event, in the present case an event that is linked to unpredictable human behaviour (Dendura).
 The employer submits that, according to the evidence, the probability that an A/O crew member will be ambushed after exiting the customer location and returning to the vehicle is low and does not constitute a threat.
 The TRAK (2015) report, states as follows at page 46:
Three specific risk factors were revealed by our research.
The first is risk of an attack on an armoured vehicle crew on a yearly basis.
Statistics show that this event has a medium to high probability potential, generally with a minor impact. In this particular event, the impact is considered at the company’s level.
Nationwide, industry statistics show that between 2000 and 2015, Brink’s has been targeted at least once in 7 years out of 15. This places the likelihood of an attack on a yearly basis in the medium to high range.
 The TRAK (2015) report also states as follows at page 26:
The data covering the 2000-2015 period (most reliable statistics) reveals that 99 attempts were made across the country. Of those, 69 (70%) were made against 2 crew members configuration and 2 All Off crews. 17 of those 99 attempts (17%) were made against Brink’s crews during that period although the company had not deployed the ALL OFF crew configuration. This might indicate a preference on the part of criminals for the ALL OFF crew configuration when planning an attack.
 The statistical breakdown of these incidents shows that thirty-nine percent (39%) of incidents occurred on location, twenty-eight percent (28%) occurred exiting the vehicle, and thirty-three percent (33%) of incidents occurred returning to the vehicle. Their statistics show that eighty- four percent (84%) of these incidents were organized attacks. The type of force used in the attacks is broken down in the statistical information, namely fourteen percent (14%) pepper spray, twenty-one percent (21%) handgun, nine percent (9%) long gun, two percent (2%) simulated, and fifty-four percent (54%) being classed as other (Ramming-Physical-NA-NS).
 In light of this evidence, the question for me to decide is whether there is a reasonable possibility that employees could be subject to a robbery attempt when they return to the armoured truck after finishing their task in the client location. The employer has not persuaded me that employees are not exposed to a serious threat when they are engaged in the activity of providing armoured car services to clients in general, and while they are engaged in the activity that is central to the present appeals. In my view, the activity of carrying cash-in-transit inherently entails the real possibility that there will be attacks and robbery attempts on employees, whether working on a 2-person or 3-person crew. The statistics cited above speak for themselves. In fact, it is the very essence of the work performed by employees in that industry. The raison d’être of armoured car services is precisely to protect cash and valuables that are in transit, from robbers and criminals and that work constantly exposes employees to the risk of being robbed. The risk is not incidental to the work, as would be the risk of slipping, or injury in handling material, etc.; it is central and inherent to the activity itself.
 The regional safety officer (predecessor to an appeals officer under the Code) stated as follows in Loomis Armoured Car service Ltd. And Canadian Brotherhood Railway, Transport and General Workers, Local 266A, Decision 93-008, at paragraph 21:
 No one would disagree, I would venture, with the premise that persons employed in the operation of armoured cars are exposed to risks on a daily basis. [...] the Canadian Labour Relations Board noted that “the risk of robbery or assault is part of armoured car service employees lives''. One can therefore conclude that danger is inherent in the operation of armoured cars, a situation which is recognized by the Code and which precludes employees from refusing to work solely because of the risk of robbery or criminal attack. However, one must also ask at what level or under which circumstances does this inherent danger become unacceptable?
 Although the probability that the hazard materializes is characterized as low by the employer, particularly in light of the most recent statistics showing no attacks on Brink’s A/O crews in the past 26 months, I remain convinced that the possibility is a real and substantial one and that is has occurred with some regularity since 2000. It is not a purely hypothetical or speculative scenario.
 In Verville v. Canada (Correctional Services), 2004 FC 767 (Verville), Gauthier, J. stated her view on the question of the reasonable expectation that a hazard materializes. I believe her thoughts continue to be relevant to the application of the new definition of danger:
(…) the injury or illness may not happen immediately upon exposure, rather it needs to happen before the condition or activity is altered. Thus, here, the absence of handcuffs on a correctional officer involved in an altercation with an inmate must be reasonably expected to cause injury before handcuffs are made available from the bubble or through a K-12 supervisor, or any other means of control is provided.
 Also, I do not believe that the definition requires that it could reasonably be expected that every time the condition or activity occurs, it will cause injury. The French version “susceptible de causer” indicates that it must be capable of causing injury at any time but not necessarily every time.
 In that respect, I do not believe either that it is necessary to establish precisely the time when the potential condition or hazard or the future activity will occur. I do not construe Tremblay-Lamer's reasons in Martin above, particularly paragraph 57, to require evidence of a precise time frame within which the condition, hazard or activity will occur. Rather, looking at her decision as a whole, she appears to agree that the definition only requires that one ascertains in what circumstances it could be expected to cause injury and that it be established that such circumstances will occur in the future, not as a mere possibility but as a reasonable one.
 The Court goes on to explain further:
 (…) If a hazard or condition is capable of coming into being or action, it should be covered by the definition. As I said earlier, one does not need to be able to ascertain exactly when it will happen. The evidence is clear that in this case, spontaneous assaults are indeed capable of coming into being or action.
 In the risk assessment report concerning the routine issue of restraint equipment dated November 8, 2001, the potential risk of confrontation between correctional officers working in the living units and the inmates is said to be high (page 20) and the risk of assault is of low frequency but high severity (page 21) . As indicated, Warden Urmson confirmed that such assaults were expected to occur and that was why handcuffs were available in the bubble.
Thus, if those assaults could reasonably be expected to cause injury, they will come within the definition of danger. However, if that danger constitutes a normal condition of his employment, the employee will not have the right to rely on it to refuse to work (s. 128(2)(b)). But, that is very different than saying that unpredictability of inmates' behaviour is alien to the concept of danger in the Code.
 I am therefore of the view that the performance of the activity described above could reasonably be expected to be a serious threat to the life or health of the respondent. There is no question that the risk of harm from a potential attack can be severe, even fatal, even if I accept the employer’s characterization of the probability of the risk as low. The litany of mitigation measures set out by the employer in its evidence and submissions are indeed designed to reduce the risk of injury to a minimum, but will not prevent attacks from occurring. I consider that the discussion on the effect of those measures is more relevant to the question of whether the hazard/activity is a normal condition of employment, as opposed to the determination of whether employees performing the task are facing a reasonable possibility of being exposed to a serious threat, and as a result, exposed to a danger as defined in the Code.
 I find the following excerpt from Martin-Ivie v. Canada (Attorney General), 2013 FC 772 (Martin-Ivie), at paragraph 47, where the Court discusses the “low frequency, high risk” principle, to be relevant to the present analysis:
 As for the Appeals Officers’ decisions, they apply the principle not in determining whether a “danger” exists, but, rather, in assessing whether a work refusal is permitted under paragraph 128(2)(b) of the Code, which prohibits work refusals – even if a “danger” exists – in situations where the danger is a normal condition of the refusing employee’s employment. These cases, as well as Verville, establish that before a risk may be said to constitute a normal condition of an employee’s employment, the employer must have taken all reasonable steps to mitigate it. In such circumstances, the reasonableness of the steps taken by the employer will depend in part on the gravity of the risk: the greater the risk the further the employer must go to mitigate it (see e.g. Armstrong at paras 62-63; Éric V at paras 295-297, 301). Thus, the “low frequency, high risk” principle is applied to the assessment under paragraph 128(2)(b) of the Code but not to determining whether a danger exists. Moreover, in applying this principle, the required analysis under the Code necessarily involves consideration first of whether a “danger” exists and then, if so, consideration of whether such “danger” is a normal condition of the employee’s employment.
 It also seems evident to me that should the possibility of an attack materialize, the employee would be exposed to the hazard before the activity could be altered, notwithstanding the measures put in place. The appeals focus on one particular aspect of the work method applied to the A/O delivery model: on the footpath returning to the armoured truck after coming out of the customer’s premises. The weight of the evidence is that this is the area where employees, whether on 3-person or 2-person crews, are the most exposed and vulnerable.
 I therefore remain of the view that the activity of returning to the truck from the client’s location exposes the employees to a serious threat of injury, and constitutes a danger to the employees concerned, as defined in the Code.
Normal condition of employment
 This takes me to the final part of the analysis: are the activity in question, and the danger that it presents a normal condition of employment? If that question is answered in the affirmative, employees cannot invoke the right to refuse under section 128 of the Code and accordingly, the ministerial delegate could not issue a direction based on a finding of danger in those circumstances.
 Subsection 128(2) reads as follows:
128.(2) An employee may not, under this section, refuse to use or operate a machine or thing, to work in a place or to perform an activity if
(a) the refusal puts the life, health or safety of another person directly in danger; or
(b) the danger referred to in subsection (1) is a normal condition of employment.
 I turn again to the Verville judgment to provide guidance on that question. At paragraphs 52 to 57, the Court states as follows:
 Turning now to the conclusion in ii) at paragraph 40 above that the risk was inherent to the applicant's employment, the applicant concedes that his job description involves a risk of possible hostage taking, injury or danger when dealing with violent and hostile offenders. But he argues that the order given to him on September 24, was a variation of his normal conditions of employment and constitutes an increase of the risk or danger described above. The applicant relies on the Public Service Staff Relations Board's decision in Fletcher v. Treasury Board (Solicitor General Canada - Correctional Service),  C.P.S.S.R.B. No. 58; Danberg and Treasury Board (Solicitor General Canada),  C.P.S.S.R.B. No. 327 and Elnicki v. Loomis Armored Car Service Ltd, 96 di 149, CLRB Decision No. 1105, in which the Board acknowledged, in the context of refusals to work by correctional officers and security guards, that even though risk of injury or death was a normal condition of employment for these employees, an increased danger resulting for example from a change in the employer's policy (such as minimum staffing), was not automatically excluded under paragraph 128(2)(b) .
 There is no indication in the decision under review that the appeal officer considered this argument. His finding appears to be based on the simple fact that a risk of assault is always present in an environment such as the Kent penitentiary. As mentioned, he could not evaluate if the increased risk of injury was a normal condition of employment because he did not consider it to be more than an unproven hypothesis.
 The customary meaning of the words in paragraph 128(2)(b) supports the views expressed in those decisions of the Board because "normal" refers to something regular, to a typical state or level of affairs, something that is not out of the ordinary. It would therefore be logical to exclude a level of risk that is not an essential characteristic but which depends on the method used to perform a job or an activity. In that sense and for example, would one say that it is a normal condition of employment for a security guard to transport money from a banking institution if changes were made so that this had to be done without a firearm, without a partner and in an unarmoured car?
 In my opinion, the decision under review is unreasonable, in particular in that the appeal officer failed to consider evidence on a core issue on which his final conclusion rests.
[Underlining and bold added]
 The issue therefore is whether the employer has taken appropriate measures to guard against the danger identified above, and to reduce it to an acceptable level such that the activity and the residual hazard that it presents (the danger) can be said to be a normal condition of employment, as provided in paragraph 128(2)(b) of the Code. I must therefore consider the particular features of the A/O model and determine whether the measures implemented by the employer mitigate the risk to an acceptable level and fulfill the employer’s obligations under the Code. Any residual hazard remaining - being subject to an attack - would be a normal condition of employment.
 In P&O Ports Inc. and Western Stevedoring Co. Ltd. v. International Longshoremen’s and Warehousemen’s Union, Local 500, 2008 FC 846, the Federal Court set out the following analysis with regard to the question as to whether a danger constituted a normal condition of employment:
 The Appeals Officer held as follows at paragraph 152:
 I believe that before an employer can say that a danger is a normal condition of work, he has to identify each and every hazard, existing or potential, and he must, in accordance with the Code, implement safety measures to eliminate the hazard, condition, or activity; if it cannot be eliminated, he must develop measures to reduce and control the hazard, condition or activity within safe limits; and, finally, if the existing or potential hazard still remains, he must make sure that employees are provided with the necessary personal protective equipment, clothing, devices and materials against the hazard, condition or activity. This of course, applies, in the present case, to the risk of falling as well as to the risk of tripping and slipping on the hatch covers.
 Once all these steps have been followed and all the safety measures are in place, the "residual" hazard that remains constitutes what is referred to as the normal condition of employment. However, should any change be brought to this normal employment condition, a new analysis of that change must take place in conjunction with the normal working conditions.
 For the purpose of this case, I find that the employers failed, to the extent reasonably practicable, to eliminate or control the hazard within safe limits or to ensure that the employees were personally protected from the hazard of falling off the hatch covers.
 Such analysis stems from sections 122 and 122.1 of the Code, which set out the purpose of the Code and the hierarchy of preventive measures an employer is required to implement:
122.1 The purpose of this Part is to prevent accidents and injury to health arising out of, linked with or occurring in the course of employment to which this Part applies.
122.2 Preventive measures should consist first of the elimination of hazards, then the reduction of hazards and finally, the provision of personal protective equipment, clothing, devices or materials, all with the goal of ensuring the health and safety of employees.
 At paragraph 214 of the Ketcheson decision, the appeals officer states as follows:
 While the evidence presented in this case has made clear that the respondent is exposed to violent inmates in the performance of his regular duties and that the possibility of an assault by an inmate is always present in a correctional institution, I was not presented with any evidence that would serve to demonstrate how the carrying of handcuff or OC Spray would prevent assaults on CMs or would decrease the level of violence from inmates particularly in light of the fact that these two pieces of equipment are already provided to CO officers. Moreover, the appellant has provided evidence to establish that numerous measures are in place at the Millhaven institution to mitigate against risk to CMs and all other staff in the performance of their duties.
 As I have noted earlier, the danger identified by the refusing employees and the ministerial delegate focusses on one specific step in the A/O process: returning to the armoured truck and the greater risk of being attacked because employees no longer have information of suspicious persons or activities occurring outside while the crew is inside the customer's location, as a result of the absence of the third person in the truck.
 As I held in Dendura, I believe that the A/O model and all of its features must be looked at in their entirety in order to make an informed determination on the specific concern which is at the source of the appeals. I am not persuaded by the argument that removing the third person to keep watch while the crew is inside the premises and to alert them if they suspect criminal activity creates in and of itself an increased risk of attack and injury. The evidence does not establish that fact with any degree of certainty. One could reach that conclusion if the removal of that person had not been accompanied by other measures purporting to mitigate the risks of this inherently dangerous activity. In fact, this is the conclusion offered by Mr. Brien in his report (TRAK (2015)) after taking into account risk assessment/management theories and the learnings of the Securicor decision. In Dendura, I stated the following:
In his analysis, Mr. Elliott relied on the appeal decision rendered in Brazeau et al. v. Securicor Canada Limited, Decision No. 04-049 (December 16, 2004) (Securicor). The appeals officer in that case found that the employer’s decision to change the delivery method from a 3-person crew to a 2-person crew (no driver staying in the truck) presented a danger to employees and was not a normal condition of employment. In his opinion, the presence of the third guard (driver) minimized the risk of successful attacks because it could radio information to the other crew members when they were at the point of returning to the armoured truck from the client location, and removing the third person increased the risk of being ambushed and subject to serious injury.
On its face, the issue is seemingly similar to the instant case. However, as I have stated earlier, the assessment of a hazard is often driven by the particular context within which the activity is taking place. Thus, it is important to note the factors set out in the appeals officer’s reasons in Securicor to reach such a conclusion: (i) the employer implemented the change without having put in place customized procedure to take into account the change in work method -- for example, the instructions to employees in case of distress included the driver leaving the premises to secure the cash and remove the possibility of hostage-taking, where in fact there was no driver in the truck; (ii) the communications tools were proven to be inadequate and deficient; (iii) the refusal took place on a particular site, where suspicious-looking individuals had been observed; (iv) the training given to employees was determined to be deficient, in that it only lasted less than one hour and was not mandatory (the appeals officer expressed “astonishment” at the systemic lack of training); (v) the employer had made the change without consultation with the health and safety committee; (vi) the hazard assessment was found to be inadequate, was generic and not site-specific and omitted a number of factors. In brief, a perfect storm in the making.
 The outcome in Securicor was essentially driven by the appeals officer’s finding that the employer had simply reduced the staff of the crew, without consultation with those affected and without focussing on mitigation measures related to this particular model.
 Mr. Brien sets out the areas that the employer must address in that respect, where Securicor had failed over a decade before. In conducting his risk assessment, he concludes that the overview of the literature does not provide any evidence that the number of armed guards or their specific role have a significant impact in explaining the frequency, severity and patterns of attacks (TRAK (2015), page 25). While statistics illustrate that 70% of the attacks between 2000 and 2015 were on 2-person crews, the contextual evidence is that a 2-person crew configuration had become the standard in the CIT-ATM industry over that period (testimony of Messrs. Doiron, [text redacted]). The likelihood that attacks will occur against 2-person crews is statistically higher, but not solely because they are 2-person crews. In fact, 30% of the attacks were on Brink’s 3-person crews during that period and there have been attacks on 4 and 5-person crews as well (testimony of Messrs. Doiron and [text redacted]). There have been no attacks on Brink’s 2-person crews over the past 26 months.
 The Union introduced a Report on Canadian Armoured Industry Statistics (2004-2014). That report was prepared by the employer in 2014 and, in short, concludes that the Garda/G4S 2-person A/O model has “clearly resulted in greater risk and exposure for employees providing the service…” It viewed Brink’s’ operating model (at the time, the 3-person crew) as providing a safer environment for employees compared to Garda’s model.
 That evidence did cause me some concern. However, in the final analysis, I do not consider it to be inconsistent with the Brink’s’ assessment of its current A/O model, as they – [text redacted] in particular - have developed it. The observations in the report were based on Garda’s model, not on the A/O that was developed by Brink’s, once the decision was made to introduce that delivery model in that company’s operations. In fact, Mr. Brien acknowledges in his evaluation that a 2-person A/O crew without the many additional protective measures to support it, would likely not meet the test of safety. [Text redacted] also explained the lower number of attacks on Brink’s in part on account of the company’s reputation to be a “hard target”, namely because of its effective training of employees and protective equipment which is also mentioned in the report.
 I also note that there are other models than the 3-person crew and A/O, as illustrated in the Pogue v. Brink’s Canada Ltd, 2017 OHSTC 27, where the two employees were operating a 2-person “One Off” model, whereby the driver of the armoured vehicle stays inside the truck while the armed messenger performs CIT deliveries and pick-ups. The danger invoked by the refusing employee in that case had to do with the greater affluence during the Christmas season at a shopping centre and the messenger’s inability to assess the crowd in the absence of a guard along with him, i.e. a third person. The appeals officer confirmed the decision of absence of danger and dismissed the appeal. The decision does not address whether the “One Off” 2-person crew configuration is inherently unsafe - let alone the A/O model - but found that Mr. Pogue was not exposed to a serious threat in the circumstances. I take this case as an illustration that the removal of the third person on a crew has not been found to necessarily constitute an elevated risk, beyond normal conditions of employment.
 In his expert report, Mr. Brien assesses the risk of attack and injury as being overall low in the armoured car industry. As stated in Martin-Ivie, this assessment is relevant in the analysis under 128(2)(b) of the mitigation measures. The Teamsters Union was consulted in the implementation of the A/O model and Mr. Doiron had no issue with the analysis of the risk presented by Mr. Brien and Brink’s management, as long as mitigating and control measures were put in place. I do not share counsel for Unifor’s view that Mr. Brien is not an independent expert witness. I found his risk assessment to be fair and honest, with appropriate nuances, and not subject to any undue influence by Brink’s management. I found him to be a credible witness, and a person for whom professionalism and integrity is critical.
 Accordingly, I must assess the A/O model as a whole, based on the evidence placed before me. Not surprisingly, the Brink’s case is substantially the same as the case it introduced in Dendura, purporting to show the manner in which the hazard is addressed under a 2-person crew configuration. However, I acknowledge that in the present appeals, unlike Dendura, the refusing employees were represented by their union Unifor, who forcefully opposes the 2-person A/O model as being unsafe and who presented a stronger case. I will discuss that evidence as I analyze the A/O model.
 Brink’s submits that such protection is ensured by an architecture of means involving physical assets [text redacted] and employees, who are provided with protective equipment [text redacted]. All those layers of protection are designed, in their fundamental purpose, to dissuade criminals from attempting a robbery by making such an attempt more difficult or risky for them, or to reduce the risk of injury to employees in case of an attack. It cannot be overstated that the risk of being attacked is an ever-present feature of the work of an armoured car employee.
 I will go over the measures outlined in the evidence.
 First, there are the technical features on A/O trucks, which are described in the testimony of [Text redacted]. The truck configuration is significantly different from trucks used in 3-person crews. They present layers of security measures that are designed to make access to the truck more effective for the crew, and more difficult for criminals. I will highlight the salient and most relevant features of the vehicle that distinguish it from vehicles used in a 3-person crew model. [Text redacted], thus allowing the employees to continue to be alert to their surroundings while opening the doors. [Text redacted].
 Witnesses called by the Union have testified on the frequent malfunctioning [Text redacted]. On balance and considering the totality of the evidence, I accept the employer’s response that those problems are sporadic and are dealt with as they occur. The evidence shows that where that situation occurred before leaving the branch, the truck – and the A/O crew configuration - would not be used. There were no [Text redacted] involved on the nights of the refusals. The Union also argues that those measures are not preventive (pro-active), but reactive, once the attack has occurred. Looked at as a whole, I am of the view that those measures constitute an important element of the infrastructure of protection for employees.
 Second, employees are provided with personal protective equipment (PPE), such as [text redacted] and firearms. There was no serious suggestion that such protective equipment is inadequate.
 Third, employees are provided with communication tools and control devices: [text redacted]. Employees must wear those devices on them at all times, allowing them to be in communication with each other and [Text redacted]. The driver/guard can trigger the alarm on the truck if he suspects something unusual or if not comfortable with the surroundings when he/she comes out of the point of service.
 The Union has provided examples of cases where the communication tools have been deficient and unresponsive. Mr. Barber [text redacted]. I accept that technology is not 100% foolproof and may fail from time to time, for various reasons. In my view and on balance, the weight of the evidence is that these matters are addressed and corrected as they occur and the A/O delivery would not be permitted in case of persistent malfunction of the equipment. Furthermore, it seems to me that all those tools would have to be deficient at the same time and the malfunction coincide with the very moment of a robbery attempt in order to present an unacceptably elevated risk. Otherwise, the threat becomes, in my view, a hypothetical or speculative possibility only. It would not be appropriate for me to determine the issue raised by the appeals on the assumption that the equipment may be defective. That being said, the importance of ensuring the good functioning of the equipment to fullest extent possible should not be minimized: instances of malfunction should be reported, reviewed and addressed by management. Had a situation of general tolerance been demonstrated, I would agree that the evaluation of the risk and hazard could be different. It is not the case here.
 Fourth, the employer has developed specific A/O operational procedures, including the [text redacted] and SRAs. It is not necessary to repeat the detailed [text redacted] procedure described by [text redacted] when crews return to the truck. The Union challenges the effectiveness of the [text redacted] in replacing the eyes of the third person. Such an evaluation is difficult to make, as we are dealing with non-scientific and somewhat intangible concepts: the assessment of security measures to guard against unpredictable criminal behaviour. The employer’s witnesses testified adamantly that the [text redacted] is an adequate measure to ensure the safe re-entry of the crew in the truck. They state that the driver/guard has a better opportunity than the third person in the truck to scan the environment and the surroundings, where the third person only has a partial view of the surroundings. In fact, it was mentioned that robbery attempts have taken place near the truck in the past without the driver even noticing them. [Text redacted]. Employees are more alert and communicate better with each other and must work with heightened vigilance rather than strictly rely on the observations of a driver. Mr. Doiron echoes that conclusion.
 That is not to say that drivers have not played an important role in the safe operation informing their colleagues coming out of the premises. They clearly have. But they have done so under a model which is entirely based on the driver/third person keeping a watch and providing information to the crew, who in turn entirely relies on that information to exit the location. Both parties agree that employees should have relevant and up-to-date information about the risks they might encounter when returning to the armoured car. Where the parties part company, though, centres on how this should be done. The A/O is a different model, with different components, that must be examined on its own value.
 Brink’s submits that it is highly unlikely that a guard doing a [text redacted] before allowing the messenger to return to the vehicle will be subject to an attack. There has been no report or evidence of a guard being attacked or harmed when he was not carrying any liability or when he was not in proximity to the messenger who was carrying liability. Understandably, as the guard does not carry any valuables and is armed, he is an unlikely target for robbers. In other words, it is statistically unlikely that a guard will be attacked or exposed to a danger while performing [text redacted] and I find this assertion to be supported by the evidence.
 Once the driver/guard [text redacted], who is also armed, the crew is in no worse or better position than a 3-person crew. [Text redacted].
 Mr. Childs withdrew his refusal and is no longer finding the A/O to be unsafe, as he is more comfortable with it. [Text redacted] testified that he is generally comfortable working on the A/O model. Mr. Armstrong’s concern with the A/O is based on his members’ reports, as he has no experience in working in the CIT industry. I find that the weight of the evidence of persons who have knowledge and experience in the CIT business supports the finding that the compulsory [text redacted] is an appropriate procedure to protect the crew when it exits a location and proceeds back to the truck.
 [Text redacted] has testified that the driver/guard is exposed when he exits the truck and performs his check of the premises. I will simply note that this is not the reason on which the employees have based their refusals in the present cases. But even so, the same considerations outlined above would apply. In addition, the driver/guard is being watched by the messenger who remains in the truck during this process and who may act as needed.
 Turning to the risk assessments of the sites, one is prepared for each point of service. They are meant to provide an assessment of the specific risks with each site, so as to inform the crew before they reach the site of the particular things that they should be alert to when servicing the site. This was found to be lacking in the Securicor case. Risk assessments of the sites are prepared jointly, by local representatives of both parties. The process for preparing them was described by a number of witnesses. That process is not a perfect one and, in the final analysis, there may be disagreements between the parties on the level of risk or safety measures to apply to a particular site. In such a case, the employer makes the final call, as the party responsible for addressing health and safety matters under the Code and pursuant to its management rights. Be it as it may, I was presented with no evidence that could lead me to a finding that the risk assessments of the site process was applied in a perfunctory manner by the employer, or not taken seriously by the parties.
 Finally, the employer provides training to employees and ensures an appropriate level of supervision. The evidence establishes that employees are subject to mandatory training, including the basic introductory employee module referred to as the [text redacted]. Employees are also given the specific [text redacted], which is an 8-9 hour in‑class module followed by an on-the-job application with a crew shadowing process and feedback for their first few shifts. The employer also provides risk assessment and related training to employees, to ensure that employees have a good understanding of the factors that may affect the risk and control measures. A considerable amount of training documentation and presentations was entered in evidence. With the exception of Mr. Barber, the weight of the evidence is that the training was viewed as being adequate. The employer also conducts sporadic street inspections meant to ensure that employees are following procedures correctly. [Text redacted] testified that the number of street inspections will be a priority in the next year as an indicator of local performance of management. I have heard no reliable evidence to suggest that the training provided to employees was inadequate or perfunctory. Employee training was found to be severely lacking in the Securicor case.
 In Dendura, I took into account the fact that the Teamsters Union had not opposed the introduction of the A/O model, albeit in conjunction with the mitigating measures described by the employer. The employer is asking me to infer from Unifor’s acceptance of an A/O premium, at the bargaining table, in October 2017 that it implicitly consented to the use of the A/O model as being safe. I am not prepared to reach that conclusion, in spite of what may appear as an action inconsistent with the Union’s position. The Union’s position was clearly explained by Mr. Armstrong, who felt that Brink’s would proceed with the A/O crew configuration regardless, and I have no reason to doubt the sincerity of the Union’s concern with the A/O model. The reference to the press release by Unifor’s president that they “would never sign an agreement that would place employees in danger” and yet signed such an agreement is perplexing, but I am prepared to consider this to be collective bargaining rhetoric rather than an admission that the A/O model is considered safe.
 The fact remains however that the Teamsters, who represent two-thirds of Brink’s employees in Canada, does not consider the A/O model to present a higher level of danger than the 3-person crew model, the two configurations being “equivalent” in the level of risks, as stated by Mr. Doiron. Mr. Doiron testified that the A/O is now the industry standard for CIT/ATM services. In fact, he ventured to say that the majority of employees of Garda, which has used the A/O model since 2005, are unfamiliar with the 3-person crew and the 2-person A/O is the only model they know. I note that Unifor also represents employees of Garda, who have been working on A/O models for some time and for whom Unifor concluded collective agreements that include provisions dealing with the A/O model operated by that employer. Mr. Doiron stated that Brink’s’ A/O model is the best of its kind in Canada. He was consulted on the mitigation/control measures developed by the employer and provided feedback, which the employer took into account. All in all, his union is not opposing the A/O model operated under those conditions and this view supports the employer’s theory. This evidence certainly bears some weight on my analysis of the issue and the judgement that I am tasked with making on the safety of the A/O model.
 I note that one province in Canada has passed a regulation aimed at the armoured car industry and set the minimum crew configuration to be two persons (s. 16(4) of the Private Investigators and Private Guards Regulations, Nova Scotia).
 All things considered, I remain persuaded, as I was in Dendura, that the implementation of the Brink’s’ A/O model has been carefully planned and the hazards duly considered by management, with the assistance of Mr. Brien, a person with considerable experience in the field of risk assessment in a context of security, policing and criminal activities. The hazard present when the crew returns to the truck is not ignored or left unaddressed. Rather, it is dealt with in a different manner, as outlined above. As I concluded in Dendura, at paragraph 182:
 The notion of whether a danger exists or is a normal condition of employment is in large part a matter of fact and context in each particular case. Evaluating the risks in the context of unpredictable human behaviour is not an exact science and is largely a matter of judgement. The judgement must be an informed one, and must rest on the totality of the evidence. This evidence includes statements made by witnesses who have considerable experience and expertise in the area of risk assessment relating to security, cash-in-transit and factors linked to criminal activities. In conclusion, I have no basis in which to find that the 2-person crew working under Brink’s’ A/O delivery model as it was presented to me, increases the risk of injury beyond the employees’ normal conditions of employment. In light of the very nature of the work, the residual hazard - the risk of being attacked - that remains after the employer has implemented the mitigating measures described above, is a normal condition of employment.
 Safety is not static. It will evolve with time and technological developments. The landscape may very well be different a few years from now, as industry statistics change. It may be that down the road, there will be compelling evidence that the mitigation measures should be revisited and increased. The employer has the responsibility under the Code to ensure that his measures remain relevant and adapt them to a changing landscape. It must do so diligently and in consultation with its policy and workplace committees and seek continuous improvement. The alternate remedies sought by Unifor in its submissions are examples of measures that could be discussed by the parties and explored, in the spirit of continuous risk mitigation and injury prevention mandated by the Code, regarding the dangers inherently associated with CIT services.
 However, at this juncture and in light of the evidence before me, I am unable to make a finding that employees are exposed to an unacceptable risk, given the combined effect of the measures implemented by the employer to mitigate the hazard. The opinions from users of the model are divided on the question, as reflected in the evidence. The determination of the appropriate level of risk acceptability in that particular context is not an easy task. I believe that it would be inappropriate for appeals officers to second-guess the safety measures taken by the employer in such a specialized area of activity, unless they are presented with compelling evidence - including expert testimony - that the measures are clearly irrelevant or inadequate. We are some distance from the scenario evoked at paragraph 55 of Verville - quoted in bold characters earlier in the present reasons – as an illustration of unacceptable risk in the armoured car industry.
 This is particularly true in light of the most serious financial and workforce consequences of a cease and desist order against the use of the A/O model altogether, as sought by the Union in the present cases, and in light of the evidence establishing that such a crew configuration is now the standard across the industry in Canada and is operated by employees represented by the Union involved in the present proceedings.
 I already voiced my concern with this situation, at paragraph 42 of my reasons in Brink’s v. Childs and Unifor (STAY):
 (…) In fact, I am somewhat troubled by the piecemeal approach adopted by the Labour Program on an issue which singularly affects the whole armoured car industry. The issue appears to have been present for some time and, in all fairness, ought to be looked at in a more comprehensive manner.
 For the reasons set out above, my conclusion is therefore that, in the absence of unusual, exceptional or abnormal circumstances, the threat to the life or health of employees as a result of potential robbery attacks when working on a 2-person A/O delivery model, as described in the present reasons, is a normal condition of employment.
The Childs appeal
 Turning back to the specifics of the Childs appeal, Mr. Childs has explained in his testimony the context in which he exercised his right to refuse. As the local union representative, he was sensitive to the fact that concerns had been expressed by some of his colleagues about the A/O model that Brink’s was about to implement. He was cognizant of his union’s stand on the safety of the model but also felt the pressure of management, whom he knew were quite keen in introducing the A/O in their operations. He was caught between a rock and hard place. He refused to work and referred to the absence of the “extra pair of eyes” (driver/third person) to provide intelligence to the crew upon exiting the service location. There was nothing unusual that night at the CIBC Hawkesbury location that placed his life or health in jeopardy, nor did Mr. Childs [text redacted] that there was. His demeanour as seen on the video is consistent with the conclusion that he was not facing a specific or imminent threat. His partner, [text redacted], did not participate in the refusal. The fact that Mr. Childs had received the appropriate training [text redacted] is not questioned. His equipment functioned correctly on the night of October 25, 2016. An SRA had been completed for the site and was available, and Mr. Childs had taken cognizance of its content. And on July 9, 2017, Mr. Childs voluntarily withdrew his refusal and stated in his testimony that he was comfortable with the A/O configuration.
 Ministerial Delegate Jenkins did not attend the refusal site. After taking cognizance of the employees’ concerns, he concluded that the basis of the refusal was the absence of a third person in the truck to provide information to the crew upon exiting the bank. That situation was identical to the Edmonton situation in Dendura and he relied on subsection 129(3.1) of the Code to render a decision of danger and issue the direction under appeal. We now know the outcome of the Dendura appeal, where I rescinded the direction. I have no reason in the present Childs appeal to be of a different view, for the reasons set out above. This outcome illustrates that reliance on subsection 129 (3.1) by ministerial delegates is fraught with difficulty, as appeals officers must inquire into the circumstances of the direction and give the parties an opportunity to be heard, and may not readily rely on that provision because of the quasi-judicial nature of the appeal process.
The Barber appeal
 Likewise in the Barber appeal, the reason for the refusal was the absence of the third person/driver in the truck. There was nothing unusual that night at the Hazeldean location that placed his life or health in jeopardy, nor did Mr. Barber claim that there was. [Text redacted], while he was [text redacted]. This problem was reported and it was addressed by a modification of the SRA for that site, in due course. His partner, [text redacted], did not participate in the refusal. The fact that Mr. Barber had received the appropriate training [text redacted] is not questioned. Mr. Barber’s criticism of the quality of the training is, in my view, unsubstantiated, considering the entire evidence on that point. His equipment functioned correctly on the night of April 3, 2017. An SRA had been completed for the site and was available for him to consult. None of the many issues he described having had with the truck or his electronic equipment were present on the night of the refusal.
 Ministerial Delegate Jenkins issued a decision of absence of danger on the basis that Brink’s had applied the conditions set out in my Order to stay the direction in the Childs appeal and after concluding that the situation essentially raised the same issues as the Childs appeal, with no unusual or abnormal circumstances relating to the site of the refusal. Given my finding outlined earlier that the Brink’s A/O model presents a danger that is a normal condition of employment, which precludes employees from refusing to work in accordance with paragraph 128(2)(b) of the Code, his decision of absence of danger cannot stand and must be varied.
 For the above reasons,
- The direction issued by the ministerial delegate in the Childs appeal, OHSTC File No. 2017-02, is rescinded.
- The ministerial delegate’s decision of absence of danger in the Barber appeal, OHSTC File No. 2017-12, is varied.
Case No.: 2017-02
Brink’s Canada Ltd., Appellant
Michael Childs and Unifor, Respondents
Case No.: 2017-12
Kevin Barber, Appellant
Brink’s Canada Ltd., Respondents
Whereas the Tribunal’s record comprises confidential and proprietary information and that during the course of the present appeal proceedings, such information will be shared between the parties and may be disclosed to the Appeals Officer as evidence;
Whereas such information reveals practices, processes and procedures related to the employer’s armoured car operations and safety and security measures designed to protect employees against possible assaults and robbery attempts, and are central to the issues that will be addressed in the appeal proceedings;
Whereas the parties have agreed to protect such information from public disclosure and have signed a “Non-Disclosure Agreement”, dated December 20, 2017, for that purpose;
And whereas the undersigned appeals officer, authorized by paragraph 146.2(h) of the Canada Labour Code to determine the procedure to be followed in the appeal proceeding, shares the parties’ view that the unrestricted disclosure of such information could result in a real and substantial risk to employee safety and its protection from public disclosure outweighs the public nature of the appeal proceedings;
I Hereby endorse the parties’ “Non-Disclosure Agreement” attached to the present Order and, as such, make it an Order of the Appeals Officer.
Issued on January 9, 2018
Report a problem or mistake on this page
- Date modified: