*Methodological Considerations*

Research in construction has challenges with loss to follow up due to the dynamic nature of construction with workers coming and going on worksites as the construction job requires specific trades during the timeline of the study [18,36]. The issue of poor follow-up rates can lead to bias; however, usually towards the null [62]. This is predominately due to the itinerant nature of construction workers [18]. This resulted in our analyses at FU2 being underpowered for some of our outcomes (such as pain and injury) where the effect size was similar to FU1. Similar findings have been found in related interventions [63].

Another challenge was the success of integrating the injury prevention and health promotion activities in this environment. Integration was achieved by linking the two programs by name and key messaging in the planning and implementation phases. In addition, messaging around Health Week included training on both injury prevention and health promotion giving workers the tools to improve their working conditions, as well as giving them control for their health. The ergonomics program prior to Health Week did have health messaging but without any specific health promotion activities.

One limitation was that our intervention depended on the participation of the general contractor safety managers, whose involvement and dedication to the study varied across sites and between general contractors. This aspect of the intervention was by design, as we considered it important for integration and sustainability into current company processes, that the ergonomics inspections were performed by the safety managers. Giving the safety managers latitude to decide how invested they were in the program allowed us to assess the feasibility of the intervention being adopted without the aid of study staff. This would ensure that our observations were realistic and representative of barriers and facilitators to the intervention's delivery by non-study staff.

Further, this study involved worksites in commercial construction only. Thus, the results may not be generalizable to other types of construction (i.e., residential or industrial). However, commercial construction accounts for a large portion of U.S. construction activities, and represents an important area for injury prevention research. Similarly, the construction workforce in the Boston metropolitan area may not be representative of commercial construction workers in other parts of the country or world, where work practices, demographics, and union membership differ.

Despite these limitations, our study had several strengths, most notably the study design and the wide variety of general contractors and sites recruited into the study. The cluster randomized control trial design is a novel approach in commercial construction. Typically, approaches to improve the health and safety of construction workers have often focused on the individual worker, targeting workers when they are enrolled in apprentice programs [64,65], targeting workers through social media campaigns via posters at worksites and/or brochures sent to union members [66,67], and engineering controls for specific tasks [68]. However, best practice involves system-level approaches that comprehensively address workplace systems relevant to the control of hazards and worker safety, health, and well-being [20]. This study was fortunate to be able to recruit five major general contractors operating in the Boston metropolitan area and gain access to ten different construction sites for the purpose of evaluating the ARM intervention. Furthermore, delivering the intervention through mid-level managers (through a combination of the general contractor safety managers and subcontractor foremen) was a strength of the study. This focused intervention efforts on those who were in the best positions to make changes to the conditions of work.

The ergonomics inspection and communication protocol provided a method to identify broad areas for improving ergonomics in the dynamic construction work environment. It is important to understand the challenges and successes of intervention delivery in order to inform and improve future worksite-based interventions. It appears that the largest barriers to the success of the intervention were the inability of subcontractors to make changes to their worksite and the variability in the involvement and dedication to the study across different worksites and general contractors. These are real-world, as well as research study challenges. Subcontractors did not have the systems in place, or the available tools, to assist in changing their working conditions. Competing safety and production priorities also influenced the level of management commitment to the study. Additionally, construction safety research may have broader implications for an increasing number of industries that are becoming as dynamic and variable as construction, as more services once housed in a single facility are outsourced to multiple employers [17].
