The construction industry has more fatalities than any other employment sector in the US economy, with falls from heights accounting for over one-third of the fatalities (CPWR - The Center for Construction Research and Training, 2013). Over half of the fatal falls occur from structures that are less than 20 feet high (CPWR - The Center for Construction Research and Training, 2013). In residential construction and the framing industries, falls account for nearly all fatalities (Bureau of Labor Statistics, 2015). The non-residential sector saw a 4% decrease in fatalities in 2012, while residential construction worker fatalities increased 82% (CPWR - The Center for Construction Research and Training, 2013). Despite employing 41% of the construction workforce, 56% of the construction worker fatalities occurred in establishments employing less than 20 employees (CPWR - The Center for Construction Research and Training, 2013).

The risk of sustaining a non-fatal work injury requiring days away from work is 78 per 100 full-time equivalent in the construction industry (Dong, et al., 2014), and falls account for 20% of work days missed (CPWR - The Center for Construction Research and Training, 2013). In the residential sector, inexperienced workers often perform risky work at heights before they have been trained in safe work methods (Kaskutas, et al., 2009, Lipscomb, et al., 2008). Residential carpenter apprentices were twice as likely to fall at work as apprentices working commercial construction (Kaskutas, et al., 2010). Most residential construction contractors employ fewer than ten workers (CPWR - The Center for Construction Research and Training, 2013) and safety problems are more prevalent in smaller construction companies (Cheng, et al., 2010, Kines and Mikkelsen, 2003, Shalini, 2009); where onsite safety professionals are rare, safety programs are lacking (Choi and Carlson, 2014), and worksite training is often inadequate (Hung, et al., 2011).

Construction workers typically learn how to perform production and safety-related tasks from an experienced worker at an active construction site (Rogers, 2007); however the quality and quantity of mentorship can be affected by staffing ratios, productivity expectations, environmental distractions, and the experienced workers’ ability to mentor (Lipscomb, et al., 2008). Hu and colleagues (2011) examined casual factors of construction worker falls in 121 peer-reviewed articles and found strong evidence that both contractor/managerial safety interventions and workers’ training and education influenced fall risks and injuries. Construction foremen and seasoned workers are often expert home builders, but many foremen lack safety communication and teaching skills (Kaskutas, et al., 2013). Evidence suggests that training can improve construction foremen’s safety communication (Hung, et al., 2011, Kines, et al., 2010, Smith, et al., 2008) and can improve construction workers’ fall prevention knowledge and behaviors (Evanoff, et al., 2012). Interventions targeting residential construction are especially timely as the industry is expecting significant growth in new workers (CPWR - The Center for Construction Research and Training, 2013) while learning to comply with more stringent federal fall prevention safety standards (Occupational Health and Safety Administration, 2010).

The goal of this research was to develop, implement, and measure the effects of a construction foremen fall prevention and safety communication intervention targeting priorities identified through needs assessment. We predicted that training foremen to better recognize hazards and train their crewmembers will increase worksite training and effectiveness, increase workers’ safety knowledge and safety behaviors when working at heights, improve perceptions of workplace safety culture, and improve overall worksite safety.

This research occurred in collaboration with contractors, apprenticeship trainers, and members of the Carpenters’ District Council of Greater St. Louis and Vicinity (CDC-StL). All procedures were approved and monitored by Washington University School of Medicine’s Institutional Review Board.

We held 6 training waves with 84 residential construction professionals, including 71 foremen, 5 superintendents, 4 owners, 3 safety directors, and 1 project manager. Table 1 includes demographics for each of the 5 measurement points. Participating foremen were not always able to be surveyed due to work schedule, availability, and because some occasionally worked as crewmembers rather than foremen due to the drop in new home construction that occurred during our study. We were able to survey crewmembers working for at least 60% of the participating foremen at each time-point; including 235 crewmembers pre-intervention, 250 post-intervention, and 93 at extended follow-up. The foremen participants had a high level of experience in the carpentry trade and a mean of over 10 years of tenure with their current employer. The crewmembers had significantly fewer years of work experience and tenure with their employer. The results of the pre-training surveys were similar between the two pre-training visits, suggesting that conditions were stable prior to the intervention, so we combined the two pre-training visits into one pre-intervention category. Six and 12-week post-training results were also very similar, so they were combined into a post-intervention group. Internal consistency of the scales measured with Cronbach’s alpha was moderate to high (Safety climate =.866 foremen and.843 crewmembers, Zohar =.762 foremen and .890 crewmembers, Behavior scale =.686 foremen and .807 crewmembers).

Table 2 demonstrates results from the hierarchical linear models with foremen nested within contractors. Pre-intervention ratings were similar between foremen and crewmembers for all areas surveyed, suggesting that the foremen’s ratings accurately reflected the crewmembers’ perceptions. Increased frequency of fall prevention behaviors was a primary goal of the intervention. The hierarchical models showed large, sustained, and statistically significant improvements in fall safety behaviors, suggesting a large effect of training on both foremen and their crewmembers. These models also demonstrated large statistically significant increases in the number of foremen delivering weekly toolbox talks and in the focus of these talks on methods to perform risky work tasks. Both foremen and crewmember knowledge about fall prevention improved, suggesting that participating foremen diffused the information learned during the training session to their crewmembers through toolbox talks and mentoring interactions. Larger improvements were noted for areas that were emphasized during the foremen’s training (fall prevention behaviors and knowledge and toolbox talk frequency and active delivery methods) and the magnitude of improvements noted were similar between foremen and crewmembers. Model estimates for the safety climate scale demonstrated improved foremen’s perception post-training and at follow-up; crewmembers’ perceptions of safety climate also improved but did not reach significance (p= 0.068). A second measure of safety climate, the Zohar scale, did not show significant changes in the composite score, though one of the items, “foreman approaches workers to discuss safety issues,” showed statistically significant improvement post-intervention by the crew members (p= 0.02), suggesting that crew members recognized a change in their foreman’s focus and communication on safety issues. The reported frequency of daily worker instruction, and beliefs that daily instruction increases safety behaviors did not change among foremen or crewmembers. (Table 2) Changes in observed behaviors were mostly in the direction of being safer; however we did not have enough observations to detect statistically significant changes.

When we examined specific behavior items independent of the hierarchical model, large improvements were noted post-intervention for areas that were emphasized in the training when compare to those that were not emphasized; many of these improvements persisted at follow-up. For example, foremen’s post-training reports of the frequency of working from the top plate of framed walls, a very unsafe behavior, decreased (odds ratio (OR) =6.0 post-training and at extended follow-up), as did the crewmember reports (OR=3.27 post-training and 2.82 extended follow-up). These self-reports were corroborated during worksite visits performed by our trained auditor, with fully safe methods observed 58% of the time prior to the intervention and 79% post-training. The OR for working while standing on a floor joist, another common unsafe behavior, demonstrated statistically significant decreases post-intervention (foremen’ OR=3.22 and crewmembers’ OR=1.65) and at extended follow-up (foremen’ OR=4.30 and crewmembers’ OR=1.55). Another method used commonly to install floor sheathing is to stand on ladders, which may seem to be an innocuous activity, however ladders account for the majority of construction worker fatalities (CPWR - The Center for Construction Research and Training, 2013). When we examined self-reported ladder behaviors, large, statistically significant increases in safe step and extension ladder set up and use were reported post-training and at extended follow-up by foremen and their crewmembers. Regarding use of personal fall arrest systems, statistically significant increases in equipment use were reported by foremen (OR=2.31 post-intervention and 2.01 extended follow-up) and crewmembers (OR=2.66 and 3.67 respectively). Worksite audits corroborate that personal fall arrest systems were being used more often, but they were often set-up incorrectly, suggesting that further training is needed. Lastly, a greater proportion of crewmembers reported that they were familiar with their company’s fall prevention plan after participant training and at extended follow-up, which also suggests that crewmember training improved.

When examining differences between levels of management and contractors participating in the intervention, we found that the foremen and upper management groups reported similar safety behaviors before and after the intervention. The safety climate was perceived to be 12-points better by upper managers when compared to foremen, and foremen perceived a better safety climate than their crewmembers, demonstrating the importance of measuring safety climate at different levels within an organization. The contractor that demonstrated the largest improvements in safety behaviors, safety climate, and toolbox talks sent all levels of management to the training and participated in the intervention just prior to the date that more stringent federal safety standards were taking effect (Kaskutas, et al., 2014).

Although we were unable to directly measure the impact of the intervention on falls from height among participating foremen and their crewmembers, our previous study among apprentice carpenters showed that a 1-point increase in the fall safety behavior scale score was associated with a 1.4% decrease in the incidence of self-reported falls in the past year (Kaskutas, et al., 2010). Extrapolating to the current project, we would expect that the observed post-intervention increase of 11.9 points on the fall safety behavior scale would be associated with a 16.6% decrease in the one year cumulative incidence of self-reported falls among apprentice carpenters following the intervention.

The intervention described in this research resulted in sustained improvements in fall prevention behaviors and safety communication in residential construction, a hard-to-reach sector of workers with excessively high morbidity and mortality due to falls. Combining fall prevention and safety communication training equipped the foremen participants with the requisite knowledge, tools, and skills needed to lead their work crews toward safer methods of performing work on elevated surfaces. Importantly, both the foremen and the workers on their crews reported improvements in fall prevention behaviors and safety communication after the intervention. These improvements seen in crewmembers’ behaviors suggest that the foremen who received training assimilated and disseminated portions of the training to their crewmembers. This view is supported by the reported increases in the frequency of toolbox talks, and their increased focus on relevant safety issues.

This research suggests that providing foremen with communication training at the time of safety-specific training will improve their abilities to influence the safety behaviors of their employees. Most construction foremen possess an excellent skill set in their building trade; however many are being placed in safety leadership roles that they may not prepared to assume. As the economy rebounds, 1.3 million workers are projected to join the construction workforce by 2020 (CPWR - The Center for Construction Research and Training, 2013). Residential contractors and foremen will need to train these new workers to safely perform the wide array of work tasks that are regularly performed at the worksite. Training foremen to juggle their production, safety, and mentorship roles can improve worker safety and ultimately prevent falls from heights at residential worksites.

Although this intervention did not target safety climate or safety culture, post-intervention ratings suggest that safety climate was improved, with greater changes observed among the foremen. Our intervention was in part aimed at improving safety communication between foremen and their crews, which likely affected perceptions of safety climate at all levels. Hahn and Murphy (2008) found that safety climate correlated strongest with effective communication and more frequent qualitative feedback between managers and employees. Differing perceptions of safety climate between front-line workers, their supervisors, and management were evident in this research, suggesting that future studies addressing safety climate should measure at multiple levels of the organization. Safety climate and culture are emerging priorities for construction safety and health (National Occupational Research Agenda, 2008) that encompasses “deeply held but often unspoken safety-related beliefs, attitudes and values that interact with an organization’s systems, practices, people, and leadership to establish norms about how things are done in the organization” (CPWR - The Center for Construction Research and Training, 2014). Our intervention was not intended to be delivered to an entire organization, but several levels of management from a few contracting companies did participate in the training.

We made multiple cross-sectional measures of each participating foreman’s crew rather than following individual crewmembers longitudinally as workers naturally flow on and off of residential work crews. Although this is a potential limitation of this research, it provided an accurate reflection of the transient nature of residential construction crews. Multiple cross-sectional measures instead of longitudinal follow-up of individual crew members is most likely to have resulted in an underestimation of the effects of the training on the crewmembers, as they would have had less exposure to the trained foreman and fewer opportunities for transfer of safety practices. The pre-post design was also a limitation of this study, as we lacked a concurrent control group. We did observe larger improvements in areas that were emphasized in the training, suggesting specificity of effect and supporting the conclusion that the observed effects are not due at least in part to the intervention. Due to the economic downturn, there were fewer active worksites than anticipated, which left us inadequately powered to detect changes in observed worksite behaviors, as the high risk behaviors that were the emphasis of this intervention occur only intermittently. Changes in the federal fall prevention safety standards for residential construction may also have affected our results; however we saw similar improvements in all six training waves over the four-year intervention period, including companies that participated prior to the changed federal standards, during the changes, and afterward.

Our sample was representative of the predominantly unionized residential construction workforce in the St. Louis metropolitan area, but not of the residential workforce across the country, which is mostly non-union and may utilize temporary day workers. Because the intervention was delivered to contractor-based groups of foremen, we believe that this intervention could readily be delivered to groups of foremen within non-union contractors, provided the contractor was willing to support the program. It is plausible that a foreman safety and communication intervention would show even larger improvements in safety behaviors if delivered to non-union foremen and crews, who typically receive much less safety training than our apprenticeship trained St. Louis workforce. Detailed intervention objectives and learning activities outlined in a training manual ensured that the intervention was consistently delivered, and increase portability of this intervention to other settings. Future research to test this intervention with non-union residential workers in other parts of the country is suggested, as well as conversion to a web-based format available in other languages to enhance wider dissemination than could be achieved with the classroom format used in our intervention.

This research adds to the growing literature demonstrating that needs-driven training can improve construction worker safety and worksite safety communication. Providing participants with baseline performance metrics and actively engaging learners in small group problem-solving are proven educational methods that are applicable to the construction sector. Most residential construction companies are small and their workforce is often transient and widely dispersed, limiting access to effective training and safety supervision. By using innovative delivery methods, we can extend the reach of safety and health training to this sector of the construction workforce with the greatest exposure to unprotected work on elevated surfaces.