During the 2015–2016 academic year, a total of 242 teachers from approximately 98 schools in 3 large school districts in urban areas in the Southeast, Central, and Northeastern United States completed the NIOSH questionnaire before participating in NIOSH-led trainings on the Talking Safety curriculum. Decisions as to which classes/tracks and at what grade level to implement the curriculum were made by administrators at the district level, prior to engagement with NIOSH. To help teachers become familiarized with the Talking Safety curriculum, they were recruited by department administrators to attend NIOSH training sessions conducted during regularly scheduled professional development days. Participants in the Southeastern school district who were selected to deliver the curriculum taught an academic subject (science) in traditional, comprehensive schools; in the Northeast, participants taught both technical/occupational and academic subjects as well as other career and technical education subjects (such as family and consumer science and health), within a large, state-wide system of career and technical schools; and participants in the central US district taught non-occupational career and technical education (business and technology) courses in comprehensive schools.

For the full sample (N = 242) used to confirm whether the questionnaire developed for the study reflects the modified, theory of planned behavior model, approximately 28% (N = 67) of teachers reported their main subject to be occupational career and technical education (such as construction, manufacturing, or culinary arts), 40% (N = 98) taught an academic subject (such as science) in a comprehensive school, and 32% (N = 77) taught non-occupational career and technical education in comprehensive schools and academic subjects (such as English or math) in a career and technical school.

Of the 242 participants, 54 individuals were excluded from further analysis because they were non-teachers, such as guidance counselors (N = 39) or teachers who did not provide any demographic information (N = 15) necessary for the statistical comparisons. Therefore, a subsample of 188 teachers who taught subjects where the curriculum was being considered for implementation were included in the ANOVAs (occupational career and technical education teachers in technical schools, N = 67; academic teachers in comprehensive schools, N = 98; and non-occupational career and technical education [business and technology teachers] in comprehensive schools, N = 23). Descriptive statistics for the subsample are reported in Table 2. The largest proportion of teachers in the subsample (26%) reported length of time teaching at more than 20 years. Approximately one-fourth of teachers in the subsample reported having experienced a work-related injury severe enough to require time off work, and roughly equal numbers of women and men were present.

Before the questionnaire was administered, content validity was established by having 3 teachers in one of the study districts and 2 school administrators review and provide feedback on all items with regards to clarity, readability, and item content. Minor revisions were suggested, mainly to clarify instructions.

Participants were administered the paper questionnaire before attending a NIOSH training on the Talking Safety curriculum, held during a regularly scheduled, professional development day in the district. Participants took about 15 to 20 minutes to complete the questionnaire. Data were deidentified and collected for analysis.

Prior to analysis, data were examined for missing values. Item level missing data ranged from 5.1% to 10.6%. To handle missing data, single imputation was conducted in Mplus (version 8)⁴⁸ using the Markov Chain Monte Carlo algorithm⁴⁹ to account for ordinal indicators and included several auxiliary variables (ie, teacher sex, years teaching, main subject taught, and injured at work).

A confirmatory factor analysis was conducted on the full teacher sample (N = 242) to assess the fit of the 3 correlated factors that measured the latent constructs of attitude, self-efficacy, and behavioral intention related to teaching workplace safety and health. A separate analysis was conducted to provide evidence that the knowledge measure reflected a single factor. The models were estimated in Mplus with weighted least squares with mean and variance correction (WLSMV) estimation. Model fit was assessed with the χ² _WLSMV statistic and its associated p value (good fit was indicated by an insignificant result, p > .05);⁵⁰ root mean square error of approximation (RMSEA), comparative fit index, (CFI), Tucker-Lewis Index (TLI), and the weighted root mean residual (WRMR). The following fit criteria and cutoff values were used: for good fit,^51,52 RMSEA ≤ 0.06, CFI ≥ 0.95, TLI ≥ 0.95, and WRMR values close to 1. Because of the way that the chi-square statistic is calculated, it is unlikely to be able to obtain a non-significant test result, but it is commonly reported despite serious limitations.⁵² Finally, internal consistency of reliability was estimated with coefficient omega (ω).⁵³

Consistent with the theory of planned behavior, a 3-factor model consisting of attitude, self-efficacy, and intention factors adequately fit the data: χ²_WLSMV(249) = 565.638, p < .001, RMSEA = 0.072, RMSEA 90% CI [0.064, 0.080], CFI = 0.995, TLI = 0.994, WRMR = 1.082. All standardized factor loadings were statistically significant (p < .05) and ranged from 0.85 to 0.98. An examination of structure coefficients indicated that no items were mis-specified (Table 3). Residual correlations were all less than |0.10|, which suggested acceptable local fit.⁵⁷ Factor correlations were 0.40 between attitude and intention, 0.43 between attitude and self-efficacy, and 0.62 between intention and self-efficacy. Sample reliability (ω) were, for attitude = 0.97, for self-efficacy = 0.98, and intention = 0.99 (Table 3).

Results for the 13-item knowledge measure indicated a one-factor model adequately fit the data: χ²_WLSMV (65) = 66.522, p = .42, RMSEA = 0.010, RMSEA 90% CI [0.000, 0.040], CFI = 0.993, TLI = 0.991, WRMR = 0.673. All standardized factor loadings were statistically significant (p < .01) and ranged from .50 to .91. Residual correlations were all less than |0.10|. Sample reliability (ω) was 0.90.

Estimated marginal means, standard errors, and confidence intervals for the theory of planned behavior and knowledge measures related to teaching workplace safety and health are reported in Table 4. Although no statistically significant main or interaction effects were found for attitude, all participants had high scores on the measure, with occupational career and technical education teachers demonstrating a more positive attitude toward teaching this topic (M = 4.76, SE = 0.08) when compared to the other respondents (M = 4.55, SE = 0.09).

A statistically significant main effect was observed for sex on self-efficacy, F(1, 177) = 4.99, p = .03, η²_Partial = 0.03 (small effect size), d = 0.34 (small effect size), indicating that on average male teachers (M = 4.24, SE = 0.13, N = 85) had higher self-efficacy scores to teach workplace safety and health than female teachers (M = 3.81, SE = 0.14, N = 100). No other main or interaction effects were observed for self-efficacy scores.

Moreover, although no statistically significant main or interaction effects were found for the behavioral intention measure, the injured at work variable approached significance on the intention outcome (p = .07), with those teachers having experienced a previous work-related injury demonstrating higher mean scores on average (M = 4.30, SE = 0.19, N = 44) when compared to the other sample teachers who had not been injured (M = 3.90, SE = 0.101, N = 144). It should also be noted that average scores (Table 4) were lower on this outcome than might be expected—especially for the non-occupational career and technical education and academic teachers (M = 3.90, SE = 0.16)—given district-wide buy-in for the program.

For the knowledge measure, a statistically significant interaction effect was found between the variables main subject taught and sex, F(1, 177) = 4.37, p = .04, η²_Partial = 0.02 (small effect size), d = 0.31 (small effect size). Simple pairwise comparison tests were conducted to examine the effects of main subject (occupational versus non-occupational career and technical education and academic) and sex on knowledge scores. Statistically significantly higher mean knowledge scores were found for male occupational career and technical education teachers (M = 12.05, SE = 0.256, N = 49) versus male non-occupational career and technical education and academic teachers (M = 10.33, SE = 0.48, N = 36), t(177) = 3.18, p = .002, d = 0.48 (medium effect size). Finally, teachers indicating a previous workplace injury had lower knowledge scores (M = 10.87, SE = 0.32, N = 44) than those who reported having no previous injury at work requiring time off work (M = 11.67, SE = 0.183, N = 141), F(1, 177) = 4.40, p = .04, η²_Partial = 0.02 (small effect size).

According the US Centers for Disease Control and Prevention, 24% to 98% (median = 82%) of large, urban school districts include injury prevention, and safety as part health education instruction.¹⁹ Risk reduction of work-related injuries among adolescents is a sub-area of injury prevention and safety,² but the extent to which students are being taught this topic, if at all, is unknown. The majority of teens work before completing high school,⁵ and many are injured due to a lack of inadequate preparation for the hazards and risks they encounter on the job.^6–14 These injuries can be life-altering, and are all the more tragic because most can be predicted and prevented. Increased recognition of the importance of and need for work safety education may result in an increased delivery of these programs in schools to teens entering the workforce.²¹ In France, Boini et al.¹⁸ found that young workers who had received workplace safety and health education at school reported 2 times fewer injuries on the job than young workers who had not received this preparation (incidence rate ratio [IRR] 0.51, 0.00–0.98). For widespread adoption of this important topic as part of health education curriculum, decision makers within the school system will need to gain awareness of the public health impact of young worker injuries and the need to prepare young people for safe and healthy work. A champion,⁶³ whether a school administrator, teacher, or parent, can be instrumental in raising awareness about the importance of occupational safety and health may facilitate the adoption of youth education in this area. School districts must also evaluate the appropriate fit, in terms of curricular area and grade level, for occupational safety and health education. For example, the Youth@Work—Talking Safety curriculum,³⁸ discussed previously, was originally developed for use in high schools. However, Talking Safety is currently being implemented in a number of settings, including in eighth grade Science (Human Growth and Development) classes, in one of the largest US public school districts.⁶⁴ Previous research suggests the need to enhance the integration of health education topics into core curriculum classrooms⁶⁵ and have shown that those health education programs integrated into normal school activities are implemented more successfully.⁶⁶ Moreover, earlier introduction of these essential life skills creates a foundation of workplace safety and health knowledge and skills before youth enter the labor force and before they are ever exposed to their first job hazard in a formal work environment.

Evidence from the public health literature suggests teachers play a critical role in ensuring the effective transfer of information on health topics,^19,20 including workplace safety and health,²¹ and that professional development/training for teachers may enhance self-efficacy and generally facilitates the successful and sustainable implementation of evidence-based, health education programs.^{23,32,61,67,68} Buy-in for new curricula should be sought and teachers provided training on new health education programs. The school districts included in this study arranged for training on the Talking Safety curriculum to be included as part of a regularly scheduled professional development day. Moreover, to achieve the long-term sustainability of new programs, it is necessary to build internal capacity to support its on-going implementation. One way to achieve this is through a training-of-the-trainer (TOT) model, which has been used to varying extents in the districts included in this study.

Institutional support, including principal support,^61,66,69 is also necessary to ensure the successful, sustainable implementation of new health education programs. Moreover, adoption of new health education programs requires there be a good fit between the innovation and local needs,⁶³ and that the innovation is adaptable to new practices, or easily integrated into current practices. Research from Parcel et al.⁷⁰ indicates that school districts able to adapt to new practices, or able to see how a new program could be integrated into current practices, were more likely to adopt an innovation. The free and widely-used NIOSH Talking Safety curriculum³⁸ may be immediately adopted by US school districts as part of existing school health education programs or easily adaptable to fit into other classes where health education topics are presented. The curriculum, customized for all US states, Puerto Rico, and the U.S. Virgin Islands, is available for free download, and is also available in Spanish. The curriculum may be used by middle school and high schools to help to prepare all young people for a lifetime of safe and healthy work.

The NIOSH Institutional Review Board (IRB)/Human Research Protection Program (HRPP) issued a research exempt determination for this project (16-EID-02XM and 14-EID-07XM) under 45 CFR 46.101(b)(1). This research was conducted in accordance with the ethical standards of the NIOSH IRB/NIOSH HRPP and with the 1975 Helsinki declaration as revised in 2000.