Health risk factors as predictors of workers' compensation claim occurrence and cost
====================================================================================

* Natalie V Schwatka
* Adam Atherly
* Miranda J Dally
* Hai Fang
* Claire vS Brockbank
* Liliana Tenney
* Ron Z Goetzel
* Kimberly Jinnett
* Roxana Witter
* Stephen Reynolds
* James McMillen
* Lee S Newman

## Abstract

**Objective** The objective of this study was to examine the predictive relationships between employee health risk factors (HRFs) and workers' compensation (WC) claim occurrence and costs.

**Methods** Logistic regression and generalised linear models were used to estimate the predictive association between HRFs and claim occurrence and cost among a cohort of 16 926 employees from 314 large, medium and small businesses across multiple industries. First, unadjusted (HRFs only) models were estimated, and second, adjusted (HRFs plus demographic and work organisation variables) were estimated.

**Results** Unadjusted models demonstrated that several HRFs were predictive of WC claim occurrence and cost. After adjusting for demographic and work organisation differences between employees, many of the relationships previously established did not achieve statistical significance. Stress was the only HRF to display a consistent relationship with claim occurrence, though the type of stress mattered. Stress at work was marginally predictive of a higher odds of incurring a WC claim (p<0.10). Stress at home and stress over finances were predictive of higher and lower costs of claims, respectively (p<0.05).

**Conclusions** The unadjusted model results indicate that HRFs are predictive of future WC claims. However, the disparate findings between unadjusted and adjusted models indicate that future research is needed to examine the multilevel relationship between employee demographics, organisational factors, HRFs and WC claims.

*   Occupational injury
*   Health risk assessment
*   Small business
*   Worksite wellness
*   Total Worker Health

### What this paper adds

*   Little is known about how health risk factors (HRFs) are related to the frequency and cost of the occupational injuries, illnesses and fatalities that result in workers' compensation (WC) claims. Previously, researchers focused on large, single-company sample populations to examine this relationship.

*   Using a diverse sample of workers and worksites, the researchers demonstrated that some HRFs are more common among those workers who subsequently experience a work-related injury that results in a WC claim. However, the analyses suggest that HRFs, demographic and work organisation factors may interact to predict the occurrence and cost of WC claims.

*   These findings reinforce the value of introducing a Total Worker Health approach whereby employers optimise the work environment to prevent and mitigate work-related injuries and poor health, especially with the goal of mitigating sources of stress, while also serving as a platform for empowering employees to adopt and practice behaviours for better personal health *and* safety.

## Introduction

Parallel trends can be observed between the rise in direct and indirect costs of healthcare and work-related injury and illness. In 2013, the US Bureau of Labor Statistics reported over 3 000 000 non-fatal workplace injuries1 and over 4000 fatal work injuries.2 Of the non-fatal injuries, approximately one-third resulted in lost work time with a median of 8 days away from work per case.3 While the frequency of occupational injuries and illnesses has declined over the past two decades, the cost per workers' compensation (WC) claim has risen.4 Occupational injuries cost US employers almost $250 billion dollars annually.5

At the same time, workers' health and the risk factors leading to poor health have also been shown to influence employer costs and employee productivity. Ward and Schiller6 estimated that a quarter of US adults have at least one chronic health condition, and another quarter has two or more. Many adults also experience depression, anxiety and other types of mental distress.7 Chronic conditions result in significant out-of-pocket healthcare costs.8 Personal health risk factors (HRFs), as assessed by an employer-sponsored health promotion programme using health risk assessments (HRAs), have previously been associated with higher healthcare and lost productivity costs. Goetzel *et al*9 ,10 found that depression, high stress, high glucose levels, high blood pressure, obesity, high cholesterol and lack of exercise were associated with high healthcare costs. Goetzel *et al*11 also found that factors related to high biometric laboratory values (eg, blood pressure), alcohol/tobacco use and emotional problems were associated with higher presenteeism (ie, poor job performance). Frey *et al*12 found that some psychosocial factors (eg, poor sleep) were associated with higher presenteeism.

On a broader scale, in 2011, total US healthcare costs exceeded $2.7 trillion with ∼84% of the costs attributable to personal healthcare and 10% attributable to prescription drug costs.13 Identifying modifiable HRFs via early detection and facilitating necessary treatment and/or behaviour change can reduce direct and indirect costs associated with chronic conditions and other preventable illnesses, and improve worker productivity.14

Organisations adopting a Total Worker Health (TWH) strategy seek to understand the relationships between health promotion and health protection and how a more holistic approach can positively influence employee health, safety and well-being.15 One way in which to understand the interplay between worker health and safety is to examine how HRFs are related to occupational injury. Although there is emerging evidence of an association between employee health and occupational injury,16 little is known about how HRFs are related to the frequency and severity of injuries, illnesses and fatalities that result in WC claims.

Previously, researchers focused on the relationship between employees' comorbid HRFs and WC based on samples from single, large companies (ie, >1000 employees).17–20 Some studies focused on evaluating the relationship between overall health risk level and WC claims by summing the number of HRFs or examining individual HRFs.18–20 Their findings have been inconsistent regarding the relationship between HRFs and WC claims,17–20 and only one study17 examined whether HRFs predict subsequent injury and cost. This study is a continuation of the Pinnacol Assurance Health Risk Management (HRM) research programme.21 The purpose of the HRM programme is to understand the impact of a worksite wellness programme offered by a WC insurer on the health, safety and productivity of covered employees. As part of this research, the present study examines the prospective relationship between individual HRFs, as defined by self-reported HRA responses, and WC occurrence and cost among a diverse sample of employees from multiple employers (see table 1). We hypothesised that HRFs related to lifestyle, psychosocial conditions and health conditions are predictive of (1) a higher odds of filing a WC claim and (2) higher medical and total costs of WC claims.

View this table:
[Table 1](http://oem.bmj.com/content/74/1/14/T1)

Table 1 
Demographic and work organisation characteristics of employees with and without a WC claim

## Methods

### Sample

The present study draws on a cohort of Colorado employees (N=16 926) from 314 companies who participated in a prospective longitudinal study from 1 May 2010 to 31 December 2014. The HRM programme included annual HRAs, feedback reports, action plans for improving wellness and reducing health risks, unlimited telephonic coaching and access to educational resources for employees. Employers with more than 50 participating employees received annual risk and recommendation reports with *aggregated* employee HRF data to highlight employee needs, ongoing feedback on participation and progress, educational content to distribute to employees and advice on programme enhancements. If employers had fewer than 50 employees, they were given an aggregated report based on their industry average to benchmark across HRFs. Pinnacol Assurance supported an external evaluation of the HRM programme by partnering with researchers from several academic institutions.21 ,22 For the purposes of this study, the unit of analysis was an employee who completed a baseline HRA questionnaire during the study period.

The data analysed in this study were subjected to a robust data linkage process to insure worker privacy. The HRA data were transferred from the wellness vendor, and the WC claims data were transferred from Pinnacol Assurance to the Integrated Benefits Institute for de-identification and then transferred to the Center for Health, Work and Environment at the Colorado School of Public Health for analysis. The Colorado Multiple Institutional Review Board determined the study to be exempt from human subjects research.

### Measures

An online, self-administered, English and Spanish HRA was offered to employees. The Wellsource HRA23 used was provided by Trotter Wellness and certified by the National Committee for Quality Assurance. The HRA included validated questions in the following categories: biographical information, health history, medical care, physical activity, nutrition, substance use, mental/social health, injury prevention practices in one's personal life and readiness to change. The HRA was supplemented with 58 additional selected questions from the WHO's Health and Work Performance Questionnaire (HPQ)24 and a shortened version of the validated HPQ, the HPQ Select,25 including demographic, health and productivity information. All responses were self-reported. For this study, the HRFs of interest were related to lifestyle, psychosocial and health condition factors. The HRFs were chosen based on their association with healthcare costs in previous literature.9 ,10 The variables, their descriptions and operational definitions of high risk can be found in online supplementary table S1. HRFs were considered present if employees indicated that they had the condition, and absent if employees indicated that they did not have the condition or if they left the question blank.

### Supplementary table

Data dictionary

[[oemed-2015-103334supp_table.pdf]](pending:yes)

WC claims were included if they were initially filed within 1 year *from* the date when the employee completed a baseline HRA and occurred at the same company in which the HRA was taken. On average, employees filed a claim 160 days (SD=105) after their baseline HRA. We excluded claims (3.0%) that had not closed within 18 months of the date of injury. For the purpose of this study, we only included compensable non-zero-cost claims (66%). Employees with a zero-cost WC claim were categorised the same way as employees without a claim because zero-cost claims do not represent compensable injuries, illnesses or fatalities. The WC variables of interest were (1) a dichotomous variable representing whether or not there was at least one compensable claim filed, (2) medical cost and (3) the total cost of the claim(s) that were filed. Total cost included all medical, indemnity and expenses (eg, legal fees) associated with the claim. Medical costs included all direct medical care costs (eg, clinical care, hospitalisation and prescriptions). All cost data were inflation-adjusted to 2013 dollars using the Consumer Price Index. Pinnacol Assurance provided all claims data.

Finally, employee demographic and work organisation variables were included as control variables in all adjusted analyses. Employee demographic variables included age, gender and education level. Work organisation variables included employment type (full time vs part time), pay scheme (hourly vs salary), occupation, income, company size (number of employees) and industry (Standard Industry Codes (SIC)). Results for all variables were derived from the HRA, except for company size and industry, which were provided by Pinnacol Assurance.

### Analysis

We generated descriptive statistics for the overall employee sample as well as for the bivariate relationship between HRFs and (1) employees who did and (2) employees who did not have a WC claim. A χ2 test was used to determine if the proportion of employees who had a WC claim was independent of demographic and HRF variables.

For all multivariate analyses, unadjusted and adjusted models predicting WC claim occurrence and cost were estimated. Unadjusted models only included HRFs, whereas the adjusted models included HRFs and demographic and work organisation control variables. We estimated the odds of filing at least one WC claim 1 year after the employees' baseline HRA using logistic regression. A sensitivity analysis was performed to determine the impact of defining WC claim occurrence as having $0 and >$0 claims instead of only >$0 claims. Finally, we used a generalised linear model (GLM) analysis to estimate the relationship between HRFs and average WC costs among employees who had a claim. The generalised gamma distribution and log link function were used in the analysis. The generalised gamma regression allows for nested comparisons of the more frequently used distribution models and can provide more efficient estimators since it is less restrictive than the nested distributions.26 Additionally, interpretation of the coefficients is not marred by re-transformation, as compared to ordinary least squares with a log-transformed dependent variable. These methods have been used previously when evaluating healthcare costs.27 Discrete differences in average marginal effects and their 95% CIs were estimated in order to facilitate interpretation of the significant and marginally significant coefficients in the GLM models. The marginal effects represent the average WC claim cost difference for employees with the HRF, as compared to employees without the HRF. Effects were considered statistically significant if the p value was <0.05. Marginally significant effects with a p value of <0.10 were reported where the clinical researchers determined that the findings had practical significance. Data management and logistic modelling were performed using Stata V.12 (StataCorp, College Station, Texas, USA). The generalised linear modelling of cost data was performed using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA).

## Results

A total of 16 926 employees completed an HRA. For a description of the study population, see table 1. Three per cent (n=533) of the employees who completed an HRA had at least 1 WC claim within 1 year of completing the HRA. Of these, 9% filed 2 or more claims in the same year. The most common claims were for contusion (26%), strain (25%), sprain (17%) and laceration (8%) injuries. Among employees who had a WC claim, the average total cost was $3971 (SD=$13 139) (median=$796, range=$19–$173 878). The average medical cost was $2413 (SD $5314) (median=$776, range=$0–$45 122).

The most common lifestyle HRFs were poor sleep (45%), the use of poor lifting practices (41%) and inadequate exercise (41%) (see table 2). The most common psychosocial HRFs were stress over finances (63%), stress at work (31%) and depression (21%). The most common health condition HRFs were abnormal body mass index (BMI) (59%), no dental examination in the past year (29%), severe headaches (25%) and no physical examination in the past 1–2 years (23%).

View this table:
[Table 2](http://oem.bmj.com/content/74/1/14/T2)

Table 2 
HRFs among employees with and without WC claims

### Unadjusted relationship between HRF and WC claims and costs

#### Bivariate analyses

A number of HRFs were significantly related to incurring a WC claim (p<0.05) (see table 2). For example, employees were more likely to file a claim within 1 year if they were smokers versus non-smokers (18% vs 13%), sick with diabetes (6% vs 4%), with abnormal BMI (63% vs 58%) and not wearing seat belts (26% vs 20%).

Three of the HRFs were statistically significantly predictive of higher mean total costs. Employees *with* diabetes incurred, on average, $12 074 (SD=$34 346) in total WC claim costs, whereas employees *without* diabetes incurred, on average, $3488 (SD=$10 519) in total WC claim costs (t=−3.51(531), p=0.00). Employees *with* arthritis incurred, on average, $6427 (SD=$21 302) in total WC claim costs, whereas employees *without* arthritis incurred, on average, $3458 (SD=$10 064) in total WC claim costs (t=−1.97(531), p=0.05). Finally, employees *with* chronic fatigue incurred, on average, $5758 (SD=$19 529) in total WC claim costs, whereas employees *without* chronic fatigue incurred, on average, $3479 (SD=$10 718) in total WC claim costs (t=−1.65(531), p=0.09).

#### Multivariate analyses

As shown in table 3, 4 HRFs were found to be significantly (p<0.05) predictive of the likelihood of having a WC claim in the unadjusted models. These included having digestive disorders (OR=1.28, 95% CI 1.00 to 1.63), poor seat belt use (OR=1.27, 95% CI 1.03 to 1.56) and exposure to secondhand smoke (OR=1.34, 95% CI 1.05 to 1.70). However, drinking and driving (OR=0.76, 95% CI 0.59 to 0.98) was predictive of a lower odds of filing a WC claim. A sensitivity analysis using $0 and >$0 claims instead of only >$0 claims for the WC claim occurrence variable revealed only two significant differences in the results of this logistic model. Safe lifting (OR=0.83, 95% CI 0.72 to 0.97) was a significant predictor, and exposure to secondhand smoke (OR=1.13, 95% CI 0.91 to 1.40) was a non-significant predictor.

View this table:
[Table 3](http://oem.bmj.com/content/74/1/14/T3)

Table 3 
Unadjusted and adjusted multivariate logistic regression models of the ability of HRFs to predict the occurrence of a WC claim (N=16 926)

Ten HRFs were significantly predictive of subsequent WC medical claim costs (see table 4). For example, smoking (β=$179, 95% CI −$13 to $372) was significantly predictive of higher medical claim costs. On the other hand, stress over finances (β=$216, 95% CI −$438 to $7) was significantly predictive of lower medical claim costs.

View this table:
[Table 4](http://oem.bmj.com/content/74/1/14/T4)

Table 4 
Unadjusted and adjusted multivariate GLM regression models of ability of HRFs to predict WC claim *medical cost*

Eleven HRFs were significantly predictive of total claim costs (see table 5). For example, pre-existing back pain (β=$150, 95% CI −$11 to $289) and heart disease (β=$634, 95% CI −$5 to $1274) were significantly predictive of higher total claim costs.

View this table:
[Table 5](http://oem.bmj.com/content/74/1/14/T5)

Table 5 
Unadjusted and adjusted multivariate GLM regression models of the ability of HRFs to predict WC claim *total cost*

### Adjusted relationship between HRF and WC claims and costs

#### Multivariate analyses

Three of the four HRFs shown in table 3 to be significantly predictive of the likelihood of a claim were no longer significant after adjusting for demographic and work organisation factors: drinking and driving (OR=0.84, 95% CI 0.65 to 1.10), poor seat belt use (OR=1.14, 95% CI 0.91 to 1.42) and exposure to secondhand smoke (OR=1.04, 95% CI 0.80 to 1.35). Two of the HRFs became marginally significant after adjusting: excessive alcohol use (OR=0.72, 95% CI 0.49 to 1.06) and stress at work (OR=1.22, 95% CI 0.98 to 1.52). A sensitivity analysis using $0 and >$0 claims instead of only >$0 claims for the WC claim occurrence variable revealed only one significant difference in the results of this logistic model. Seat belt use (OR=1.21, 95% CI 1.01 to 1.46) was a significant predictor.

As shown in table 4, 8 of the 10 HRFs that were significantly predictive of medical claim costs were no longer significant after adjustment: poor sleep, smoking, high blood pressure, high cholesterol, heart disease, lung disease, osteoporosis and stroke (marginal effects for insignificant variables not shown). One HRF became significant after adjustment: stress at home (β=−$206, 95% CI −$109 to $521). Two HRFs became marginally significant after adjustment: severe headaches (β=$132, 95% CI −$67 to $331) and irritable bowel disorder (β=−$160, 95% CI −$421 to $101).

As shown in table 5, 9 of the 11 HRFs that were significantly predictive of total claim costs, no longer significant after adjustment: poor sleep, smoking, back pain, high blood pressure, high cholesterol, heart disease, irritable bowel disorder, lung disease and stroke (marginal effects for insignificant variables not shown). One HRF became significant after adjustment: stress at home (β=$156, 95% CI −$37 to $349). One HRF became marginally significant after adjustment: overall health (β=−$56, 95% CI −$117 to $4).

## Discussion

Using a diverse sample of 16 926 employees from 314 companies across a variety of industries, we found a number of HRFs that were predictive of the occurrence and/or increased cost of WC claims. In the unadjusted models, 4 HRFs were predictive of future WC claim occurrence, and 10 and 11 HRFs were predictive of future WC total and medical claim costs, respectively. However, after adjusting for demographic and work organisation factors, most HRFs were no longer predictive of future WC claim occurrence or cost. In the adjusted models, employee-reported stress was predictive of future WC claim occurrence and cost. Notably, the type of stress mattered. Stress at work predicted higher WC claim occurrence, whereas stress at home predicted higher WC claim costs and stress over finances predicted lower WC claim costs. These findings demonstrate a need to understand the interactive relationship between employee HRFs, demographic and work organisation factors, and WC claims.

### Opportunities for health promotion

In the bivariate analyses, we found that workers who subsequently experienced a WC claim were positively and significantly more likely to report smoking, exposure to secondhand smoke, an abnormal BMI, stress over finances, poor seat belt use, no dental examination in the past 2 years, diabetes, digestive disorder and overall health rating. In general, workers reporting being in better health incurred fewer claims. Previously, using a national database of WC claims, the National Council on Compensation Insurance (NCCI) found that the WC claims with comorbidity diagnoses are on the rise.28 These findings demonstrate that there is an opportunity for professionals who interact with injured workers such as occupational medicine physicians, other healthcare providers, and safety and health managers to not only provide injury and illness care but also to offer health promotion interventions.

### Predictive relationship between HRFs and WC claims

Turning to the relationships between HRFs and WC claims, this prospective study across a wide range of industries shows that several HRFs are predictive of WC claims. Previously, researchers reported contradictory results regarding the relationship between HRFs and WC claim occurrence and cost. Musich *et al*18 found no significant relationship between HRFs and WC claim occurrence, but Kuhnen *et al*17 did. Kuhnen *et al*17 and Wright *et al*19 found no significant relationship between HRFs and WC claim cost,17 ,19 but Henke *et al*29 and Musich *et al*18 did. For example, Henke *et al*29 found that WC claim cost was significantly associated with obesity where WC costs were 46%, 59% and 135% higher for workers with Class I (BMI=30.0–34.9), Class II (BMI=35.0–39.9) and Class III (BMI=40.0+) obesity, respectively. In the present study, our unadjusted models demonstrated that HRFs were more predictive of WC claim cost, rather than WC claim occurrence. However, after adjusting for confounders, they became non-significant, suggesting that certain demographic and work organisational factors were more important predictors of subsequent claims and costs. The disparate findings among these studies may be due to a difference in HRAs, underlying sample, measurement or statistical methods.26

Stress was the only HRF to display a predictive relationship with WC claim occurrence and cost in our analysis. Similar to prior research that has demonstrated a relationship between stress and work-related injury,30 ,31 we found that *stress at work* was marginally predictive of increased odds of filing a WC claim. Work-related stress can stem from poorly functioning aspects of the psychosocial and physical work environment such as poor supervisory safety leadership, among other factors.32 We also found that *stress at home* predicted higher WC claims costs and, inversely, that *stress over finances* predicted lower WC claims costs. We speculate that workers who are experiencing stress over finances may return to work sooner to avoid lost wages or job loss.33 On the other hand, workers who are experiencing stress at home may have low social support, an important predictor of return to work.33 These findings support the need for employers to consider TWH strategies that reduce workplace and financial stress, and identify ways to assist workers in managing life stressors.15

### Strengths and limitations

Our study used a large sample of employees with different occupations and employment types from multiple companies of varying size and industries. However, there are four main limitations to our analysis. First, the results of our study may not be generalisable because samples were drawn from a non-randomised self-selected employer and employee population willing to participate in the HRM programme. However, we consider our findings more broadly generalisable than much of the published literature on this subject, especially as compared to single-company studies. Furthermore, it is important to note that we consider our study reflective of employers and employees who would willingly participate in a WC insurer-sponsored HRM programme. This universe of employers interested in engaging in health promotion is of practical interest to researchers as well as practitioners, especially as the number of organisations offering comprehensive worksite wellness programmes grows in the USA and abroad.

Second, measurement error may have inhibited detection of significant effects. Many of the HRFs typically associated in prior research with increased healthcare costs were not significantly predictive of WC claims in our prospective study.9 These disparate findings may stem from measurement issues when using a self-report HRA, as well as differences in HRAs being used. The HRF variables were self-reported and potentially subject to recall bias, which would have biased our results towards the null. This is supported by research showing that regardless of a good or bad health test result, individuals are more likely to recall a value as being better than it actually is.34 Measuring HRFs with limited response scales may have limited our ability to detect HRFs and their relationship with WC claims. It may be useful for future research to investigate more subjective self-report symptom screening tools as well as objective records (eg, biometrics) of HRFs. Although it may be difficult to link WC and biometric data from healthcare data among small businesses due to issues of privacy and buy-in from all stakeholders involved, it may be an important next step.

Third, the claims represented in the present study are mostly reflective of acute injuries rather than chronic health conditions in a 1-year post-HRA timeframe. Furthermore, the claims are also likely to underestimate the true cost of injury including lost wages over a lifetime. Therefore, the WC data in this study likely underestimate the true occurrence and cost of work-related injuries and illnesses. Thus, we may have been unable to detect a relationship between the HRFs and latent occupational illnesses and injuries.

Finally, it should be noted that we performed multiple comparisons in our models and some significant findings may have arisen by chance. Indeed, the negative predictive relationship observed between no physical examination in the past 2 years and cost was counter intuitive.

While the intent of this paper is to focus on the relationship of HRFs to WC claims, our findings in relation to the control variables cannot be ignored. The adjusted models suggest that by and large, HRFs are not highly predictive of subsequent WC claims, after accounting for differences between employees and work organisation factors. However, it is likely that our adjusted models were over-adjusted, which may have obscured the true relationship. We hypothesise that the control variables do play an important role in the relationship between HRFs and WC claims; however, this relationship may only hold true in specific instances. For example, Ostbye *et al*35 found that employees with a BMI of >30 who were employed in high-risk jobs had a relative risk of 7.04 (95% CI 5.95 to 8.33) for WC claims, compared to employees with normal BMIs and who worked in low-risk jobs.

## Conclusions

Our study demonstrates that the aetiology of occupational injuries may involve organisational and individual risk factors. Work organisation factors such as industry-related as well as employee-related HRFs including employee-perceived stress were predictive of the occurrence and cost of WC claims in the present study. This study is a starting point for examining the interplay between employee demographics and work organisation, HRFs and WC claims, and how interventions involving employee and employer can be integrated to promote TWH. Future research should consider these multilevel relationships.

## Footnotes

*   Twitter Follow Natalie Schwatka at [@nvschwatka](http://twitter.com/nvschwatka)

*   Contributors Each author has made substantial contributions to this study, provided help revising this paper for important intellectual content, gave final approval for this version of the paper and agree to be accountable for all aspects of this work.

*   Funding This study was funded by Pinnacol Assurance.

*   Competing interests All coauthors have filled out the ICMJE form. The competing interests include support from Pinnacol Assurance to conduct this study.

*   Provenance and peer review Not commissioned; externally peer reviewed.

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: [http://creativecommons.org/licenses/by-nc/4.0/](http://creativecommons.org/licenses/by-nc/4.0/)

## References

1.  Bureau of Labor Statistics. Employer reported workplace injury and illness summary. 2014. [http://www.bls.gov/news.release/osh.nr0.htm](http://www.bls.gov/news.release/osh.nr0.htm) (accessed 12 Jan 2015).
    
    

2.  Bureau of Labor Statistics. Census of fatal occupational injuries summary, 2013. 2014. [http://www.bls.gov/news.release/cfoi.nr0.htm](http://www.bls.gov/news.release/cfoi.nr0.htm) (accessed 12 Jan 2015).
    
    

3.  Bureau of Labor Statistics. Nonfatal occupational injuries and illnesses requiring days away from work, 2013. 2014. [http://www.bls.gov/news.release/pdf/osh2.pdf](http://www.bls.gov/news.release/pdf/osh2.pdf) (accessed 20 Mar 2015).
    
    

4.  Davis J, Bar-Chaim Y. Workers compensation claim frequency. 2011. [https://www.ncci.com/Articles/Documents/II\_2011\_Claim\_Freq\_Research.pdf](https://www.ncci.com/Articles/Documents/II\_2011_Claim_Freq_Research.pdf) (accessed 13 Mar 2016).
    
    

5.  Leigh JP. Economic burden of occupational injury and illness in the United States. Milbank Q 2011;89:728–72. [doi:10.1111/j.1468-0009.2011.00648.x](http://dx.doi.org/10.1111/j.1468-0009.2011.00648.x)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1111/j.1468-0009.2011.00648.x&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=22188353&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000298357400007&link_type=ISI) 

6.  Ward BW, Schiller JS. Prevalence of multiple chronic conditions among US adults: estimates from the National Health Interview Survey, 2010. Prev Chronic Dis 2013;10:120203. [doi:10.5888/pcd10.120203](http://dx.doi.org/10.5888/pcd10.120203)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.5888/pcd10.120203&link_type=DOI) 

7.  Center for Behavioral Health Statistics and Quality. Behavioral health trends in the United States: results from the 2014 National Survey on Drug Use and Health. 2015. [http://www.samhsa.gov/data/](http://www.samhsa.gov/data/) (accessed 29 Apr 2015).
    
    

8.  Paez KA, Zhao L, Hwang W. Rising out-of-pocket spending for chronic conditions: a ten-year trend. Health Aff (Millwood) 2009;28:15–25. [doi:10.1377/hlthaff.28.1.15](http://dx.doi.org/10.1377/hlthaff.28.1.15)
    
    [Abstract/FREE Full Text](http://oem.bmj.com/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToiaGVhbHRoYWZmIjtzOjU6InJlc2lkIjtzOjc6IjI4LzEvMTUiO3M6NDoiYXRvbSI7czoxOToiL29lbWVkLzc0LzEvMTQuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

9.  Goetzel RZ, Pei X, Tabrizi MJ, et al. Ten modifiable health risk factors are linked to more than one-fifth of employer-employee health care spending. Health Aff (Millwood) 2012;31:2474–84. [doi:10.1377/hlthaff.2011.0819](http://dx.doi.org/10.1377/hlthaff.2011.0819)
    
    [Abstract/FREE Full Text](http://oem.bmj.com/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToiaGVhbHRoYWZmIjtzOjU6InJlc2lkIjtzOjEwOiIzMS8xMS8yNDc0IjtzOjQ6ImF0b20iO3M6MTk6Ii9vZW1lZC83NC8xLzE0LmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

10. Goetzel RG, Anderson DR, Whitmer RW, et al. The relationship between modifiable health risks and health care expenditures. An analysis of the multi-employer HERO health risk and cost database. J Occup Environ Med 1998;40:843–54. [doi:10.1097/00043764-199810000-00003](http://dx.doi.org/10.1097/00043764-199810000-00003)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/00043764-199810000-00003&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=9800168&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000076449700003&link_type=ISI) 

11. Goetzel RZ, Carls GS, Wang S, et al. The relationship between modifiable health risk factors and medical expenditures, absenteeism, short-term disability, and presenteeism among employees at Novartis. J Occup Environ Med 2009;51:487–99. [doi:10.1097/JOM.0b013e31819eb902](http://dx.doi.org/10.1097/JOM.0b013e31819eb902)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/JOM.0b013e31819eb902&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=19337132&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

12. Frey JJ, Osteen PJ, Berglund PA, et al. Predicting the impact of chronic health conditions on workplace productivity and accidents: results from two US Department of Energy national laboratories. J Occup Environ Med 2015;57:436–44. [doi:10.1097/JOM.0000000000000383](http://dx.doi.org/10.1097/JOM.0000000000000383)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/JOM.0000000000000383&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=25654634&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

13. National Center for Health Statistics. Health, United States, 2013: with special feature on prescription drugs. 2014. [http://www.cdc.gov/nchs/hus/previous.htm#editions](http://www.cdc.gov/nchs/hus/previous.htm#editions) (accessed 13 Mar 2016).
    
    

14. The Kaiser Family Foundation. Preventative services covered by private health plans under the Affordable Care Act. 2015. [http://kff.org/health-reform/fact-sheet/preventive-services-covered-by-private-health-plans/](http://kff.org/health-reform/fact-sheet/preventive-services-covered-by-private-health-plans/) (accessed 21 Jul 2015).
    
    

15. Schill AL, Chosewood LC. The NIOSH Total Worker Health™ program: an overview. J Occup Environ Med 2013;55:S8–11. [doi:10.1097/JOM.0000000000000037](http://dx.doi.org/10.1097/JOM.0000000000000037)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/JOM.0000000000000037&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=24284752&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

16. Kubo J, Goldstein BA, Cantley LF, et al. Contribution of health status and prevalent chronic disease to individual risk for workplace injury in the manufacturing environment. Occup Environ Med 2014;71:159–66. [doi:10.1136/oemed-2013-101653](http://dx.doi.org/10.1136/oemed-2013-101653)
    
    [Abstract/FREE Full Text](http://oem.bmj.com/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NToib2VtZWQiO3M6NToicmVzaWQiO3M6ODoiNzEvMy8xNTkiO3M6NDoiYXRvbSI7czoxOToiL29lbWVkLzc0LzEvMTQuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

17. Kuhnen AE, Burch SP, Shenolikar RA, et al. Employee health and frequency of workers’ compensation and disability claims. J Occup Environ Med 2009;51:1041–8. [doi:10.1097/JOM.0b013e3181b32071](http://dx.doi.org/10.1097/JOM.0b013e3181b32071)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/JOM.0b013e3181b32071&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=19687757&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

18. Musich S, Napier D, Edington DW. The association of health risks with workers’ compensation costs. J Occup Environ Med 2001;43:534–41. [doi:10.1097/00043764-200106000-00005](http://dx.doi.org/10.1097/00043764-200106000-00005)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/00043764-200106000-00005&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=11411325&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000169158900003&link_type=ISI) 

19. Wright DW, Beard MJ, Edington DW. Association of health risks with the cost of time away from work. J Occup Environ Med 2002;44:1126–34. [doi:10.1097/00043764-200212000-00006](http://dx.doi.org/10.1097/00043764-200212000-00006)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/00043764-200212000-00006&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=12500454&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

20. Yen L, Schultz A, Schnueringer E, et al. Financial costs due to excess health risks among active employees of a utility company. J Occup Environ Med 2006;48:896–905. [doi:10.1097/01.jom.0000235987.75368.d0](http://dx.doi.org/10.1097/01.jom.0000235987.75368.d0)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/01.jom.0000235987.75368.d0&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=16966956&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000240527500005&link_type=ISI) 

21. Newman LS, Stinson KE, Metcalf D, et al. Implementation of a worksite wellness program targeting small businesses: the Pinnacol assurance health risk management study. J Occup Environ Med 2015;57:14–21. [doi:10.1097/JOM.0000000000000279](http://dx.doi.org/10.1097/JOM.0000000000000279)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/JOM.0000000000000279&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=25563536&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

22. Goetzel RZ, Tabrizi M, Henke RM, et al. Estimating the return on investment from a health risk management program offered to small Colorado-based employers. J Occup Environ Med 2014;56:554–60. [doi:10.1097/JOM.0000000000000152](http://dx.doi.org/10.1097/JOM.0000000000000152)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/JOM.0000000000000152&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=24806569&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

23. Wellsource. What is a health risk assessment? 2015. [http://www.wellsource.com/health-risk-assessments.html](http://www.wellsource.com/health-risk-assessments.html) (accessed 3 Apr 2015).
    
    

24. Kessler RC, Barber C, Beck A, et al. The World Health Organization Health and Work Performance Questionnaire (HPQ). J Occup Environ Med 2003;45:156–74. [doi:10.1097/01.jom.0000052967.43131.51](http://dx.doi.org/10.1097/01.jom.0000052967.43131.51)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/01.jom.0000052967.43131.51&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=12625231&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000180965800006&link_type=ISI) 

25. Wang PS, Beck A, Berglund P, et al. Chronic medical conditions and work performance in the health and work performance questionnaire calibration surveys. J Occup Environ Med 2003;45:1303–11. [doi:10.1097/01.jom.0000100200.90573.df](http://dx.doi.org/10.1097/01.jom.0000100200.90573.df)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/01.jom.0000100200.90573.df&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=14665817&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000187227300012&link_type=ISI) 

26. Manning WG, Basu A, Mullahy J. Generalized modeling approaches to risk adjustment of skewed outcomes data. J Health Econ 2005;24:465–88. [doi:10.1016/j.jhealeco.2004.09.011](http://dx.doi.org/10.1016/j.jhealeco.2004.09.011)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1016/j.jhealeco.2004.09.011&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=15811539&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000228715000003&link_type=ISI) 

27. Liu L, Strawderman RL, Cowen ME, et al. A flexible two-part random effects model for correlated medical costs. J Health Econ 2010;29:110–23. [doi:10.1016/j.jhealeco.2009.11.010](http://dx.doi.org/10.1016/j.jhealeco.2009.11.010)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1016/j.jhealeco.2009.11.010&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=20015560&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000275587300008&link_type=ISI) 

28. Laws C, Colon D. Comorbidities in workers compensation. 2012:1–27. [https://www.ncci.com/documents/Research-Brief-Comorbidities-in-Workers-Compensation-2012.pdf](https://www.ncci.com/documents/Research-Brief-Comorbidities-in-Workers-Compensation-2012.pdf) (accessed 13 Mar 2016).
    
    

29. Henke RM, Carls GS, Short ME, et al. The relationship between health risks and health and productivity costs among employees at Pepsi Bottling Group. J Occup Environ Med 2010;52:519–27. [doi:10.1097/JOM.0b013e3181dce655](http://dx.doi.org/10.1097/JOM.0b013e3181dce655)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1097/JOM.0b013e3181dce655&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=20431407&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

30. Nakata A, Ikeda T, Takahashi M, et al. Impact of psychosocial job stress on non-fatal occupational injuries in small and medium-sized manufacturing enterprises. Am J Ind Med 2006;49:658–69. [doi:10.1002/ajim.20338](http://dx.doi.org/10.1002/ajim.20338)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1002/ajim.20338&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=16758484&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000239290300008&link_type=ISI) 

31. Leung MY, Chan IYS, Yu J. Preventing construction worker injury incidents through the management of personal stress and organizational stressors. Accid Anal Prev 2012;48:156–66. [doi:10.1016/j.aap.2011.03.017](http://dx.doi.org/10.1016/j.aap.2011.03.017)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1016/j.aap.2011.03.017&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=22664679&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 

32. Kelloway EK, Barling J. Leadership development as an intervention in occupational health psychology. Work Stress 2010;24:260–79. [doi:10.1080/02678373.2010.518441](http://dx.doi.org/10.1080/02678373.2010.518441)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1080/02678373.2010.518441&link_type=DOI) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000283246900004&link_type=ISI) 

33. Dembe AE. The social consequences of occupational injuries and illnesses. Am J Ind Med 2001;40:403–17. [doi:10.1002/ajim.1113](http://dx.doi.org/10.1002/ajim.1113)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1002/ajim.1113&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=11598991&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000171326500009&link_type=ISI) 

34. Croyle RT, Loftus EF, Barger SD, et al. How well do people recall risk factor test results? Accuracy and bias among cholesterol screening participants. Health Psychol 2006;25:425–32. [doi:10.1037/0278-6133.25.3.425](http://dx.doi.org/10.1037/0278-6133.25.3.425)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1037/0278-6133.25.3.425&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=16719615&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000237761300020&link_type=ISI) 

35. Ostbye T, Dement JM, Krause KM. Obesity and workers’ compensation: results from the Duke Health and Safety Surveillance System. Arch Intern Med 2007;167:766–73. [doi:10.1001/archinte.167.8.766](http://dx.doi.org/10.1001/archinte.167.8.766)
    
    [CrossRef](http://oem.bmj.com/lookup/external-ref?access_num=10.1001/archinte.167.8.766&link_type=DOI) 
    
    [PubMed](http://oem.bmj.com/lookup/external-ref?access_num=17452538&link_type=MED&atom=%2Foemed%2F74%2F1%2F14.atom) 
    
    [Web of Science](http://oem.bmj.com/lookup/external-ref?access_num=000245903100004&link_type=ISI)