Guzman, Jaime MD, MSc; Ibrahimova, Aybaniz MD; Tompa, Emile PhD; Koehoorn, Mieke PhD; Alamgir, Hasanat PhD
Economic evaluations of occupational health and safety (OHS) programs are increasingly reported and used to inform program choices and funding decisions.1 Nevertheless, published economic evaluations do not include the nonwage losses experienced by workers as a result of work injury. These focus primarily on losses accruing to an employer or insurer.2 At best, these include a broad estimate of nonwage losses, such as 1 or 1.5 times the wage-replacement and health care payments.3 Nonwage losses include the time lost from household and community involvement by the worker and by the persons assisting the worker, out-of-pocket payments, and the loss of quality of life (QOL). Ignoring nonwage losses makes it possible to conclude that an OHS intervention is cost-beneficial simply by shifting the costs to the worker and family.
The aim of the study was to record the economic and QOL impacts of the most common types of injuries experienced by health care workers in British Columbia, Canada, on the basis of a direct report from the workers. The study objectives were as follows:
1. To estimate the typical nonwage economic and QOL losses experienced by health care workers who suffer occupational injuries.
2. To define how type of injury and the place of residence affect these losses.
3. To identify other factors affecting these losses.
The following findings were expected: large nonwage losses for health care workers (between 50% and 100% of wage losses); substantial initial decreases in QOL that disappear for most but not all by the 12th week; and that losses would vary with type of injury, the region of residence, and other factors; with larger losses for back injuries and the lower mainland region of British Columbia (higher costs of goods and services).
The study used an inception cohort design and included consenting health care workers who (1) filed a claim with the provincial workers' compensation agency for a loss-time or medical-aid occupational injury within 2 weeks of the incident; (2) were full-time, part-time, or casual workers of a regional health care authority in British Columbia; and (3) reported one of the following six injury types of injury: musculoskeletal back injury (MSI-back), musculoskeletal neck and/or upper limb injury (MSI-neck/upper limb), other musculoskeletal injuries (MSI-other); exposures to blood and body fluids; bruises, contusions, and cuts without exposure to blood and body fluids; or irritations and allergies to skin, respiratory tract, and mucous membranes. Workers not employed by the regional health care authorities (physicians, students, and subcontracted facility workers), those unable to complete reports in English, and those who reported the injury 2 weeks or more after the incident were excluded from the study. A stratified sampling design was used to obtain representation from each of the six types of injury and each of the four regions of the province: Lower Mainland, Interior, Vancouver Island, and Northern. This study was approved by the Behavioural Research Ethics Board of the Office of Research Services, University of British Columbia.
The study was conducted in the province of British Columbia, Canada, with participants recruited between October 2009 and March 2011. Once every week during the study period, the provincial workers' compensation agency produced a list of health care workers having submitted a claim during the previous week (both time-loss and medical-aid claims). The individuals in the list were contacted by the agency staff to ascertain consent to provide their contact information to the research team. Then, a research assistant mailed a recruitment package to the prospective participants and followed up with them by phone.
After obtaining the appropriate informed consent, study participants were asked to record economic information daily in a questionnaire developed for this study and to report the information (online, by phone, or by mail) to researchers once a week for the first 12 weeks after the injury (the data reported in this study). A few study participants, who did not fully recover after 12 weeks, provided additional data until 12 months after the injury. Nevertheless, these data were not used in this study because of a small number of such participants.
The study questionnaire was pilot-tested on a separate sample of 12 injured workers and refined on the basis of their feedback, before the main study. The questionnaire collected daily economic information on seven sections: (1) out-of-pocket payments for injury-related health care services and supplies, (2) out-of-pocket payments for injury-related non–health care expenses, such as transportation for medical visits, (3) regular and overtime hours of work lost because of injury, (4) hired-help payments, (5) number of hours friends or relatives assisted the injured worker, (6) time taken off work by others (partner or family member), and (7) hours of leisure and volunteering lost because of injury. The impact of the injury on health-related QOL of the workers was documented by completing the European Quality of Life–5 Dimensions (EQ-5D) questionnaire once a week. This is a validated five-item questionnaire that also provides a visual analogue scale (EQ-VAS) to assess overall perceived health status by using a 0- to 100-point scale (0 is equivalent to worst and 100 to best possible health). The EQ-5D is a widely used instrument in economic evaluations of medical interventions.4
Nonwage losses for each worker were calculated by the sum of out-of-pocket payments and hired-help payments, plus the estimated dollar value of the hours of nonhired help, the time taken off work by a partner or family member, and the hours lost from leisure and volunteering. The value of an hour of leisure or volunteering was estimated at the worker's hourly rate of pay, and the value of the time taken off by others was calculated by using the average hourly salary rate in British Columbia for 2010 (Can$20.32).5 Workers also reported the hours lost from regular and overtime work. The value of an hour of regular work was calculated at the hourly rate of pay, and the value of an overtime hour was calculated at 1.5 times the hourly rate.
The EQ-5D completed once a week was scored to derive the corresponding utility (EQ-5D Index) as per the EuroQol manual (EuroQol Group, 2010) and applied to the preceding week, at each interval for 12 weeks.6 The quality-adjusted days of life lost were estimated in two ways: assuming that the injured workers before the injury had the same EQ-5D Index score as the general Canadian population aged 35 to 44 years7 (a priori analysis); and assuming that the best EQ-5D Index score reported by the participants after injury was their preinjury baseline (secondary post hoc analysis). The 35- to 44-years-old group was used for comparison because the mean age for the Canadian health care workforce is 41.9 years.8 Results of perceived health status (EQ-VAS) were analyzed with descriptive statistics but not included in the calculation of the quality-adjusted days of life lost.
Mean with standard deviation and median with 25th and 75th percentiles of the losses were reported in 2010 Canadian dollars. To estimate the mean economic losses per episode of injury in the province, weights were attributed to each type of injury in each region on the basis of injury rates published previously by Alamgir et al9 and 2010 workforce totals from the provincial workers' compensation agency Web site. Only MSI losses were estimated, because the low number of participants in the non-MSI categories would make the estimates for such injuries unreliable. Thus, exposures to blood and body fluids; bruises, contusions, and cuts without exposure to blood and body fluids; and irritations and allergies to skin, respiratory tract, and mucous membranes were collapsed in a single non-MSI category. To examine the possibility that study participants chose to respond because they had more severe injuries than nonparticipants, we used claim payments by the workers' compensation agency as a proxy for severity of injury. We compared claim payments for study participants and nonparticipants by using the Wilcoxon median test. Because of confidentiality restrictions, the linkage between our information and workers' compensation agency payments was probabilistic, and a match for 69% of participants and 65% of nonparticipants was possible.
Two-way analysis of variance was used to assess whether there was a statistically significant difference in the economic and QOL losses by type of injury and region of the province (prespecified analysis, hypothesis testing). Other factors influencing the economic and QOL losses were analyzed by multiple linear regression (exploratory analysis). Regression models included type of injury, geographical region, sex, occupation, and employment status. To avoid overfitting the models, we limited the number of independent variables and collapsed categories as follows: type of injury (back, neck/upper limb, lower limb, and non-MSI), occupation (registered nurse, licensed practical nurse, care aide, and other), and employment status (full-time and casual/part-time). Logarithmic transformation of dependent variables was done for all regression analyses because the data were skewed. Significance level was set at P < 0.05, and 95% confidence intervals were calculated. Statistical analyses were carried out, using the Statistical Package for the Social Sciences (Version 19.0, 2010; Statistical Products and Service Solutions, Inc, Chicago, IL).
During participant recruitment (October 2009 to March 2011), the provincial workers' compensation agency provided the contact information of 651 potential participants. Of those 651 potential participants, 222 consented to participate in the study (34% recruitment rate). Ultimately, 146 workers provided complete records (65.8% retention rate). The recruitment target of the study was 30 participants for each of the six injury types and 45 participants in each of four geographical regions, for a total of 180. These recruitment targets were not reached in the non-MSI group and in the most remote region of the province.
As expected, the majority of participants were women (n = 133; 91%), because most health care workers in Canada are women.8 Almost half of the study participants were registered nurses or licensed practical nurses (Table 1). The majority of the study participants were employed in acute care (59.6%), followed by long-term care (22.6%) and community care (17.8%). Among all study participants, 58.9% were full-time workers, 26% were part-time, and 15.1% were casual. Overall, 84% (n = 123) of the participants had MSI: MSI-neck/upper limb (40.4%), MSI-back (30.8%), and MSI-other (13.0%). The high proportion of MSI in the study sample corresponds to the pattern of occupational injury among health care workers reported by other sources.9,10
The study participants did not differ from those who declined to participate, on the basis of region of residence and sex, but study participants were less likely to have non-MSIs (16% vs 25%; P < 0.05). The study participants with MSI did not statistically differ from the nonparticipants with MSI in total claim payments provided by the provincial workers' compensation agency during the first 12 weeks after injury (P = 0.12). The median total payments during the 12 weeks after an MSI were Can$5520 (Can$1850 to Can$7825 for 25th and 75th percentile) for study participants, Can$4204 (Can$1162 to Can$6563) for study nonparticipants, and Can$2541 (Can$750 to Can$5812) for all health care workers across the province. Median wage replacement payments were Can$4453 for participants and Can$3403 for nonparticipants (P = 0.22). Median health care payments were Can$861 for participants and Can$570 for nonparticipants (P = 0.01).
Mean nonwage economic loss within the 12 weeks after MSI was Can$3131 (95% confidence interval, Can$3035 to Can$3226). These included Can$338 in out-of-pocket expenditures, 66 hours lost from leisure and volunteering, and 39.5 hours spent by others assisting the injured worker (Table 2). Nonwage losses corresponded to about 59% of the lost wages, resulting in a total mean economic loss of Can$8417 (95% confidence interval, Can$7213 to Can$9620). Median and 25th and 75th percentiles are also reported in Table 2.
Quality of Life
Results from the EQ-5D showed that during the first week after injury, 41.5% of the participants reported mobility problems, 59.8% problems with self-care, 85.7% problems with their usual activities, 87.2% pain or discomfort, and 50.4% feeling anxious or depressed. As expected, EQ-5D Index scores increased over time as workers recovered from their injury (better QOL). Shortly after the injury, workers with MSI reported worse EQ-5D Index scores than workers with other types of injury. Workers with exposure to body fluids reported lesser impact on QOL (Fig. 1).
The mean level of perceived health status (EQ-VAS) of study participants improved steadily from 55 in week 1 to 78 in week 12. For comparison, the mean EQ-VAS for the general Canadian population aged 40 to 49 years was 81.2 in 2000.11 Perceived health status varied, depending on type of injury. Workers with back injuries reported the lowest mean perceived health status, 43.2 in week 1 rising to 74.6 in week 12; whereas workers exposed to blood and body fluids reported the highest, 90.7 in week 1 rising to 93.6 in week 12.
Assuming that before the injury, study participants had the same QOL as the general Canadian population aged 35 to 44 years (mean EQ-5D Index score of 87.4),7 workers lost an equivalent of 1.0 quality-adjusted days of life in the first week and 0.2 days in week 12. Overall, each MSI resulted on average in 7.9 quality-adjusted days of life lost. Assuming that the best EQ-5D Index score achieved by the participant after injury was his or her preinjury baseline, each MSI resulted on average in 13.7 quality-adjusted days of life lost.
Impact of Type of Injury and British Columbia Region
The type of injury and British Columbia region had an impact on the total economic losses (two-way analysis of variance: F = 3.774; P < 0.001) and nonwage economic losses (two-way analysis of variance: F = 2.132; P = 0.013). The MSI-neck/upper limb had the highest economic loss in absolute terms; however, the differences when compared with MSI-back and MSI-other were not statistically significant. Nonmusculoskeletal injury was associated with significantly smaller total and nonwage economic losses. For total economic losses, injuries occurring in the Lower Mainland were associated with the highest losses. Nevertheless, there were no statistically significant differences for nonwage economic losses among regions. The QOL losses were lower for non-MSI and for health care workers residing in the Interior region.
Other Factors Affecting Economic and QOL Losses
Table 3 shows the results of the exploratory multiple regression analysis of characteristics associated with the total economic losses in the cohort. The model was significant (P < 0.001) with adjusted R2 = 0.255, indicating that the model explains about 25% of the variation in total losses. In addition to the type of injury and region of the province, occupation was important, with greater economic losses among registered nurses. When the regression analysis was repeated by using only nonwage economic losses as the dependent variable, the model was still significant. As with the previous model, non-MSIs were associated with less nonwage losses and nurses had higher nonwage losses (P < 0.001). The higher nonwage losses for nurses were still statistically significant when leisure and volunteering time was valued at the same rate for all occupations (Can$20.32), rather than at the rate of pay for the injured worker.
Table 4 shows results of the multiple regression analysis for QOL losses, assuming preinjury QOL was that of the average Canadian aged 35 to 44 years. This model was significant (P < 0.001), with an adjusted R2 = 0.17. In addition to the type of injury and region of the province, no other factor was found to be associated with the magnitude of the QOL losses. Using the highest (best achieved) postinjury EQ-5D Index score as the worker's baseline, the analysis revealed that none of the predictors had significant impact on QOL losses (R2 = 0.02).
This cohort study sought to determine the nonwage losses (both economic and QOL) associated with occupational injuries among health care workers in a Canadian province and to explore factors affecting the magnitude of these losses. Musculoskeletal injuries among health care workers in this Canadian province resulted in a mean nonwage economic loss of Can$3131 in 2010. The corresponding loss in QOL was 7.9 quality-adjusted days of life per MSI, when the health state of the general Canadian population aged 35 to 44 years was used as the preinjury baseline. This impact increased to 13.7 quality-adjusted days when the best health state achieved by the worker after injury was used as their baseline. Nonwage economic losses included a mean of Can$338 in out-of-pocket expenditures, 66 hours lost from leisure and volunteering, and 39.5 hours spent by others assisting the injured worker. Nonwage economic losses corresponded to 59% of the value of lost wages, which is in line with the study hypothesis. The type of injury, region of the province, and occupation influenced the magnitude of these losses. It was not possible to estimate detailed losses associated with specific types of non-MSI because of the small number of workers with these injuries in the cohort.
Countries, provinces, and states differ in their policies and regulations governing compensation for occupational injuries.12 In North America, workers' compensation agencies provide wage replacement benefits calculated as a defined portion of the lost wages. Nonwage losses do not enter in these calculations. Some nonwage losses might be eligible for reimbursement; but for the most part, nonwage losses are borne by the worker and his or her family and community.
Jurisdictions also differ in how QOL losses (often referred to as pain, suffering, and loss of enjoyment of life) from occupational injury are compensated. In some jurisdictions, no compensation is provided; whereas in others, a lump sum is given if workers do not fully recover from their injury (ie, they experience a permanent impairment).13 For injuries that do not fall under the legal umbrella of workers' compensation, it is not unusual for courts to award large amounts to compensate for pain, suffering, and loss of enjoyment of life.14
The study findings confirmed that work-related injuries substantially affect the QOL, although by week 12, the majority of the respondents recovered. Rather than assigning a dollar value to QOL losses, the quality-adjusted days of life lost as a result of injury were reported by using a common and well-validated utility measure, the EQ-5D. Ideally, to calculate the quality-adjusted days of life lost, the QOL of the person before the injury needs to be determined. Participants' retrospective reports about their QOL in the weeks before the injury were considered unreliable because of a recall bias. Instead, the quality-adjusted days of life lost were calculated assuming that before the injury, the respondents were as healthy as an average Canadian aged 35 to 44 years. A previous study in Alberta reported that the mean value of the EQ-5D Index score for that age group was 87.4 in 2000.7 As the results were analyzed, it was realized that by week 12 after injury, the EQ-5D Index scores in the participants were significantly larger than 87.4. Therefore, a post hoc analysis was performed assuming that the QOL before the injury was equal to the best health state achieved by the participant during the 12 weeks after the injury. Values under both the assumptions were presented: 7.9 and 13.7 days. There is a cogent argument to rely primarily on the latter estimate, because it is logical to assume that the QOL before the injury was better than the best health achieved during the 12-week postinjury recovery period (unless one proposes that having an injury improves the QOL). Nevertheless, the counterargument would be that the high EQ-5D Index scores observed during recovery were an artifact of multiple applications of the same instrument week after week, because participants wanted to reflect the gradual improvements experienced after injury.
Nonwage losses were highly skewed, with median values that were substantially smaller than the mean values, because few workers incurred very large losses. In fact, a small number of participants continued to incur losses well beyond 12 weeks after the injury. This study cannot offer precise estimates for these high-cost chronic disabling injuries, because it concentrated on the first 12 weeks after injury. The natural skewness of losses after an occupational injury needs to be considered in economic evaluations and in setting compensation policies.
Factors Affecting Economic and QOL Losses
Nonwage losses were similar for all MSIs and perhaps larger for neck and upper limb injuries, rather than low back injuries as hypothesized. As a group, non-MSI did show statistically significant lower nonwage losses than MSI. It is well known that average payments for wage replacement and health care provided by workers' compensation systems are higher for MSI than non-MSI.10 Other studies indicate that exposures to blood and body fluids are less likely to be reported than other injuries; and when reported, health care workers do not seem to place a lot of importance on the injury.15 This might partially explain the difficulty in recruiting workers with non-MSI. The findings emphasize the importance of prevention of MSI among health care workers because they are associated with greater costs. Potential benefits from such prevention programs would be much greater when the nonwage losses for the workers and their families are taken into account.
There were significant regional variations in total losses, but these were no longer significant when only nonwage losses were considered and models were adjusted for other variables. It was hypothesized that regional differences existed because of differences in days lost from work after injury, the local cost of goods and services, and salaries across the province. Differences in days lost from work after injury across regions in this province have been noted (WorkSafeBC, unpublished data, 2009), although the reason for such differences is unclear. The observed regional differences, across a single province in a single industrial sector with the same injury reporting and compensation systems, should alert researchers to the need to explore regional differences in their own setting when planning economic evaluations that include data for a large geographical area.
The observed differences on losses by type of injury and occupation are informative and could be used in future research to adjust the valuation of consequences, where these data cannot be collected directly. At first, it was felt that the greater total losses reported by registered nurses were explained by their higher rate of pay than other occupations represented in the study. But the difference did not disappear when the analyses were repeated on nonwage losses only or using the same hourly rate of pay for all occupations. The effect of occupation on losses after occupational injury should be studied in future research.
The QOL losses varied with type of injury and region of the province but not with any of the other factors examined. The effects of these factors on QOL losses were not statistically significant when the best-achieved postinjury health status was used as a baseline. The reason for this was not clear.
The reviews by Weil16 and Reville et al17 provide a solid framework for understanding options for valuation of the consequences of work injury: documenting household income before and after injury, documenting how household members changed the time allotted to valued activities as a result of injury, and documenting the global loss in QOL experienced by the worker.16 This study provides information on the latter two ways of valuing consequences to the worker.
The study estimates of total losses for health care workers after occupational MSI (Can$8417 in the first 12 weeks) were in line with published studies. The consequences of injury across sectors and all occupational injuries have been reported to be about Can$8000 to Can$10,000 in the first year after injury.18 The literature also shows that workers' compensation does not cover all costs incurred by injured workers, because many injuries do not result in a compensation claim19; and even for compensated claims, compensation amounts do not cover all losses experienced by the worker.1
Study Strengths and Limitations
This study has a number of strengths. It is the first study to systematically quantify nonwage losses after an occupational injury in an inception cohort of health care workers. It contributes unique new estimates of those nonwage economic losses, their relative composition (out-of-pocket payments, time lost from leisure and volunteering, and time spent by others assisting the worker), and the relative value of nonwage losses in relation to wage losses determined with the same methods in the same population. Data were recorded daily and reported weekly, so that the effect of recall bias is minimized. In addition, QOL losses were estimated using a validated and standardized method based on the EQ-5D Index.
This study also has several limitations. The original study design called for recruitment of an equal number of participants in each injury category across all study regions; however, this proved impossible to achieve during the study time frame. Musculoskeletal injuries were overrepresented in the study cohort, making up 84% of the total. Musculoskeletal injuries are known to be very frequent among health care workers and often require a longer recovery time than other injuries. The low number of non-MSI included in the cohort did not allow detailed examination of each type of such injuries to be conducted.
Our estimates are based on self-report of workers accepting to participate in the study. Self-report was considered the only feasible way of assessing these nonwage losses. Although it is possible for injuries to aggravate preexisting conditions, we asked workers to specifically report expenses because of the injury and did not ask about the presence of any preexisting conditions. It is possible that the workers accepting to participate in the study did so because they had more severe injuries than the workers who decided not to participate; this would result on overestimation of the impact of the injury. Our comparison of MSI claim payments provided by the provincial workers' compensation agency, as a proxy for injury severity, supports this possibility. Although the total claim payments were not statistically significantly different, claim payments of nonparticipants were about 24% lower than payments of participants. Because there is no ethical way to ascertain nonwage losses without the informed consent of the worker, no study can entirely eliminate this kind of selection bias. One way of dealing with this in economic evaluations would be to conduct sensitivity analyses, using our estimates as the upper bound of MSI nonwage losses and using values 25% lower as the lower bound.
The study estimates did not include the cost of most medical care provided for the injuries. In Canada, physician and hospital services are not paid by the injured worker; these are directly covered by the universal health care system in the province and then charged to the provincial workers' compensation agency. There are no copayments or deductibles paid by the worker, and there are no coverage maximums or differential premiums based on age or previous health status. A number of other health care services (which during the time of the study in this province included chiropractic, physiotherapy, and other rehabilitation services) are either paid by extended benefit plans or paid out-of-pocket by the worker. The cost of these services may then be reimbursed to the worker within prescribed limits. Studies in other jurisdictions may need to add health care deductibles or copayments, if paid directly by the worker. In 2010, the provincial workers' compensation agency paid a mean of Can$592 for the health care provided to health care workers with MSI in this province (WorkSafeBC, unpublished data, 2012). We did not have access to records of the payments provided by the universal health care plan or extended benefits plans in our cohort, and the assessment of cost-shifting across these systems was beyond the scope of this study.20
The study did not include all people who work in health care but included only those considered employees of the regional health authorities. A large number of housekeeping workers and food service workers were contracted out, and thus, not considered employees at the time of this study. Physicians, residents, and students were also excluded.
No data on the impact of the injury on specific roles such as child care, home maintenance, or community service were available. These were grouped together in global estimates of the time needed for other persons to undertake those roles while the injured worker recovered and in a general category of leisure and volunteering.
A final limitation is that the study included only injuries, resulting in the filing of a loss-time or medical-aid claim with the provincial workers' compensation agency. Nothing could be said about the losses associated with injuries that did not result in such a claim.21
Most published economic analyses of OHS interventions consider economic losses for the employer or the insurer but ignore the nonwage losses that remain uncompensated consequences for workers experiencing occupational injuries. The information reported in this study can be used to integrate workers' perspectives into comprehensive economic evaluations of OHS programs to prevent MSI. Future economic evaluations should take into account these substantial additional losses borne by the workers and their families, which corresponded to 59% of wage losses in this study. Furthermore, governments and organizations determining compensation policies for injured workers should consider these nonwage losses in calculations of program benefits. Consideration of nonwage losses is likely to make the cost–benefit ratio of occupational injury prevention programs more favorable, thus making investments in such programs more justifiable at the societal level.
We thank the following people for their support in this study: Jeremy Parr, Boris Kuzeljevic, Saleema Dhalla, Isabelle Lindon, Elizabeth Wilcox, Ron Kelly, and Carol Shafer.
1. Tompa E, Dolinschi R, de Oliveira C. Practice and potential of economic evaluation of workplace-based interventions for occupational health and safety. J Occup Rehabil. 2006;16:375–400.
2. Tompa E, Dolinschi R, de Oliveira C, Irvin E. A systematic review of occupational health and safety interventions with economic analyses. J Occup Environ Med. 2009;51:1004–1023.
3. Alamgir H, Yu S, Fast C, Hennessy S, Kidd C, Yassi A. Efficiency of overhead ceiling lifts in reducing musculoskeletal injury among carers working in long-term care institutions. Injury. 2008;39:570–577.
4. Räsänen P, Roine E, Sintonen H, Semberg-Konttinen V, Ryynänen OP, Roine R. Use of quality-adjusted life years for the estimation of effectiveness of health care: a systematic literature review. Int J Technol Assess Health Care. 2006;22:235–241.
6. Feeny D. A utility approach to the assessment of health-related quality of life. Med Care. 2000;38(suppl):151–154.
7. Johnson JA, Pickard AS. Comparison of the EQ-5D and SF-12 health surveys in a general population survey in Alberta, Canada. Med Care. 2000;38:115–121.
8. Canadian Institute for Health Information. Canada's Health Care Providers. Ottawa, ON: Canadian Institute for Health Information; 2007:65.
9. Alamgir H, Cvitkovich Y, Yu S, Yassi A. Work-related injury among direct care occupations in British Columbia, Canada. Occup Environ Med. 2007;64:769–775.
10. Ngan K, Drebit S, Siow S, Yu S, Keen D, Alamgir H. Risks and causes of musculoskeletal injuries among health care workers. Occup Med (Lond). 2010;60:389–394.
12. Anema JR, Schellart AJ, Cassidy JD, Loisel P, Veerman TJ, van der Beek AJ. Can cross country differences in return-to-work after chronic occupational back pain be explained? An exploratory analysis on disability policies in a six country cohort study. J Occup Rehabil. 2009;19:419–426.
13. Spieler EA, Barth PS, Burton JF Jr, Himmelstein J, Rudolph L. Recommendations to guide revision of the Guides to the Evaluation of Permanent Impairment. American Medical Association. JAMA. 2000;283:519–523.
14. Boyce CJ, Wood AM. Money or mental health: the cost of alleviating psychological distress with monetary compensation versus psychological therapy. Health Econ Policy Law. 2010;5:509–516.
15. Makary MA, Al-Attar A, Holzmueller CG, et al. Needlestick injuries among surgeons in training. N Engl J Med. 2007;356:2693–2699.
16. Weil D. Valuing the economic consequences of work injury and illness: a comparison of methods and findings. Am J Ind Med. 2001;40:418–437.
17. Reville RT, Bhattacharya J, Sager Weinstein LR. New methods and data sources for measuring economic consequences of workplace injuries. Am J Ind Med. 2001;40:452–463.
18. Boden LI, Biddle EA, Spieler EA. Social and economic impacts of workplace illness and injury: current and future directions for research. Am J Ind Med. 2001;40:398–402.
19. Shannon HS, Lowe GS. How many injured workers do not file claims for workers' compensation benefits? Am J Ind Med. 2002;42:467–473.
20. Asfaw A, Souza K. Incidence and cost of depression after occupational injury. J Occup Environ Med. 2012;54:1086–1091.
21. Azaroff LS, Lax MB, Levenstein C, Wegman DH. Wounding the messenger: the new economy makes occupational health indicators too good to be true. Int J Health Serv. 2004;34:271–303.
Copyright © 2013 by the American College of Occupational and Environmental Medicine