Abstract
BACKGROUND:
Migrant workers have been identified in Europe, North America, Asia and Australia as a particularly vulnerable working population with a higher risk of work-related injury and mortality compared to non-migrant workers. Lack of English language proficiency is associated with an increased risk of work-related injury. Whether lack of English proficiency influences post-injury recovery or return to work outcomes remains unknown.
OBJECTIVE:
Using administrative data from a population based workers’ compensation dataset in the state of Victoria, Australia, we aimed to examine work-related injury rates, worker characteristics and compensation outcomes in workers who were not proficient in English. We hypothesized that the use of an interpreter service would be associated with a poorer post-injury recovery profile and worse return to work outcomes.
METHODS:
WorkSafe Victoria accepted non-fatal claims for injuries and illnesses reported between January 1, 2003, and December 31, 2012 by workers aged 15 to 74 (n = 402, 828 claims) were analysed. Consistent with prior research, we selected “use of an interpreter service” as the indicator of English language proficiency. The total and categorical compensable cost of recovery was used as recovery outcomes.
RESULTS:
Of these claims, 16,286 (4%) involved the use of an interpreter service (LOTE workers). Our analysis revealed that Victorian injured LOTE workers have significantly different demographic, occupational and injury characteristics compared to non-LOTE injured workers. Furthermore, we present novel evidence that LOTE status was associated with poorer long-term injury outcomes, observed as a greater healthcare utilisation and larger paid income benefits, after controlling for occupation, employment status and injury type compared to non-LOTE injured workers.
CONCLUSIONS:
These data suggest that English language proficiency is associated not only with the risk of work-related injury but also to the long-term recovery outcomes. We conclude that despite access to language interpreter services, injured LOTE workers experience English language proficiency dependent, and injury severity independent, recovery barriers which need to be overcome to improve long term recovery outcomes.
Introduction
Global migration is increasing with 28% of the resident population of Australia estimated to have been born overseas [1]. In 2013-14, Australia welcomed the arrival of 212,700 migrants [1]; with 82% of these migrants of working age between the ages of 20–44 [2]. Migrant workers have been identified in Europe, North America, Asia and Australia as a particularly vulnerable working population with a higher risk of work-related injury and mortality compared to non-migrant workers [3, 4]. The increased risk of work-related injury and mortality has been attributed to a number of risk factors including language barriers, limited occupational health and safety training, absence of workplace support, poor employment conditions and employment in occupations with a high level of injury risk [5–17]. Pransky et al. observed that only 31% of surveyed migrant workers in the U.S.A reported having any job safety training. Alarmingly, 25% of these workers who reported receiving job safety training revealed that their training was only delivered in English [11]. In another study of Latino immigrant construction workers, the median length of job safety training was reported to be less than one hour and workers with difficulties speaking/understanding English reported receiving even less training [10]. These findings demonstrate that a lack of language proficiency is a particularly important risk factor as it may influence other risk factors such as employment in high risk manual jobs [18] as well as exposure to higher risks within these roles (including higher workload) due to poor comprehension of occupational health and safety training, high workload and access to workplace support [19].
There is strong evidence to support that there is an increased injury risk in migrant workers internationally and a growing body of literature reporting the barriers and facilitators of migrant worker vulnerability. Whether the injury rates and characteristics of migrant workers in Australia reflect the data re-ported internationally is unclear due to the limited research evidence of work-related injury rates and post-injury outcomes in vulnerable migrant workers in Australia. In particular, the long-term outcomes and recovery profiles of these injured migrant workers in Australia are relatively unknown; with only three publications since the 1980s which report work-related injury rates using hospital admission and work-related mortality rates and no investigation oflong-term outcomes [3, 21]. These studies havereported that work-related mortality was greater inrecently arrived migrant workers (classified as resident in Australia for <5 years) compared to Austr-alian-born workers [3]. However, more recent findings using national death records, hospital admission records and national household surveyfindings report no difference in work-related mor-tality or injury between migrant and non-migrant workers [20, 21]. To our knowledge, there have been no studies that have examined recovery and returnto work outcomes in injured Australian migrantworkers.
Although it is widely accepted that effective com-munication and implementation of occupational he-alth and safety messages require both training andworkload management in order to reduce work-re-lated injury risk, limited research has been published investigating the association between English proficiency and work-related injury rates. Marvasti et al. and Premji conducted analysis of this kind in a population from the U.S.A and Canada respectively, and provided the first empirical evidence that lack of English proficiency (irrespective of native language) was positively associated with work-related injury risk [22, 23]. A recent Australian study revealed that 48% of newly arrived migrants who are employed reported use of a language other than English in their workplace [24] suggesting that English proficiency is likely to be a major factor contributing to worker vulnerability in Australia.
Using administrative data from a population based workers’ compensation dataset in the state of Victoria, Australia, we aimed to examine work-related injury rates, worker characteristics and long-term compensation outcomes in workers who were not proficient in English. Consistent with prior research, we selected “use of an interpreter service” as the indicator of English language proficiency [22]. Based on the international evidence, we hypothesise that use of an interpreter service will be associated with the characteristics of the injury and a poorer post-injury recovery profile. Characterisation of the injury rates and outcomes among vulnerable workers in the Victorian context will assist with determining the need for changes to public policy to address the occupational, social and economic vulnerability of migrant workers to reduce injury and improve outcomes in this segment of the labour force.
Methods
Setting and sample
Located in the south east of Australia, Victoria is the second most populous state with the population of 6,039,100 as of March 2016. WorkSafe Victoria (WSV) is a state-government organisation responsible for providing compensation services to injured workers such as treatment rehabilitation services, disability services, income benefits, and household support services irrespective of fault status. It is funded by a premium paid by the employer and is estimated to cover 90% of the working population in Victoria. The employer is liable for AUD$682 (as of 2016) of medical excess and the first 10 working days of time loss.
The initial sample included de-identified WorkSafe Victoria accepted non-fatality claims for injuries and illnesses reported between January 1, 2003, and December 31, 2012 by workers aged 15 to 74 (n = 402, 860 claims); 32 claims were excluded due to missing values. The final sample included 402, 828 claims.
Data and variables
The data was accessed through the Compensation Research Database (CRD). The CRD is an administrative database “established at the Institute for Safety Compensation and Recovery Research at Monash University, which holds over 20 years of population-based data for transport and workplace injury in the state of Victoria, Australia” [25].
English proficiency
WorkSafe Victoria does not systematically collect information on workers’ language and English proficiency and the CRD does not contain information regarding the migration status of injured workers such as the country of birth, visa status, or ethnicity. In the absence of this information, we aimed to use a surrogate marker which would most appropriately identify the most vulnerable migrant workers in the database. Given the well-reported impact of English language proficiency on migrant worker vulnerability, the use of a language interpreter service was used to categorise the most vulnerable migrant injured workers in our population. Therefore, in the current study we compared injured workers who accessed an interpreter service (language other than English (LOTE) injured workers- identified according to payment(s) made to interpreter services during the claim duration) with all other injured workers (non-LOTE (NLOTE) injured workers). We acknowledge that our approach will exclude workers with low English language proficiency who have not used formal interpreting services due factors such as presence of social support (family members, friends etc.), lack of awareness regarding availability of interpreter services or other access barriers. Nevertheless, this approach captures a subset of vulnerable migrant workers.
Factors and outcomes
The claim data included the following items: age at the time of accident, gender, occupation, workplace industry, size of employer, and information about injury such the mechanism and nature or method of injury. Age was grouped into four 15-year bands so the potential non-linear impacts could be better identified. Occupation coding followed Australian and New Zealand Standard Classification of Occupations (ANZSCO) [26], industry coding was based on Australian and New Zealand Standard Industrial Classification, and employer size was determined using the amount of annual remuneration paid to WorkSafe (< $1 million AUD=Small, $1–20 million AUD=Medium,> $20 million AUD=Large). All injury information followed the Type of Occurrence Classification System (TOOCS) coding system[27].
Cost outcomes are the compensated cost in Aust-ralian dollars. Six major cost categories were in-cluded in this paper to measure the recoveryoutcomes. This included the total cost of recovery which represents the overall burden of injury, as well as categorical costs: 1) income benefits, 2) medical costs, 3) hospital costs, 4) allied health costs, and 5) occupational rehabilitation cost.
The income benefit was the total amount paid toworkers to compensate for work absence due to work-related injury. Since the income benefits were calculated based on workers’ pre-injury income, wecontrolled for pre-injury weekly income in our models to ensure that the cost of income benefit rep-resented the duration of time off work and not thelevel of pre-injury salary. Allied health costs in-cluded payments for services such as physiotherapy, chiropractic, dental, speech therapy, psychology, osteopathy, dietitian services. Hospital costs included both inpatient and outpatient payments to private and public hospitals.
The decision to use cost of health service utiliza-tion rather than count was to ensure that the aggregated outcome measures do not give the same weight to the wide range of services included in each and so more accurately represent the burden of injury and complexity of recovery.
Statistical analysis
Four sets of analyses were undertaken in this study. First, we aimed to examine whether the work, workplace and injury characteristics of LOTE injured workers were similar to NLOTE injured workers. For this purpose, we compared the distribution of LOTE workers across the potential injury characteristic variables. To identify statistical significance, we ran a univariate logistic regression for each factor where the outcome is one if the claim is marked as a being LOTE, otherwise zero. We then ran three multivariate logistic models using the same outcome. The first model includes age, gender and work and workplace factors, the second model includes age, gender and injury factors, and the final model includes all the variables except the industry.
To study the association between cost outcomes and status of the worker, we used the logged value of the cost due to the skewed distribution of cost across the sample. Two main models were implemented: in the first model, we included all workers and in the second one we only included workers who have used services from that category. For example, when measuring the impact of LOTE worker status on hospital cost, we first included the whole sample regardless of using hospital services. We then ran an adjusted model where workers who did not receive any hospital services were removed and only workers who have a hospital cost greater than zero were analysed. To adjust for the difference in the follow-up periods, the year of injury report and its squared value are included in the analysis.
Ethics
The Monash University Human Research Ethics Committee has granted approval for the CRD project to collect, use, and disclose de-identified information (CF09/3150 – 2009001727).
Results
Our analysis identified 402,828 accepted workers’ compensation claims lodged between January 1st 2003 and December 31st 2012. Of these claims, 16,286 (4% ) involved the use of an interpreter service (LOTE workers). Between 2003 and 2012, the rate of workers receiving an interpreter service declined from 4.56% of total claims lodged in 2003 to 3.48% of total claims lodged in 2012.
Worker, work and workplace
Table 1 presents the distribution of LOTE workers across different factors associated with the worker, work and workplace as well as results of a univariate logistic model to test the significance of the difference within each factor.
Distribution of LOTE-workers based on the charactristics of worker, work and workplace
Distribution of LOTE-workers based on the charactristics of worker, work and workplace
Across the sample, 4.04% of the workers used interpreter services. When stratified by gender, the share is slightly higher among men than women (OR = 1.07, 95% CI = (1.04, 1.11)). The LOTE workers were also significantly more likely to be older than NLOTE workers; with 1.09% of workers in the 15–29 years of age bracket, 3.05% of workers between 30 and 44, 5.60% of workers 45–59, and 8.37% of workers 60–74. Moreover, LOTE workers were more likely to be employed in medium- sized workplaces and blue-collar occupations such as labourers (OR = 6.76, 95% CI = (6.05, 7.58)) or technicians and trades workers (OR = 3.06, 95% CI = (2.73, 3.44)). Manufacturing (9.86% being LOTE), followed by administrative and support services (8.11% ) had the highest portion of LOTE claims while mining (0.3% ) and financial and insurance services (0.34% ) had the lowest.
Table 2 presents the distribution of LOTE workers based on agency, mechanism and nature of their workplace injury. As the table shows, the major agency of injury for LOTE workers is machinery and fixed plant (OR = 1.33, 95% CI = (1.25, 1.42)) and less likely to be classed as caused by mobile plant and transport, unpowered tools or an animal (which includes humans and biological factors). Machinery and fixed plant agencies include subcategories such as cutting, slicing, sawing, crushing, pressing, rolling machinery, heating, cooking, baking, cooling equipment as well as conveyors, lifting, filling and bottling/packaging plant machinery.
Distribution of LOTE-workers based on the charactristics and setting of the injury
Distribution of LOTE-workers based on the charactristics and setting of the injury
The main mechanism of injury for LOTE workers was identified as sound and pressure (OR = 6.67, 95% CI = (6.21, 7.15)) and body stressing (OR = 1.87, 95% CI = (1.78, 1.96)). Body stressing includes subcategories such as muscular stress while lifting, carrying, putting down objects, handling objects and repetitive movement.
Finally, when compared to factures, nervous system and sensory organ diseases (OR = 4.83, 95% CI = (4.46, 5.24)) was the most likely nature of injury among LOTE workers. Fractures were selected as the reference group as they have a relatively consistent setting in terms of the severity of fracture injuries with a large sample size to ensure robust statistical ana-lysis against the other injury types. Furthermore, the analysis revealed that the subcategories of the nature of injury, deafness and carpal tunnel syndrome as well as traumatic amputation and disc displacement had higher than average share of LOTE workers compared to NLOTE workers.
Table 3 presents the adjusted models which re-vealed four significant insights. Firstly, the LOTE workers were more likely to be female (Male OR =0.626, 95% CI = (0.602, 0.650)) and aged between 45–59 (OR = 1.896, 95% CI = (1.825, 1.970)) or 60–74 (OR = 2.412, 95% CI = (2.283, 2.548)) compared to NLOTE who were most likely to be aged between30–44 years. Using managers as the reference, LOTE workers were much more likely to be employed in blue collar occupations such as labourers (OR =6.993, 95% CI = (6.256, 7.845)), machinery operators(OR = 5.394, 95% CI = (4.813, 6.068)) or techniciansand trade workers (OR = 3.861, 95% CI = (3.441, 4.347)). Moreover, in the adjusted model, the mainmechanisms of injury were identified to be sound and pressure (OR = 2.384, 95% CI = (2.114, 2.689))and body stressing (OR = 1.448, 95% CI = (1.361, 1.541)) followed by being hit by moving objects (OR = 1.169, 95% CI=(1.081, 1.265)) and mental stress (OR = 1.284, 95% CI = (1.038, 1.594)). Nerv-ous system and sensory organ diseases (OR = 1.384, 95% CI = (1.231, 1.556)), musculoskeletal and con-nective tissue diseases (OR = 1.156, 95% CI = (1.069,1.252)) and traumatic joint/ligament and muscle/tendon injury (OR = 1.108, 95% CI = (1.023,1.201)) were the most likely types of injuries and illnesses sustained.
The relationship between the LOTE status of the injured workers and their demographic, work, workplace and injury characteristics using multivariate regression
The relationship between the LOTE status of the injured workers and their demographic, work, workplace and injury characteristics using multivariate regression
Table 4 presents that the average total cost of recovery was $32,096 AUD for NLOTE workers compared to $105,064 AUD for LOTE. We observed that LOTE workers had consistently higher average costs of recovery for income benefits, hospital services, medical services, allied health services and occupational rehabilitation services compared to NLOTE workers. After controlling for worker characteristics, work type and injury, the total cost of recovery was on average four times higher for LOTE workers (4.02, 95% CI = (3.89, 4.15)) than NLOTE workers. On average, an injured LOTE worker received close to five times higher income benefits (4.93, 95% CI = 4.60, 5.28), 1.6 times higher hospital costs (1.59, 95% CI = (1.51, 1.68)), and al-most 2.5 times higher medical costs (2.44, 95% CI = (2.32, 2.57)). Furthermore, LOTE workers were identified to have received 2.84 (2.70, 3.00) times greater allied health service costs and almost five times greater (4.96, 95% CI = (4.71, 5.23)) occupational rehabilitation costs, respectively.
Total and categorized cost of recovery after workplace injury and the LOTE status
Total and categorized cost of recovery after workplace injury and the LOTE status
*The column presents the coefficients of regressing the log of cost values after adjusting for age, gender, occupation, pre-injury income and nature and mechanism of injury.
Furthermore, LOTE workers are more likely to re-ceive any income benefits and occupational rehabi-litation services (Table 4). This is further explored in Table 5 where the coefficients from Table 4 werecopied next to the coefficients from new modelswhere only injured workers who had used services from each category were included. We observed that the coefficient for the total cost does not change butthe coefficients for medical health and allied health services slightly increase. Among the injured workers who received income benefits, we observed thatLOTE workers received 3.38 (95% CI = (3.27, 3.49)) times greater benefits compared to NLOTE workers which was less than the results from the previousmodel with all workers included (4.93, 95% CI =(4.60, 5.28)). The data on pre-injury weekly ordinary income was available for workers who received income benefit. When we controlled for pre-injury income, the coefficient was 3.58 (95% CI = (3.47, 3.7)). This indicates the higher cost of income benefit for LOTE workers was not due to their higher pre-injury income, but more likely due to longer duration of time loss.
Total and categorized cost of recovery after workplace injury and LOTE status
*The column presents the coefficients of regressing the log of cost values after adjusting for age, gender, occupation, pre-injury income and nature and mechanism of injury.
Finally, the coefficients for hospital services dro-pped from 1.59 (95% CI = (1.51, 1.68)) in NLOTE workers to 1.33 (95% CI = (1.28, 1.38)) in LOTE workers and the coefficient for occupational rehabi-litation substantially decreased to 1.16 (95% CI =1.13, 1.19) in the LOTE workers.
Our analysis revealed that Victorian injured LOTE workers have significantly different demographic, occupational and injury characteristics compared to NLOTE injured workers. Furthermore, we present novel evidence that LOTE status was associated withpoorer long-term injury outcomes, observed as agreater number of days off work and greater heal-thcare utilisation, which was independent of occupation, employment status and injury severity com-pared to NLOTE injured workers. Conducting research to better understand the reasons underlying the increased work-related injury risk is notoriously challenging as migrant workers are transient, are more likely to work outside of standard working hours and are poorly represented in jurisdictional injury datasets (due to lack of visibility or limitations in data collection regarding migration status) [28]. One of the major findings of our analysis of injured workers with low English language proficiency was that we identified a population of 16,286 LOTE workers with injuries. The most recent Australian labour force estimates demonstrate that 28% of the Australian population over the age of 15years were born overseas [29]. Furthermore, 9% of the Australian population are reported to have been over the age of 15 at the time of arrival [29], including 5.9% arriving from non-English speaking countries [29]. 2011 Australian census data revealed that 11.5% of migrants who arrived in the past 5 years reported lack of English proficiency [30]. In terms of the broader Australian population, 1.6% of the Australian population over the age of 15, reported they speak English not well or not at all [29]. We observed that 4% of all injured workers between 2003–2013 were LOTE with injuries. Therefore, our data suggests that our analysis identified a greater proportion of LOTE with injuries in the workers’ compensation claims database than LOTE migrants in the Australian population. Whilst this is consistent with the literature demonstrating that workers with a lack of language proficiency are at a greater risk of work-related in-juries [22], underreporting of work-related injuries in the migrant worker population is well documented [20] and it is therefore possible that reported work-related injury rates in the current study underestimate the actual injury rates and may instead over-repre-sent the contribution of more severe injuries and outcomes. Indeed, we are confident that not all migrant workers with injuries will have accessed language interpreters and therefore our approach will further underestimate the number of migrant workers with injuries in the Victorian setting. Nevertheless, the population identified in our analysis can be used to begin to understand the association between English language proficiency and work-related injury outcomes and reiterates the challenges in studying this vulnerable population.
We observed a rightward shift in the age distribution of LOTE workers with injuries towards older age brackets, LOTE workers with injuries were overrepresented in the older age brackets with 40% of all LOTE injured workers in the 40–49-year-old age category. This is in contrast to the Australian migrant population labour force participation statistics which report a similar pattern of participation in each worker age between migrant and non-migrant workers [29]. This data suggests that either older migrant workers are at an increased risk of work-related injury or that younger migrant workers face additional barriers to reporting work-related injury. Both younger age and job inexperience have been reported in the literature to be factors associated with increased work-related injury vulnerability. Previous Australian research has shown that 49% of migrant workers report employment in a field non-commensurate with the skills, tra-ining or qualifications attained in their country of origin [31] suggesting that inexperience which increases the risk of work-related injury is possible. The findings of Reid together with the rightward shift in the age distribution of injured LOTE workers in our population highlights the need to address the mismatch between previous training/experience and current employment in LOTE workers. Whether barriers to reporting are age-specific in migrant workers remains unclear. Previous research by Akhavan et al. uncovered that female immigrants in Sweden over the age of 50 reported a lack of access to skills training programs compared to younger immigrants and native workers [32]. With this in mind, it is possible that the LOTE workers in our setting also experience access barriers to skills and training which may increase their risk of occupational injuries. In addition to a differential age distribution between LOTE and NLOTE workers with injuries, we observed that LOTE workers with injuries were more likely to be male compared to NLOTE workers with injuries which is consistent with previously published migrant worker statistics [33]. This data suggests that the gender distribution in our LOTE compensation dataset is representative of the Australian migrant population.
Next we explored the occupation profiles of LOTE and NLOTE workers with injuries. This analysis extends previously published Australian research examining work-related injury rates in migrant workers compared to non-migrant workers which was unable to assess occupation or the causes of work-related injuries in this population [3, 21]. We observed LOTE workers with injuries were over- represented in high risk industries such as manufacturing. This finding is consistent with data from the U.S.A which reveal that transportation and construction report both the highest rates of work-related fatalities and comprise the greatest proportion of migrant workers across all industries [34, 35]. In the UK however, migrant workers were reported to be employed mainly in health and social work, manufacturing, real estate, renting and business activity and wholesale, retail and motor trade sectors [36], hig-hlighting the differences in migrant populations in-ternationally and confirming the need for targeted local data analysis. We observed that the injury typeand cause of injury was significantly different bet-ween LOTE and NLOTE workers with injuries, with LOTE workers with injuries more likely to report injuries caused by sound and pressure, body stressing and mental stress than NLOTE workers with injuries. These findings are consistent with the differences in occupation observed. Dembe et al. reported that musculoskeletal injuries, cuts and bruises and other traumatic injuries were the leading injury types in workers with injuries aged 33–41, irrespective of migrant status (classified as being “foreign-born”) but dependent on family income, occupation and specific hazardous job activities [37]. This data suggests that the LOTE workers with injuries identified in our analysis may represent workers at an increased work-related injury risk due to other factors which are independent of language proficiency and that lack of language proficiency may in fact be a predictor of occupation, family income and/or hazardous job activities. Further qualitative research exploring work-related injury characteristics in large populations with similar occupation and income status between migrant and non-migrant workers is required to dissect the contribution of these factors to the work-related injury risk profile.
One of the most striking findings of our analysis was that after adjusting for employment status, occupation type and injury severity factors, LOTE workers with injuries were associated with poorer recovery outcomes (such as a greater number of days off work and greater total healthcare expenditure) compared to NLOTE workers with injuries. This finding may suggest that the barriers and challenges to return to work and recovery may be different between LOTE and NLOTE workers with injuries. The poorer recovery outcomes we identified in the LOTE workers with injuries irrespective of injury severity support the findings of Premji that workers who lack English language proficiency experience structural and interpersonal barriers to recovery services and benefits [38] and extend these findings by demonstrating the long-term consequences of these barriers. The study by Premji highlights the bias of systemic factors such as return to work policies and practices which fail to account for the issues faced by workers with lack of English proficiency and instead hinder recovery and return to work in this vulnerable population [38]. With this in mind, our analyses suggests that a review of the systemic factors with a view to addressing the language proficiency limitations has the greatest potential to significantly improve long-term recovery and return to work for LOTE workers with injuries.
We acknowledge that the LOTE workers with injuries identified in the present study likely underestimate the LOTE worker with injuries population in Victoria as we are unable to determine whether all injured workers who lack English language proficiency 1) accessed workers’ compensation or 2) accessed interpreter services. Nevertheless, we suggest that the LOTE workers with injuries identified in this study are a population which benefit from access to formal language support yet still experience poorer recovery outcomes suggesting that language support per se is not sufficient to improve LOTE worker with injuries recovery profiles to NLOTE worker with injuries levels.
Importance of English proficiency/language-specific safety messaging
Lack of English language proficiency is associated with an increased risk of work-related injury [20]. Whilst it is important to ensure safety and risk management that is not dependent on language or cultural biases, such as employing graphics to convey important safety messages [39], our analysis extends this research by demonstrating that workers with injuries who utilise an interpreter service have a poorer long-term recovery from work-related injury. Therefore, we suggest that English language proficiency is associated not only with the risk of work-related injury, but also with recovery outcomes. A number of potential reasons for the increased work-related injury risk in migrant workers have been proposed such as overrepresentation in high risk occupations and longer working hours. Certainly, our analysis revealed that LOTE workers with injuries were more likely to be employed in high risk occupations such as manufacturing.
In terms of work-related injury, one of the major research limitations in this field is the lack of visibility of vulnerable workers (including migrant workers) at a jurisdictional level and challenges in identifying and accessing these migrant workers for targeted research. In addition, appropriately defining and identifying vulnerable migrant workers is recognised as especially difficult [40–42]. What we do know is that migrant workers have been shown in Australia, Canada and Sweden to face the highest risk of work-related injury during the first 5 years foll-owing migration, after which time the risk of injury is similar to the native population [15, 43]. In the absence of information regarding migration status or time since arrival, we suggest that lack of English proficiency may be an appropriate surrogate marker to identify the most vulnerable migrant workers; both for research purposes and also for identifying injured workers who would benefit from access to additional resources to improve recovery and facilitate earlier return to work.
Conclusion
Lack of English language proficiency is associated with an increased risk of work-related injury [20]. Our analysis extends this research by demonstrating that injured workers who utilise an interpreter service have a poorer long-term recovery from work-related injury. Therefore, we present evidence that English language proficiency is associated not only with the risk of work-related injury but also to the long-term recovery outcomes. These findings suggest that despite access to language interpreter services, LOTE workers with injuries experience English language proficiency dependent, and injury severity independent, recovery barriers which need to be overcome to improve long term recovery outcomes.
Conflict of interest
The authors declare no conflicts of interest.