Article Text
PDF
Abstract
Objective To identify, appraise and synthesise studies that have examined the degree to which new workers are at an elevated risk of work-related acute injuries and musculoskeletal (MSK) injuries.
Method We searched three relevant electronic databases for studies published between 1995 and early 2018. Fifty-one studies using multivariate analyses met our relevance and quality appraisal criteria. These studies examined two different work outcomes: acute injuries (eg, cuts, burns and falls) and MSK injuries (eg, repetitive strain).
Results In four of six studies looking at acute work injuries, new workers were found to be at an elevated risk of injury (ie, moderate supportive evidence of new worker risk). In another six studies looking at MSK symptoms, injuries or disorders, evidence of an elevated risk among new workers was insufficient or limited.
Conclusions Our review has potential implications for the prevention of work injuries, providing policy-makers and workplace parties with supportive evidence about the importance of prevention efforts focused on new workers, such as developing workplace policies that emphasise hazard exposure reduction, hazard awareness, hazard protection and worker empowerment.
- epidemiology
- health and safety
- injury
Statistics from Altmetric.com
Key messages
What is already known about this subject?
Many studies find new workers at increased risk for occupational injury compared with more experienced workers. However, the studies vary in their definitions of a new worker and the types of injuries examined. Also, labour market shifts to short-tenure jobs increase the importance of understanding and mitigating new worker injury risks for occupational health and safety practitioners and policy-makers.
What are the new findings?
The systematic review found that four of six acute work injury studies showed new workers at an elevated risk. In contrast, musculoskeletal injury studies showed insufficient to limited evidence of new workers at an elevated risk.
How might this impact on policy or clinical practice in the foreseeable future?
The findings are consistent with instituting workplace policies that assist new workers with reduced hazard exposures, as well as increased hazard awareness and protections. Our results also underscore the importance of reducing job turnover, which would decrease the amount of time individuals spend as new workers over the course of their working lives. Given the potential for shorter work tenure to be related to increased risk of acute injury, greater thought on how best to operationalise job tenure in future studies would improve our understanding of the mechanisms leading to this elevated risk.
Background
Work-related acute injuries (eg, cuts, burns and falls) and musculoskeletal (MSK) injuries (eg, repetitive strain and lower back pain) continue to be a substantial source of lost work days and disability around the world.1 Many studies have found that inexperience or being new on the job is a risk factor for work injuries.2–4 Furthermore, temporary employment has become more common in developed economies.5 Changing firms and jobs more often means that people are spending proportionally more of their working lives as new workers.6 Finally, although work injury rates have been decreasing in many jurisdictions for several years,7 8 the gap between the elevated injury rate for new workers and their more experienced counterparts has remained.6
New worker injury vulnerability has been attributed to lack of familiarity with job tasks, inability to handle unexpected events, difficulty accurately appraising hazards, increased exposures to hazardous conditions and lower awareness of workplace safety policies and protections.9–13 These injury mechanisms lead to a hypothesis that an elevated risk of injury among new workers would be more strongly associated with acute work injuries such as cuts, burns and falls than it would be for MSK injuries, which are often the result of long-term, cumulative exposures.
A complication is the diversity of ways to define a ‘new worker’. The most common definition is the length of time a person has been employed at a firm (ie, could include multiple jobs within the firm). However, one also finds new workers defined by time at a particular job or time in an industry (eg, construction). To allow us to describe the heterogeneity of new worker definitions, the literature search included all three types of tenure. However, our evidence synthesis excluded studies defining new workers using industry tenure because the job tasks and working conditions across workplaces would be too varied.
Accordingly, the primary objective of this review is to identify, appraise and synthesise studies that have examined the degree to which new workers are at an elevated risk of work-related acute injuries and MSK injuries. A secondary objective was to describe and discuss more generally the heterogeneity of types of work tenure measures (ie, time in job, firm or industry tenure) and the diversity of thresholds for defining ‘new worker’ (eg, ≤12 months).
Methods
Systematic review steps
This systematic review included the following steps: developing the review question, conducting a literature search, screening articles for their relevance, appraising the quality of the relevant articles, extracting data from the articles that met quality criteria and synthesising the evidence.14 15 The review team consisted of six researchers with expertise in the fields of occupational health and safety, epidemiology, psychology, ergonomics, health sciences, information science and systematic review methodology.
Literature search
Search terms were developed for three domains of the PICO framework: population terms for workers; exposure terms for job, firm or industry tenure; and outcome terms related to the work injury.
The following databases were searched for peer-reviewed articles in English: MEDLINE (OVID), Embase (OVID) and Health and Safety Abstracts (see the online supplementary appendix A). The initial search was conducted for articles published between 1995 and 2015 and later updated in January 2018. We also searched the reference sections of the articles in table 1 for additional relevant studies.
Supplementary file 1
Summary of study characteristics
Relevance screen
Articles were considered relevant if they included the following:
a population sample engaged in paid work.
a measure of the length of time working at a particular job, firm or industry.
a measure of a work-related injury.
a quantitative study design.
a multivariable analysis (or other method for controlling for confounding factors) that examined the association between tenure measures and work injury.
The detailed inclusion and exclusion criteria are summarised in the online supplementary appendix B. The screening tool was pilot-tested to ensure consistency in applying the criteria. Both the titles and abstracts and the full-text articles were screened by two rotating pairs of reviewers. For all the review stages below, disagreements were resolved between the reviewers, and when consensus could not be reached, pairs consulted with a third reviewer for resolution. When necessary, the review team attempted to contact a study’s primary author for further information.
Quality appraisal
We used the Quality in Prognosis Studies (QUIPS) tool16 to assess quality in this review because work tenure as a predictor variable is functionally similar to a prognostic factor. Also, the QUIPS tool has been specifically designed for assessing the method issues relevant to observational studies, while many systematic review quality appraisal tools are designed for intervention studies. Further, the QUIPS tool is comprehensive in that it evaluates six methodological issues: study participation, study attrition, predictive factor measurement, outcome measurement, confounding measurement, and analysis/reporting (for details about QUIPS dimensions and study quality algorithm, see the online supplementary appendix C). The team pilot-tested this tool on two sample articles and resolved differences to ensure a consistent application among team members.
Using the QUIPS tool, all relevant articles were evaluated by the lead author, with the other team members serving as second reviewers. For each of the six dimensions of the QUIPS, reviewers assigned a risk of bias (ie, low, moderate or high) for that methodological issue (for QUIPS QA ratings for each study, see the online supplementary appendix D).
The following guidelines were used to define study quality:
High quality: all dimensions rated low bias OR one-dimensional rated moderate bias.
Medium quality: one-dimensional rated high bias OR two-dimensional to three-dimensional rated moderate bias OR a combination of one-dimensional rated high bias and one-dimensional rated moderate bias.
Poor quality: two-dimensional or more rated high bias OR four-dimensional or more rated moderate bias (ie, a majority of dimensions showing moderate bias) OR a combination of one-dimensional rated high bias and at least two-dimensional rated moderate bias.
Those studies rated as poor quality did not move to the data extraction phase.
Data extraction
The goal of the data extraction phase was to collect information on the key characteristics of the study, sample, predictive factor (ie, how work tenure was operationalised), covariates, relevant multivariate analyses of new worker variable and work-related injury outcome. The data extraction guide was developed and pilot-tested on two articles to ensure reviewers’ consistency. Data extraction was performed independently by a primary reviewer who had a background in research methods (FCB, PS, ETM, ML). Three secondary reviewers checked and confirmed the data extracted by the primary reviewer (FCB, JD, QM).
Evidence synthesis
We employed an evidence synthesis approach that uses an algorithm for categorising the level of support from ‘strong’ to ‘insufficient’ and, accordingly, for providing guidance to Occupational Health and Safety (OHS) practitioners.14 17 The data extracted from the relevant studies were combined and organised to determine the quantity, quality and consistency of evidence on the association between tenure measures and work injury (for evidence synthesis algorithm, see the online supplementary appendix E). For example, if a set of studies agree consistently enough to meet criteria for a ‘strong’ level of evidence, then that would indicate more confidence in the finding and in potential recommendations and implications of the finding.
Our initial relevance criteria led to a very heterogeneous set of studies with regard to new worker definitions and work injury outcomes. To provide a meaningful evidence synthesis on a subset of more homogeneous studies, we included in our synthesis only studies that used a measure of job tenure or firm tenure (ie, studies using industry tenure were excluded from further analyses). In addition, we only included studies that operationally defined new workers as ≤12 months in the job/firm. If the tenure measure was continuous, the study was not included in the final evidence synthesis because it did not provide a clear demarcation for which part of the sample was considered a new worker and continuous variables assume a linear relationship between the predictor and outcome.
To define elevated injury risk for each study, we focused on the ORs, relative risk and CIs from multivariable logistic regressions, where elevated injury risk meant that the newest workers showed a statistically significant, higher likelihood of injury compared with the reference group in the study.
If a study included an overall multivariable analysis as well as stratified analyses by specific subgroups (eg, gender, occupation and subcategories of acute injuries), then the overall analysis was used to determine whether it was supportive of the notion that new workers were at elevated risk for injury or not. When no overall analysis of the association between tenure and work injury was included and the majority of stratified analyses were supportive, then the study was counted as supportive.
To reflect the different mechanisms potentially associated with MSK outcomes, we subdivided the MSK evidence synthesis into those studies examining MSK symptoms, those examining MSK work injuries and those examining MSK disorders (ie, included clinical examination).
We also identified and sought to summarise the studies that conducted formal statistical tests of the moderating effect of demographic (eg, gender) and work-related (eg, industry) variables. We identified only three studies that included formal statistical tests of the moderators of the new worker–injury association4 18 19 in the Reference section. However, the studies assessed different moderators (eg, gender and industry sector), so no evidence synthesis was conducted on this issue.
Results
Literature search
The literature search until January 2018 identified a total of 3521 references after the results from the different databases were combined and duplicates were removed (see figure 1 for Preferred Reporting Items for Systematic Reviews and Meta-Analyses). A total of 418 references met the title and abstract inclusion criteria and were retrieved for full-text review. Of these, 128 studies met the relevance criteria and moved to the quality appraisal phase. Data were extracted and synthesised from 51 studies of high and medium quality (see the online supplementary appendix F for study references).
PRISMA flow Diagram. MSK, musculoskeletal.
Of the 51 articles, 24 were rated as high quality and 27 were rated as medium quality. Studies originated from several geographic regions: Europe (n=10); USA (n=20); Canada (n=4); South America (n=1); Africa (n=1); Middle East (n=7); Asia (n=7) and Australia/Pacific (n=1).
Study design characteristics
Cohort and longitudinal designs are generally considered to have fewer threats to validity than cross-sectional studies.20 The 51 studies included 18 retrospective cohorts, 6 longitudinal designs (ie, prospective cohorts), 1 case–control study and 26 cross-sectional studies (see table 1 for a summary of the study characteristics).
Measurement of tenure and work injury
Information about the duration of job, firm or industry tenure and work injury was collected through self-report in most of the studies (30 out of 51). Other studies obtained information on tenure duration and work injury through administrative firm records or workers’ compensation authority data (21 out of 51).
Several studies used a continuous measure of tenure in their analysis of the association between tenure and work injury (n=10). When studies used work tenure categories, new workers were defined as working from as little as 1 month since the date of hire to <258 months (21.5 years) of tenure. Only 26 studies of the 41 studies using work tenure categories specified new workers as those having worked ≤1 year.
Across the 51 studies, 20 studies used compensation claims or firm records to define the presence of a work injury. Twelve studies used a standardised self-report measure such as the Nordic Musculoskeletal Questionnaire to define their outcomes.21 Another 14 studies used other self-report measures with no established reliability or validity evidence to define their outcome. Four studies focused only on MSK injuries and included a clinical examination to confirm the presence of a MSK disorder. One study used both a standardised self-report measure and a clinical examination to confirm the outcome measure.
Covariates examined in tenure-work injury studies
Many different types of covariates were included across the studies examined. Included in tables 1–3 is a column that shows whether the study controlled for demographic, work-related or other (eg, geographic and temporal) factors. The most common type of covariate was a demographic covariate such as age or gender. Almost all studies also included some kind of work-related covariate (eg, type of occupation) or some kind of workplace exposure (eg, work schedule).
Studies examining acute injury outcomes where newest worker group is ≤12 months at the job or firm
Studies examining MSK-related outcomes where newest worker group is ≤12 months at the job or firm
Evidence synthesis
Acute work injury synthesis
Six multivariate studies examined acute work injury outcomes where the newest worker group was defined as ≤12 months at the job or firm (see table 2). Of the six studies, four showed a significantly elevated risk for new workers. According to the systematic review evidence synthesis guidelines, having two-thirds or more medium-quality and high-quality studies agree constitutes moderate evidence that new workers are at elevated risk for an acute work injury.
Work-related MSK outcomes synthesis
Two multivariate studies examined work-related MSK symptoms where the newest worker group was defined as ≤12 months at the job or firm (see table 3). Of the two studies, only one showed a significantly elevated risk for new workers. According to the systematic review evidence synthesis guidelines, having findings from two high-quality studies that are contradictory constitutes mixed evidence that new workers are at elevated risk for work-related MSK symptoms.
Two multivariate studies examined MSK work injuries from compensation claims where the newest worker group was defined as ≤12 months at the job or firm (see table 3). Both the medium-quality and high-quality studies showed a significantly elevated risk for new workers. According to the systematic review evidence synthesis guidelines, having a medium-quality and a high-quality study agree constitutes limited evidence that new workers are at elevated risk for work-related MSK injuries.
Two multivariate studies examined work-related MSK disorders where the newest worker group was defined as ≤12 months at the job or firm (see table 3). Of the two studies, neither showed a significantly elevated risk for new workers. According to the systematic review evidence synthesis guidelines, having findings from two high-quality studies that are not supportive constitutes insufficient evidence that new workers are at elevated risk for work-related MSK disorders.
Discussion
This is the first systematic review to synthesise the existing evidence on the degree to which new workers are at an elevated risk of work-related acute injuries and MSK injuries. The present review performed an evidence synthesis on six acute work injury studies and six MSK outcome studies that were sufficiently homogeneous with regard to their definition of a new worker and their type of injury. Overall, we found moderate evidence that new workers are at an elevated risk of acute work injuries and limited to insufficient evidence that new workers are at an elevated risk of work-related MSK outcomes. There were too few formal statistical examinations of moderators of new worker–injury association to perform an evidence synthesis.
Interpretation of results
In seeking to explain why new workers would be at elevated risk for acute work injuries, the mechanisms could be broadly classified as new workers having differential exposure or differential responses compared with their more experienced counterparts. Differential exposure refers to the possibility that even when they have the same job title, new workers may be exposed to more hazardous tasks and working conditions than more experienced workers.22 Indirect evidence of this possible mechanism is that newer, younger workers report encountering many hazardous working conditions.13 23 24
Differential responses refer to the possibility that lack of familiarity with OHS procedures (eg, inadequate OHS training), greater hesitation to speak up about hazards or lack of skill in implementing OHS procedures (eg, inadequate coping response to unexpected events) can indirectly increase hazard exposure, even when new workers are assigned similar work tasks in similar working conditions.25 For physically demanding work, new workers might be more susceptible to early injuries due to physiological factors such as lack of strength, insufficient psychomotor coordination or lack of physical resilience.
It is difficult to interpret the limited to insufficient evidence that new workers are at elevated risk of MSK outcomes. For example, studies of MSK injuries were based on workers’ compensation claims, which typically include a mix of short-latency MSK injury mechanisms (eg, sprains and strains) as well as more cumulative, long-term MSK injury mechanisms (eg, repetitive strain injuries). Consistent with the notion that new workers are not at immediate risk for more cumulative, long-term MSK injuries, no studies that included a clinical examination to diagnose MSK disorders found new workers at elevated risk. However, this does not imply that reducing more cumulative long-term MSK hazards among new workers should be ignored. While MSK disorders do not manifest quickly, early exposure to MSK hazards can still contribute to overall exposure and certain vulnerable populations (eg, young workers) may be particularly affected by early exposure to MSK hazards.
Strengths and limitations
Strengths of this systematic review included using state-of-the-art methods, specifically an established quality appraisal tool and transparent evidence synthesis procedures.14 16 The strength of evidence synthesis was enhanced by only including medium-quality and high-quality studies. Also, all studies included in the review conducted multivariate analyses including many covariates, reducing the risk of confounding affecting results.
Limitations of a systematic review include the risk of publication bias. We attempted to reduce the risk of publication bias by including studies where work tenure was a covariate (ie, not the focal predictor) and by contacting authors to request additional information when tenure information in multivariate analysis was missing. Also, while a strength of the evidence synthesis approach is that it was developed in collaboration with Ontario-based OHS stakeholders (eg, employers and OHS practitioners), the generalisability of this approach could depend on the needs of knowledge users and legislation in other jurisdictions.
Notable limitations are present in the methodologies of the included studies. Common limitations include the potential biases of cross-sectional designs (eg, direction of causality), longitudinal studies (eg, attrition bias) and unmeasured covariates (eg, foreign-born workers), which may lead to studies producing biased results on the new worker–injury risk relationship. Another limitation is that those studies based on workers’ compensation claims can be biased due to under-reporting. In addition, substantial heterogeneity was present across the included studies regarding type of tenure (ie, job, firm and industry tenure) and how ‘new worker’ was defined.
A more general methodological issue for observational studies is the healthy worker effect (HWE). The HWE is a type of selection bias related to the tendency of workers with high biological and behavioural resilience to continue working at the same job, whereas less resilient workers who experience an injury or health event may leave their jobs.26 Thus, an alternative explanation for the elevated risk among new workers is that the HWE is an unobserved third variable (eg, resilience) that confounds the association between work tenure and the risk of acute injury. However, given the rise of part-time and temporary work, the HWE may have lessened recently because of proportionally more non-injury-related reasons that workers have for leaving jobs early (eg, contract terminated) than in the past. The possibility that the HWE is at play underscores the need for longitudinal designs and for assessing time-varying job exposures in more detail.26
Conclusions
It is important not to overstate the implications of the preliminary findings of this systematic review. If the new worker–acute injury finding is confirmed, prevention efforts such as developing workplace policies that emphasise reduced hazard exposure, greater hazard awareness and greater worker empowerment for new workers may help reduce some of the potential mechanisms underlying the acute injury risk for new workers.11 Our results also underscore the importance of reducing job turnover, which would decrease the amount of time individuals spend as new workers over the course of their working lives. This issue is particularly important in industries with moderate-risk to high-risk hazards where the level of temporary employment is increasing.
One recommendation for improving OHS research on this topic is to provide a more detailed description of the work tenure measure, and when using a self-report measure, to establish its reliability and validity. Also, any threshold or cut-off for operationally defining ‘new worker’ should be a theoretically informed decision regarding, for example, when job tasks can be reasonably mastered, and not solely due to statistical considerations (eg, dichotomous splitting of the sample). Further, studies that assess some type of work tenure should start including measures with greater specificity in terms of job-specific hazard exposures over time.
Future studies should also explicitly provide analyses stratified by key moderating variables such as gender or industry and should conduct formal statistical tests to examine whether the moderation effect meets some level of significance, when appropriate. Also, to ensure accuracy in terms of documenting when the injury occurred during a worker’s tenure, it is essential that future studies employ longitudinal designs and assess for time-varying job exposures in more detail. Following these recommendations in future studies would aid in turning work tenure from a general variable reflexively included in analyses that maybe a proxy for many other factors to measures that help elucidate the specific reasons new workers are at elevated risk.
Acknowledgments
The authors would like to acknowledge Emma Irvin, Kim Cullen, Joanna Liu and Sara Morassaei for their assistance with this systematic review. Their help was greatly appreciated.
References
Footnotes
Contributors FCB, PMS and ML: developed the research question. QM and JD: completed the literature search. FCB, JD and QM: carried out the evidence synthesis, with feedback from PMS, ML and ETM. FCB, ETM and QM: wrote the manuscript in consultation with PMS, ML and JD. FCB: responsible for the overall content of the review. All authors were involved in the relevance screening, quality appraisal and data extraction steps.
Funding This research was supported by internal funding from the Institute for Work and Health, Toronto, Ontario, Canada.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Patient consent for publication Not required.