Article Text

other Versions

Download PDFPDF
Original research
Predictors of subsequent injury at work: findings from a prospective cohort of injured workers in New Zealand
  1. Helen Harcombe1,
  2. Ari Samaranayaka2,
  3. Emma H Wyeth3,
  4. Gabrielle Davie1,
  5. Ian D Cameron4,
  6. Rebbecca Lilley1,
  7. Sarah Derrett1
  1. 1Injury Prevention Research Unit, Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand
  2. 2Centre for Biostatistics, Division of Health Sciences, University of Otago, Dunedin, Otago, New Zealand
  3. 3Ngāi Tahu Māori Health Research Unit, Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand
  4. 4John Walsh Centre for Rehabilitation Research, Sydney Medical School Northern Faculty of Medicine and Health, Sydney, New South Wales, Australia
  1. Correspondence to Dr Helen Harcombe, Injury Prevention Research Unit, Department of Preventive and Social Medicine, University of Otago, Dunedin 9054, New Zealand; helen.harcombe{at}otago.ac.nz

Abstract

Objectives People who have experienced a work-related injury can experience further work injuries over time. This study examines predictors of subsequent work-related injuries over 24 months among a cohort of injured workers.

Methods Participants were those recruited to the earlier Prospective Outcomes of Injury Study (POIS) who had a work-related injury (the ‘sentinel’ injury). Data from POIS participant interviews were combined with administrative data from the Accident Compensation Corporation (New Zealand’s no-fault universal injury insurer) and hospital discharge data. Modified Poisson regression modelling was used to examine whether presentinel injury sociodemographic and health, sentinel injury or presentinel injury work-related factors predicted subsequent work-related injuries.

Results Over a third of participants (37%) had at least one subsequent work-related injury in 24 months. Factors associated with an increased risk of work-related subsequent injury included being in a job involving carrying or moving heavy loads more than half the time compared with those in jobs that never involved such tasks (RR 1.42, 95% CI 1.01 to 2.01), having an inadequate household income compared with those with an adequate household income (RR 1.33, 95% CI 1.02 1.74) and being aged 50–64 years compared with those aged 30–49 years (RR 1.25, 95% 1.00 to 1.57).

Conclusion Subsequent work-related injuries occur frequently, and presenting with a work-related injury indicates a potentially important intervention point for subsequent injury prevention. While the strength of associations were not strong, factors identified in this study that showed an increased risk of subsequent work-related injuries may provide a useful focus for injury prevention or rehabilitation attention.

  • epidemiology
  • injury
  • longitudinal studies

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

What is already known about this subject?

  • Work-related injuries have large social and economic costs.

  • Following a work-related injury, some workers go on to have further work-related injuries.

  • Presenting to healthcare professionals with a work-related injury is a potentially important intervention point for subsequent injury prevention.

What are the new findings?

  • A substantial proportion (37%) of workers had at least one subsequent work-related injury in the 24 months following a previous work-related injury.

  • There was an increased risk of work-related subsequent injury for those in a job involving carrying or moving heavy loads more than half the time compared with those in jobs that never involved such tasks.

  • Those with an inadequate household income were more likely to have a subsequent work-related injury, as were those aged 50–64 years compared with those aged 30–49 years.

How might this impact on policy or clinical practice in the foreseeable future?

  • Factors that showed an increased risk of subsequent work-related injuries identified in this study may provide a useful focus for injury prevention and/or rehabilitation interventions.

Background

‘Good work’ is generally considered beneficial for health and well-being, providing opportunities for productive social and work participation, economic stability and positive interpersonal interactions1 2 It is important, however, to recognise that not all work is beneficial to health and well-being. When safety is compromised at work, injuries related to that work can and do occur. In New Zealand, a country of approximately 5 million people, the Accident Compensation Corporation (ACC), the country’s universal no-fault injury insurer, had 238 800 work-related injury claims in 2018.3 Work-related injury plays a major role in reducing work and social participation with the social and economic costs of workplace injury and disease burden estimated at $NZ3.5 billion (or 2% of gross domestic product) per annum.4 Costs of injury are not only financial and fall on the worker, their families, communities and wider society.5

People can incur more than one work-related injury over time. For example, 37% of work injury claimants in a study of workers’ compensation data in Victoria, Australia, filed a second claim in the following 5 years,6 and 49% of claimants in a study of Workers Compensation Board claimants in Alberta, Canada, had at least two claims in a 10-year period.7 Understanding the risk factors associated with subsequent work-related claims, and then providing interventions to prevent subsequent injuries from occurring, provides an opportunity to help address this substantial burden. To date, research examining subsequent injuries in a work context has largely drawn on workers’ compensation claims data, examining the risk factors associated with further work-related claims.7–11 Other studies, such as that by Keeney et al,12 have focused on a single injury type such as back injuries. A study by Galizzi13 used linked self-reported panel survey data of workers aged 14–22 years in 1979 and workers compensation claims to examine subsequent occupation injuries sustained in the period 1988 to 2000. This study was not limited to a specific injury type and found that low education, short job tenure, occupation and job demands were associated with recurrent injury.13 The Prospective Outcome of Injury Study (POIS), with comprehensive measures of sociodemographic, injury and workplace characteristics and follow-up to 24 months postinjury has already provided insights into the social, injury and vocational predictors of return to work following injury in New Zealand.14–16 The current examination of work injuries in the Subsequent Injury Study (SInS) expands on the POIS interview data and includes compensation claims’ data for subsequent work-related injuries in the 2 years following the injury for which participants were recruited (known as the ‘sentinel’ injury). The aim of the analyses reported in this paper is to investigate whether particular presentinel injury sociodemographic and health, sentinel injury or presentinel injury work-related factors are predictors of subsequent work-related injuries in the 24 months following an ‘initial’ (sentinel) work-related injury.

Methods

The POIS recruited participants soon after an injury event (referred to as the ‘sentinel’ injury event for the purposes of this paper) involving a claim to ACC between 2007 and 2009.17 To be eligible for POIS, the injury event needed to involve an entitlement claim with ACC. Entitlement claims indicate that the injury was considered by ACC as likely to require financial compensation for at least a week off work or some other sort of rehabilitative assistance (eg, home help and travel assistance). Entitlement claims contrast with medical treatment only claims where ACC does not consider the injury likely to require financial compensation or the need for rehabilitative supports.18 POIS recruited a cohort of injured people of working age, which included all injury diagnoses occurring in all settings (eg, road, work and recreation). Sensitive claims, such as sexual assaults, were excluded.17 In the SInS,19 data from POIS participant interviews were combined with ACC administrative data for the sentinel and any subsequent injury ACC claims over the following 24 months and, for injuries involving hospitalisation, hospital discharge data from the National Minimum Dataset. This paper focuses on subsequent work-related injury claims for the subset of participants recruited to POIS following a work-related sentinel injury event and examines whether there are particular sociodemographic, sentinel injury or work-related factors that predict subsequent work-related injuries.

Outcome variable

Work-related subsequent injuries were identified using ACC’s ‘at work’ indicator (a binary indicator of whether the subsequent injury was work related or not). All subsequent work-related ACC claims were included (eg, entitlement claims, medical fees only and sensitive claims).

Potential explanatory variables

A number of the potential explanatory variables in this paper have been described in detail previously.20 21 Presentinel injury sociodemographic and health variables were from POIS interviews undertaken (on average) 3 months after the sentinel injury event and included: age at the time of the sentinel injury event, sex, ethnicity, education, adequacy of household income and chronic conditions. Adequacy of household income for everyday needs was assessed using a single question with responses ‘more than enough’, ‘enough’ and ‘just enough’ classified as ‘adequate’, with ‘not enough’ classified as ‘not adequate’.22 Participants were asked about the presence of 21 specific chronic health conditions diagnosed by a doctor and that had lasted, or were expected to last, at least 6 months23 with these classified as ‘no chronic conditions’, ‘one chronic condition’ or ‘≥2 chronic conditions’.24

Sentinel injury characteristics were obtained from the 3-month POIS interview data, ACC data and hospital discharge data and included: injury type (combination of nature and body region), injury severity, hospitalisation and trouble accessing healthcare services. Twelve binary (‘yes’ or ‘no’) variables were used to capture injury type as each injury event could involve more than one type of injury, for example, in a fall, someone may sustain an upper extremity fracture and an intracranial injury. Sentinel injury severity was derived from New Injury Severity Scores (NISS)25 categorised as NISS 1–3 (lower threat to life severity), 4–6 and >6. Hospitalisation was defined as being admitted to hospital or treated at an emergency department for ≥3 hours within 7 days of the injury event. A single question asked participants if they had ‘trouble getting to or contacting health services’ for their sentinel injury with response options ‘yes’ and ‘mixed’ combined for analyses.26

Presentinel injury work-related factors were also derived from the 3-month POIS interviews and included physical, psychosocial and organisational factors, many of which have been described previously.15 Occupation27 was coded to the New Zealand Classification of Occupation (NZSCO)28 and classified as ‘professional’ (NZSCO major level 1–3), ‘technical’ (level 4–5), ‘trade or manual’ (level 6–9) or unclassified. Physical work factors were: ‘working in painful or tiring body positions’, ‘carrying or moving heavy loads’, ‘repetitive hand or arm movements’, ‘standing or walking’ and ‘physical exertion more than walking, standing and moving in a normal way’.15 29 Participants were asked whether their work involved each of these physical factors (yes/no). Those who answered ‘yes’ to any of these activities were then asked how much of the time they did that task with possible responses: ‘all to ¾ of the time’, ‘about ½ to ¼ of the time’ or ‘occasionally or sometimes’.15 For analyses these were categorised as ‘more than half the time’, ‘up to half the time’ and ‘never’16

Psychosocial work factors were: job strain, social support, job security and job satisfaction. Job demands, job control and social support were assessed using the Whitehall II psychosocial work questions (adapted from the Job Content Questionnaire).30 31 Job strain was determined by combining job demands and job control, with the combination of low control and high demands classified as ‘high job strain’, low control and low demands ‘passive job strain’, high control and high demands ‘active job strain’ and high control and low demands classified as ‘low job strain’.32 Social support was categorised as quartiles for analyses.15 16 Job security was assessed by asking participants how secure they felt in their main job with responses categorised as ‘insecure’ and ‘secure’. Job satisfaction was assessed using a single question with responses collapsed into three categories for analyses: ‘satisfied’ (‘completely satisfied’ and ‘mostly satisfied’), ‘neither satisfied nor dissatisfied’ and ‘dissatisfied’ (‘mostly dissatisfied’ and ‘completely dissatisfied’).29

Work organisational factors were: contract type, hours of work and days worked per week. Contract type was categorised as: ‘permanent’, ‘temporary/casual/fixed term’, ‘self-employed’, ‘employer’ and ‘other/no formal contract’. Hours of work were categorised as ≤30, 31–45 and ≥46 hours per week.15 Days worked per week were classified as 1–5 or 6–7 days.15 Where participants had more than one job, work-related aspects were assessed for the participants’ ‘main job’; however, we also asked if participants worked multiple jobs.29

Analyses

Analyses have been restricted to those in POIS who had a work-related sentinel injury and who reported they had returned to paid work in some capacity at either the 3, 12 or 24 month POIS interview. Modified Poisson regression modelling33 was used to estimate relative risks and 95% CIs. Each potential explanatory variable was initially examined in univariable models. As well as age, sex and occupation, potential explanatory variables with a p value ≤0.3 in univariable analyses were then included in a stepwise backwards elimination to determine a final multivariable model of variables with a p value <0.1. Analyses were carried out using Stata V.15.1.34

Results

Of the 2856 POIS participants, 754 had a work-related sentinel injury. Of these, 650 reported they had returned to part-time or full-time employment in at least one of the follow-up interviews and were therefore deemed to be ‘at risk’ of a subsequent work-related injury and included in these analyses.

Of the cohort of interest, 243 (37%) had at least one work-related subsequent injury in the 24 months following their sentinel work-related injury. The majority of participants had one (n=85, 35%) or two (n=79, 33%) work-related subsequent injury claims in this period. However, the number of work-related subsequent injury claims per person ranged from 1 to 10 (with a median of two claims). Among the 243 people with work-related subsequent injuries, 78 had entitlement claims and 165 had other (non-entitlement) claim types. Participants were aged between 18 and 64 years (mean age 43.6 years) at the time of their sentinel injury and 74% were male. Table 1 describes the presentinel sociodemographic and health characteristics for those who had, and those who did not have, at least one subsequent work-related injury claim in the 24 months following their sentinel injury event. Univariable analyses found that women were less likely than men to have at least one subsequent work-related injury in the 24 months following their sentinel work-related injury (RR 0.74, 95% CI 0.57 to 0.95), those of Pacific ethnicity were more likely to have subsequent work-related injuries compared with those of New Zealand European ethnicity (RR 1.41, 95% CI 1.01 to 1.96) and those with an inadequate household income were more likely to have subsequent work-related injuries compared with those with an adequate household income (RR 1.40, 95% CI 1.08 to 1.82). There was no evidence to suggest that educational level, or number of chronic conditions, predicted subsequent injury.

Table 1

Univariable associations between presentinel injury sociodemographic and health characteristics and work-related subsequent injury

Sentinel work-related injuries included a wide range of injury types including intracranial injuries, fractures and dislocations/sprains/strains. Table 2 describes sentinel injury characteristics for those who had, and those who did not have, at least one subsequent work-related injury claim in the 24 months following their sentinel injury event. Those with a lower extremity fracture were less likely to have a subsequent work-related injury compared with those who did not have this type of sentinel injury (RR 0.63, 95% CI 0.41 to 0.97).

Table 2

Univariable associations between sentinel injury characteristics and work-related subsequent injury

Table 3 describes presentinel injury work-related factors for those who had, and those who did not have, at least one subsequent work-related injury claim. In univariable analyses, trade/manual workers were more likely to experience at least one subsequent work-related injury compared with professionals (RR 1.44, 95% CI 1.07 to 1.93) as were those in jobs that involved physical exertion more than half the time compared with those in jobs that never involved physical exertion (RR 1.34, 95% CI 1.02 to 1.76). (table 3). There was no evidence that job strain predicted subsequent injury.

Table 3

Univariable associations between presentinel-injury work-related factors and work-related subsequent injury

Table 4 describes the findings from the final multivariable model assessing predictors of having at least one subsequent work-related injury claim in the 24 months following the sentinel work-related injury event. Those aged 50–64 years were 1.25 times as likely to have a subsequent work-related injury compared with those aged 30–49 years (95% CI 1.00 to 1.57). Trade/manual workers were 1.34 times as likely to have a subsequent work-related injury compared with professionals (95% CI 0.98 to 1.82). Participants of Pacific or ‘Other’ ethnicities were more likely to have a subsequent work-related injury compared with those of New Zealand European ethnicity (RR 1.39, 95% CI 1.00 to 1.95, and RR 1.27, 95% CI 0.98 to 1.64, respectively). Workers with an inadequate household income were 1.33 times as likely to have a subsequent work-related injury compared with those with an adequate household income (95% CI 1.02 to 1.74). Workers whose sentinel injury involved a lower extremity fracture were less likely to have a subsequent work-related injury in the subsequent 24 months (RR 0.67, 95% CI 0.44 to 1.02). Those whose job involved carrying or moving heavy loads more than half the time were 1.42 times as likely to have a subsequent work-related injury compared with those whose job never involved carrying or moving heavy loads (95% CI 1.01 to 2.01).

Table 4

Final multivariable model: associations between presentinel injury work-related factors and work-related subsequent injury*

Discussion

A range of variables including sociodemographic, sentinel injury-related and work-related factors were predictive of subsequent work-related injury. Factors associated with an increased risk of work-related subsequent injury that had the greatest strength of association were: being in a job that involved carrying or moving heavy loads more than half the time, being of Pacific ethnicity, having an inadequate household income and being aged 50–64 years (compared with those aged 30–49 years). However, when accounting for all the other factors in the final multivariable model, these relationships were not strong.

Carrying/moving heavy loads is a known risk factor for injury.35 However, our finding that those whose jobs involve these types of tasks are at increased risk of subsequent work-related injuries may mean that if these tasks are able to be modified in the workplace, then this potentially may reduce work-related subsequent injuries. Those of Pacific ethnicity were more likely to have a subsequent work injury compared with those of sole New Zealand European ethnicity; however, it is unclear why this might be.

Participants with an inadequate household income to meet ‘everyday needs’ were on average 1.33 times as likely to have a subsequent work-related injury compared with those who reported that their household income was adequate. In this study, all participants’ sentinel injuries involved an ACC entitlement claim, which means that if people required a week or more off work then ACC would cover up to 80% of people’s weekly income.36 However, for those with an ‘inadequate household income’, 80% of their income would be even less sufficient to meet their everyday needs. Although we can only speculate, this may result in people going back to work, or to full work duties/hours, earlier than they might otherwise (ie, before they have fully recovered or completed their planned rehabilitation or treatment from their sentinel work-related injury), potentially placing themselves at risk of a subsequent work injury. In addition, in New Zealand, although ACC covers the majority of treatment costs, often there are copayments required from those who are injured for treatments such as physiotherapy or general practitioner visits. An inability to afford copayments may potentially mean that those with an inadequate household income might not complete their full rehabilitation, which may then again potentially put them at increased risk of injury at work. For injured workers with less adequate incomes, and receiving only 80% weekly income replacement, their ability to afford other activities or treatments (eg, access to swimming pools or other exercise facilities) that may help facilitate rehabilitation may also be affected.

Those aged 50–64 years were 1.25 times as likely to have a subsequent work injury compared with those aged 30–49 years. This finding is in contrast to several earlier studies among workers compensation claimants in Australia8 and Canada7 that found second work injury claims were less likely among those of older age groups. These studies attributed this to older workers being exposed to potential risk for a shorter time due to retirement. However, direct comparison with our findings is limited as in the current cohort only those workers who reported being in employment in at least one follow-up interview were included in the analysis, and our study was only over a 24-month period. Given that New Zealand has a rapidly ageing workforce and an increasing level of work participation after the age of 55 years,37 this suggests a need for further investigation. It may however indicate that more attention should be paid to supporting those older workers injured in the workplace who are returning to work.

This study had a number of strengths including having a population-based sample across all industries, a prospective design, the inclusion of several sources of baseline data including information from the injured people themselves and baseline variables reflecting multiple domains of interest. A further strength is that we have complete data on all work-related subsequent injuries involving an ACC claim, and ACC is a no-fault universal compensation system. There were however some limitations of the study. We do not know the specific date of return to work for participants, or whether they returned to the same hours or tasks at that point, as participants’ work status was collected via self-report at the time of the POIS interviews (on average 3, 12 and 24 months following the sentinel injury event). Those who reported not working at all of these interviews were excluded from these analyses on the basis that they were deemed to be not working and therefore not exposed to work hazards. However, given that work status was measured at set time points in a time continuum and we do not know about work status for the periods of time between the interviews, it is possible that some misclassification could have occurred. In addition, people may have been working for varying lengths of time over the 24 months of follow-up and therefore the time at risk of a work-related subsequent injury is unlikely to be consistent among all participants. We also do not know if the work-related factors considered, or some of the sociodemographic and health characteristics, changed over time, for example, we do not know about work accommodations that may have been made to allow an effective return to work. In addition, for those with a work-related subsequent injury in the 3-month period between their sentinel injury event and their first interview, having a subsequent injury could potentially have influenced their responses about work-related factors in their interview. However, there were only 38 participants who had an SI in this time period and any effect is likely to be small. Finally, duration of job tenure at the time of the work-related sentinel injury was not available, so we were unable to explore this risk factor.

In conclusion, in this cohort of injured workers, over a third went on to have at least one subsequent work-related injury in the following 24 months. This is important in itself and indicates a potentially important intervention point for work-related injury prevention. This is additionally important within the New Zealand context given the considerable financial support that ACC provides.38 A number of factors were associated with a subsequent work-related injury claim and these factors may be useful for those involved in the treatment and rehabilitation of people with work-related injuries. Having an inadequate household income was associated with work-related subsequent injury. This may indicate a quicker return to work than optimal, or inadequate rehabilitation, for those less financially secure, and more attention to this aspect is warranted. However, there is no clear dominant relationship with work-related subsequent injury and the preferred approach may be to view the initial injury as the stimulus to try to prevent subsequent injuries.

Acknowledgments

The authors are grateful to the study participants for sharing their information.

References

Footnotes

  • Contributors HH and SD were colead investigators of the Subsequent Injury Study (SInS); SD was the lead investigator of the Prospective Outcomes of Injury Study (POIS); HH wrote the draft of the manuscript; AS analysed the data; all coauthors have contributed to the writing and editing of the manuscript, have contributed to the interpretation of the findings and have read and approved the final manuscript.

  • Funding The SInS was funded by the Health Research Council of New Zealand (2015–2017). The POIS was funded by the Health Research Council of New Zealand (2007–2013) and cofunded by the Accident Compensation Corporation, New Zealand (2007–2010).

  • Competing interests None declared.

  • Patient consent for publication Not required.

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

  • Data availability statement No data are available. Due to ethical constraints the data cannot be shared but anyone interested in pursuing collaborative research should contact sarah.derrett@otago.ac.nz.