Predictors of Return To Work Following Motor Vehicle Related Orthopaedic Trauma
Predictors of Return To Work Following Motor Vehicle Related Orthopaedic Trauma
Predictors of Return To Work Following Motor Vehicle Related Orthopaedic Trauma
Abstract
Background: Work disability following motor vehicle related orthopaedic trauma is a significant contributor to the
burden of injury and disease. Early identification of predictors for return to work (RTW) is essential for developing
effective interventions to prevent work disability. The study aim was to determine the predictors (including
compensation related factors) of time to RTW following motor vehicle related orthopaedic trauma.
Methods: Admitted patients were recruited prospectively from two trauma hospitals with upper and/or lower
extremity fractures following a motor vehicle crash. Baseline and follow up data were collected by written
questionnaire. For baseline, this occurred in person within 2 weeks of injury. For follow up, this occurred by mail at
six, 12 and 24 months. Additional demographic and injury-related information was retrieved from hospital
databases. Analysis involved: descriptive statistics; logrank test to detect survival distributions of categorical
variables; and Cox proportional hazards regression models for risks of time to RTW using baseline characteristic and
compensation related variables (at 6 months).
Results: Of 452 study participants 334 (74 %) were working pre-injury: results are based on this subset. Baseline
characteristics were mean age 36 years (13.9 Standard Deviation [SD]), 80 % male; 72 % self-assessed very good-
excellent pre-injury health, 83 % household income > AU$40,000 (Australian Dollar). Follow up data was available for
233 (70 %), 210 (63 %), and 182 (54 %) participants at six, 12 and 24 months respectively.
Significant risks of a longer time to RTW were greater injury severity, as measured by the New Injury Severity Score
(NISS) (Hazards Rate Ratio [HRR] = 0.54, 95 % CI 0.35-0.82); and lower occupational skill levels (HRR = 0.53, 95 % CI
0.34-0.83). Significant risks of a shorter time to RTW were: recovery expectations for usual activities within 90 days
(HRR = 2.10, 95 % CI 1.49-2.95); full-time pre-injury work hours (HRR = 1.99, 95 % CI 1.26-3.14); and very good
self-assessed pre-injury health status (HRR = 1.41, 95 % CI 0.98-2.02). Legal representation (analysed at six months
only) was not associated with time to RTW. At each time period, there were 146 (63 %), 149 (71 %), and 137 (76 %)
working participants.
Conclusions: A longer time to RTW was associated with greater injury severity and lower occupational skill levels;
while a shorter time to RTW was associated with recovery expectations for usual activities within 90 days, full-time
pre-injury work hours, and very good self-assessed pre-injury health status. Our findings reinforce existing research.
There is an opportunity to trial interventions that address potentially modifiable factors. The issues surrounding
legal representation are complex and require further research.
Keywords: Compensation and redress, Wounds and injury, Multiple trauma, Return to work
* Correspondence: dmur0062@uni.sydney.edu.au
5
John Walsh Centre for Rehabilitation Research, The University of Sydney,
Kolling Institute, Royal North Shore Hospital, Pacific Hwy, St Leonards, NSW
2065, Australia
Full list of author information is available at the end of the article
© 2016 Murgatroyd et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 2 of 13
Western Sydney Local Health District, South Eastern Alcohol consumption was measured using the first
Sydney Local Health District, and The University of three questions of the Alcohol Use Disorders Identifica-
Sydney). tion Test: Self-Report Version (AUDIT-C) [26]. The
word ‘standard’ and ‘in the past year’ were added. Risk of
Injury related factors long/short term harm due to alcohol consumption was
Injuries were coded using the Abbreviated Injury Scale assessed with the National Health and Medical Research
(AIS) (1990 Revision, Update 98) [19]. The AIS ranks in- Council (NHMRC) levels [27]. Because these levels were
juries from one to six (six is not survivable). The Injury mismatched with the AUDIT-C categories, an algorithm
Severity Score (ISS) and New Injury Severity Score was used based on the Bettering the Evaluation of Care
(NISS) were calculated by summing the squares of the and Health (BEACH) Survey, (Associate Professor K
three highest AIS scores from different body regions Conigrave, personal communication March 19, 2007).
(ISS), and regardless of body region (NISS). They are Categories for other study factors are explained in the
indicators of potential mortality [20]. Injuries were Tables.
classified as minor – moderate [1–8], serious [9–15]
or severe – critical (16–75) [21]. Compensation related measures
The majority of compensation related factors were re-
Socio-demographic factors corded at six months because most questions would
Socio-demographic factors included age, gender, marital have been unanswerable at baseline. The following
status, occupation, and education. Income was measured questions were asked: claim made (Y/N); claim type
exclusive and inclusive of household structure to allow (Compulsory Third Party [CTP]/Workers Compensa-
for potential differences in income distribution. An tion [WC]/other); claim accepted (Y/N/don’t know);
adjusted income (inclusive of household structure) was and legal representation obtained (Y/N). Claim made
calculated by dividing the income by the sum of points: ‘Yes’ was defined as making a personal injury claim of
1 for the first person aged ≥15 years; 0.5 for each add- any type; which included a CTP Accident Notification
itional person aged ≥15 years; and 0.3 for each person Form (ANF) for expenses less than AU$5,000 (Australian
aged <15 years [22]. Dollar) within 28 days of injury. At baseline self-reported
fault of the driver was measured (i.e. whether the driver
Health related factors considered that they caused the crash). Passengers and
Self-reported chronic illnesses were measured as an indi- pedestrians were considered not at fault.
cator of baseline health status, they were asthma, cancer, In NSW, CTP personal injury insurance is a privately
heart and circulatory conditions, diabetes, arthritis, underwritten, statutory, modified common law scheme.
osteoporosis, mental and behavioural problems, and All motor vehicles travelling on public roads must be
neck/back disorders. These self-reported illnesses were registered and insured for CTP. A CTP claim is made
compatible with the National Health Priority Areas ini- against the owner or driver of the vehicle at fault. Since
tiative (conditions that imposed high social and financial April 2010, regardless of who was at fault, anyone
costs on Australian society) [23]. A chronic condition injured in a motor vehicle crash can access limited
was defined as one which the patient currently has, and entitlements (medical expenses and lost wages up to
which has lasted or is expected to last for six months or AU$5,000). The WC scheme is publically underwritten
more, from the Australian Bureau of Statistics (ABS) with statutory benefits and administered by private in-
Health Survey [22, 23]. Other factors included: recent surers. To make a claim for injury the motor vehicle
injuries (other than the motor vehicle crash) in the last crash must have occurred during travel between place of
4 weeks requiring medical intervention or a decrease in employment, home and/or any work-related place and a
usual activities; medication use in the last 2 weeks for a person injured (regardless of fault). Further, the insurer
chronic illness; and smoker status [22]. must be notified of an injury within 48 hours and there
Previous research has found an association between is a legal obligation under the NSW WC legislation for
poor expectations for recovery and poor RTW and/or employers to accommodate RTW of an injured em-
health outcomes, but there was an absence of validated ployee, although there is no obligation under the NSW
measures [9, 18, 24, 25]. Therefore, we used two applic- CTP legislation [28, 29]. In 2015, the government regula-
able measures from a large Canadian study of soft tissue tors of these schemes merged to form the State Insurance
injuries [24]. The questions asked were: If you were Regulatory Authority (SIRA).
working before the motor vehicle accident, do you think For both schemes, a claim must be lodged within six
you will recover enough to return to your usual job (Y/N); months of injury and the insurer has three months to
and How long do you think it will take for you to return determine final liability (accept or deny the claim).
to your usual activities (number of days). Provisional acceptance of liability enables earlier
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 4 of 13
payment for medical expenses, and for WC weekly wage can be regarded as noise and excluded to avoid over fit-
benefits based on work capacity and weeks since injury ting in the model [31]. Furthermore, explained variance
[29]. In CTP, lump sum payments are available on a using R-squared describes the relative importance of
case-by-case basis for financial hardship. Entitlements adding each variable into the Cox regression model.
include past and future losses across each scheme (e.g. Data from participants where the endpoint (RTW) had
medical expenses, loss of income, and pain and suffer- not occurred or was unknown at 24 months were con-
ing/impairment) [28, 29]. Legal representation can also sidered censured. In studies of survival, in which the
be obtained at any time for either scheme. outcome is death, Hazard Rate Ratios (HRR) greater
than 1 indicates risk. However, in this study, the fewer
Outcome measure - return to work cumulative days of time taken to RTW, the more posi-
There are no standardised measures for RTW. Those tive the outcome, in terms of injury recovery/RTW, and
used in this study encapsulated self-reported duration the higher the HRR. Therefore, a HRR less than 1 indi-
and level of work [6, 7]. The primary measure was time cates higher risk and a longer time taken to RTW. A test
(days) to return to work (i.e. from date of injury to date of proportionality was performed on all predictors, and
of RTW). At each time period work status (Y/N) was claim made and legal representation. The assumption of
measured. Working participants were then asked the proportionality was not violated (p > 0.05) [33].
date of RTW, if they were working full/modified duties A separate Cox proportional hazards regression ana-
(e.g. lifting restrictions), and full-time (usually working lysis was done for compensation related variables (claim
at least 35 hours per week) or part-time (usually working made and legal representation), as these variables were
1–35 hours per week) [30]. These questions were asked measured at the six month time point and only a portion
pre-injury (baseline) and post-injury (six, 12 and of participants made a claim and sought legal represen-
24 months). Participants were also asked if their inability tation. In this Cox regression, claim made and legal rep-
to RTW was crash-related, and if they had changed their resentation were added to the final variables in the
occupation following injury. baseline RTW model. All data analysis was performed
using SPSS statistical software version 22 (SPSS Inc,
Data analysis USA).
RTW baseline characteristics, including full/modified
duties and full/part-time, were summarised using de- Results
scriptive statistics. Outcomes were assessed using sur- From November 2007 to February 2011, 840 eligible par-
vival analysis with Cox proportional hazards regression ticipants were admitted to hospital across both sites, 491
models employed to determine the multivariate predic- were screened (349 eligible participants missed being
tors of time to RTW. The Cox model is considered an screened due to resource limitations), and 452 (92 %)
appropriate approach to accounting for time to an event consented to participate. There were 31 refusals and
[31]. The variables selected for the model have been eight who were discharged and unable to be contacted.
shown to be independent predictors of RTW and/or po- Additional information about recruitment and follow up
tential confounders of poorer outcomes in other research for study participants is shown in Fig. 1. There were sig-
[7, 16–18]. Similarly, compensation-related factors were nificant differences (p < 0.05) in baseline characteristics,
selected for the same reasons [2, 4, 7, 8]. namely socio-demographic and socio-economic factors,
Selection of variables for the Cox model was based on between those working and not working pre-injury.
associations between baseline characteristics, including These differences were expected and people not working
compensation related factors and time to RTW. These were not included in the analyses (data not shown). Of
were assessed using the logrank test to detect differences the 452 participants, our subsequent results are based
in the survival distributions across categorical variables. on the subset of 334 (74 %) participants who worked up
All variables with p-value ≤ 0.20 were entered into the Cox to the time of injury.
regression model using a backward elimination process
with an entry p-value < 0.05 and an exit p-value < 0.10. Baseline characteristics
Variable selection was confirmed through explained Baseline characteristics were: mean age 36 years (13.9
variation and predictive accuracy using R-squared Standard Deviation [SD]); 80 % male; 72 % self-assessed
values calculated with the Cox and Snell R-squared ap- very good-excellent pre-injury health; 83 % annual
proach and a concordance index [32]. The concordance household income > AU$40,000. Follow up data was
index is a widely applicable measure with progressive available for 233 (70 %), 210 (63 %), and 182 (54 %) par-
addition of factors that improve discrimination of the ticipants at six, 12 and 24 months respectively. There
model. When a variable is added and the c-index plat- were significant differences between responders and
eaus or decreases, that variable and additional variables non-responders at six, 12 months and 24 months; this is
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 5 of 13
explained in Table 1. For all other variables there was no 24 months, and two participants who had returned to
significant difference (p > 0.05) (data not shown). In work at six months no longer worked at 24 months.
addition, there were significant differences between
those participants that made a claim at six months (n = Predictors of time to return to work
140) and those that did not (n = 91). This reflected eligi- For all 334 study participants, the median time to RTW
bility to claim under the NSW legislation: those partici- was 231 days (95 % CI 190.05-271.95). For RTW, the
pants more likely to make a claim were not at fault probability of participants working at six months was
(78 %), had crashed on a public road (94 %), and worked 40.6 %, at 12 months was 62.2 %, and at 24 months was
pre-injury (77 %). For all other variables there was no 74.2 %. This is based on the Kaplan-Meier estimates of
significant difference (p > 0.05) (data not shown). the survival curve as shown in Fig. 2. Associations be-
tween baseline characteristics and time to RTW are
Characteristics of return to work shown in Table 2. The significant variables identified in
At baseline, of the 334 who worked, 83 % were full-time the logrank test, including age and sex, were entered
and 96 % performed full duties. At six months, of the into the Cox proportional hazards regression model.
146 (63 % of responders) who worked, 65 % were full- Based on the variables identified from the backwards
time and 64 % performed full duties. At 12 months, of elimination process, Table 3 shows the concordance (c-
the 149 (71 % of responders) who worked, 73 % were index) and R-squared of each of the variables as they
full-time and 69 % performed full duties. At 24 months, were added to the Cox model. The c-index plateaued at
of the 137 (75 % of responders) who worked, 81 % were the variable of smoking history; the remaining variables
full-time and 79 % performed full duties. In addition, at were not included in the model. Of the variables that
six months, failure to RTW was related to the crash for were not significant only age and sex were deemed ne-
84 %, and 10 % had changed occupation. At 12 months, cessary to be included in the Cox model.
failure to RTW was related to the crash for 80 %, and The Cox proportional hazards regression model for
16 % had changed occupation. At 24 months, failure to risks of time to RTW is presented in Table 4. The sig-
RTW was related to the crash for 81 %, and 22 % had nificant risks of taking a longer time to RTW were:
changed occupation. greater injury severity (NISS), namely those with severe-
Overall, there were nine participants who initially critical injuries as compared to those with minor-
returned to work at either six or 12 months but did not moderate and serious injuries, and lower occupational
remain at work during the subsequent follow up pe- skill levels as compared to managerial or professional
riod(s). Of these, five participants worked at six and skill levels. In the same model, the significant risks of
12 months but not at 24 months. Two participants taking a shorter time to RTW were: full-time pre-injury
returned to work at 12 months but no longer worked at work hours compared to part-time pre-injury work
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 6 of 13
Table 1 Baseline characteristics and health status of participants in the study compared to non-participants at six, 12 and 24 month
follow up
Participation at six months Participation at 12 months Participation at 24 months
a a
Variable No (n = 101) Yes (n = 233) P No (n = 124) Yes (n = 210) P No (n = 151) Yesa (n = 182) p
Age (years), Mean (SD) 31.9 (12.0) 38.2 (14.2) ** 31.7 (11.5) 39.0 (14.4) ** 31.5 (11.7) 40.3 (14.3) **
New Injury Severity Score, No. (%) NS * NS
Minor - moderate 1-8 22 (21.8) 41 (17.6) 27 (21.8) 36 (17.1) 33 (21.9) 30 (16.5)
Serious 9-15 50 (49.5) 94 (40.3) 61 (49.2) 83 (39.5) 71 (47.0) 73 (40.1)
Severe - critical 16-75 29 (28.7) 98 (42.1) 36 (29.0) 91 (43.3) 47 (31.1) 79 (43.4)
Marital status, No. (%) ** ** **
Single 60 (60.0) 87 (37.5) 68 (54.8) 79 (38.0) 86 (57.0) 61 (33.9)
Married/defacto 36 (36.0) 131 (56.5) 53 (42.7) 114 (54.8) 61 (40.4) 105 (58.3)
Divorced/widowed 4 (4.0) 14 (6.0) 3 (2.4) 15 (7.2) 4 (2.6) 14 (7.8)
Occupation skill levelb, No. (%) * NS *
Managers/professionals 20 (19.8) 58 (25.0) 23 (18.5) 55 (26.3) 60 (26.1) 38 (17.1)
Tradespersons 26 (25.7) 85 (36.6) 36 (29.0) 75 (35.9) 71 (30.9) 55 (24.8)
Intermediate clerical 17 (16.8) 35 (15.1) 21 (16.9) 31 (14.8) 27 (11.7) 37 (16.7)
Elementary related 38 (37.6) 54 (23.3) 44 (35.5) 48 (23.0) 53 (23.0) 72 (32.4)
c 2
Body Mass Index (BMI) (kg/m ) * NS NS
<18.50 (underweight) 4 (4.0) 2 (0.9) 4 (3.2) 2 (1.0) 4 (2.6) 2 (1.1)
18.50-24.99 (normal) 45 (44.6) 78 (33.8) 50 (40.3) 73 (35.1) 64 (42.4) 58 (32.2)
≥25.00 (overweight) 35 (34.7) 89 (38.5) 47 (37.9) 77 (37.0) 53 (35.1) 71 (39.4)
≥30.00 (obese) 17 (16.8) 62 (26.8) 23 (18.5) 56 (26.9) 30 (19.9) 49 (27.2)
Smoking history, No. (%) * NS NS
Current smoker 38 (38.0) 51 (22.0) 40 (32.5) 49 (23.4) 48 (32.0) 40 (22.1)
Ex-smoker 23 (23.0) 67 (28.9) 32 (26.0) 58 (27.8) 36 (24.0) 54 (29.8)
Never smoked 39 (39.0) 114 (49.1) 51 (41.5) 102 (48.8) 66 (44.0) 87 (48.1)
Self-reported chronic 17 (16.8) 76 (32.6) ** 26 (21.0) 67 (31.9) * 35 (23.2) 58 (31.9) NS
illnesses (yes) No. (%)
Medication use (current), No. (%) 11 (10.9) 56 (24.1) ** 13 (10.5) 54 (25.8) ** 19 (12.6) 48 (26.5) **
Vehicle type, No. (%) * * *
Motor vehicle 61 (60.4) 114 (48.9) 71 (57.3) 104 (49.5) 87 (57.6) 88 (48.4)
Motorcycle 31 (30.7) 109 (46.8) 42 (33.9) 98 (46.7) 53 (35.1) 87 (47.8)
Bicycle 9 (8.9) 10 (4.3) 11 (8.9) 8 (3.8) 11 (7.3) 7 (3.8)
a
Participation status ‘yes’ was measured using the information recorded in variables - work status at six, 12 and 24 months, and the Short Form-36 Version 2.0
(SF36v2), Physical Component Score (PCS) at six, 12 and 24 months respectively
** P < 0.01, *P < 0.05, NS not significant
b
The measure for occupation is from the Australian Standard Classification of Occupations (ASCO), Cat. No. 1220.0, Australian Bureau of Statistics 1997. See
Table 2, Occupational skill level for all categories
c
BMI classification is from the Global Database on Body Mass Index, World Health Organisation
hours; recovery expectations for usual activities of ≤90 days claim, 95/140 (68 %) sought legal representation at six
compared to recovery expectations for usual activities months. Making a claim at six months was not associ-
of ≥90 days, and having very good self-assessed pre- ated with time to RTW, the HRR was 0.89 (95 % CI
injury health status as compared to having excellent 0.60-1.33). The logrank analysis of seeking legal repre-
self-assessed health status. sentation at six months was associated with a longer
In terms of compensation related factors, overall time to RTW (see Table 2). However, in the Cox regres-
140/231 (60 %) made a claim at six months (there was sion with baseline variables included, legal representa-
missing data for 2/233 responders for the compensation tion was not associated with time to RTW; the HRR was
related questions at six months). Of those who made a 0.81 (95 % CI 0.54-1.21).
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 7 of 13
Fig. 2 Kaplan-Meier estimate of the cumulative time (days) to return to work for study participants (n = 334)
Discussion injury work hours and very good pre-injury health status
Time to RTW was associated with both injury and non- is likely to be dependent on the study population. These
injury related factors in a cohort with motor vehicle re- factors have been reported as predictors of a shorter time
lated moderate-severe orthopaedic injuries. The main to RTW and recovery [4, 7, 16, 36]. Again, results are in-
findings were that greater injury severity and lower oc- consistent and measures vary. For example, higher base-
cupational skill levels were significant risks of a longer line income or job involvement is measured instead of
time to RTW. Whereas, recovery expectations for usual pre-injury work hours [4, 7].
activities of ≤90 days, full-time pre-injury work hours, Recovery expectations and illness perception, and low
and very good self-assessed pre-injury health status were self-efficacy, are significant predictors of RTW rates
significant risks of a shorter time to RTW. Legal repre- and/or recovery across a range of injuries and illnesses
sentation at six months was not associated with time to [7, 24, 25, 37, 38]. This shows they are robust predictors
RTW. of RTW that relate to the individual rather than a spe-
cific diagnosis. These predictors are complex and multi-
Predictors of time to return to work dimensional [24, 37, 39]. Theoretically, self-efficacy (i.e.
In our study, the significance of injury severity as a risk person’s belief in their own competence) materialises
of time to RTW was driven by those with severe-critical during childhood and evolves throughout life. Those
injuries (ISS 16–75). Similarly, the significance of occu- with strong self-efficacy master problems, recovering ex-
pation was driven by those with lower occupational skill peditiously; those with weak self-efficacy avoid chal-
levels such as elementary workers and tradespersons. lenges, focusing on negative outcomes [40]. Similarly,
Existing research confers that injury severity and occu- illness perception is based on a self-regulatory model
pational skill level are predictors of RTW, particularly that appraises a person’s response to their illness event
for lower limb injuries [3, 4, 8, 34, 35]. This result is not [41]. In other words, how well you think you will recover
unforeseen given the socio-demographic profile of the can influence how well you actually recover.
cohort – mean age 36 years, 80 % male, and 33 % trade- The association between making a claim and legal rep-
spersons, advanced clerical or service workers. Adequate resentation, and poor RTW rates or recovery is well doc-
physical function is likely to be an important component umented [7, 10]. As before, results vary according to the
of work. In other research, these factors have been inde- study population, outcome measures, and possibly the
pendent predictors of RTW, although the level of evidence compensation scheme. In this study, the logrank test
is variable [7]. Likewise, the significance of full-time pre- between legal representation and time to RTW was
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 8 of 13
Table 2 Baseline characteristics and time to RTW of study Table 2 Baseline characteristics and time to RTW of study
participants (n = 334) participants (n = 334) (Continued)
Variable No. Median Logrank Recovery expectations for work 0.08
(%) time to Test
RTW p value Yes 298 212
Age 334 No 32 280
Injury Severity Score 0.09 Recovery expectations for usual activities <0.001
Minor - moderate 1-8 84 204 ≤90 202 177
Serious 9-15 198 240 >90 111 455
Severe - critical 16-75 52 259 Language other than English 0.08
New Injury Severity Score 0.002 Yes 108 250
Minor - moderate 1-8 63 204 No 226 231
e
Serious 9-15 144 203 Total yearly household income (before tax, 0.17
AU) excluding number of people in
Severe - critical 16-75 127 305 household
Index of Relative Socioeconomic 0.57 ≤$39,999 53 302
Disadvantagea
$40,000-$79,999 110 250
Most disadvantaged 89 215
≥$80,000 148 194
Disadvantaged 34 365
Total adjusted yearly household incomee 0.47
Average 64 270 (before tax, AU) including number of
Advantaged 78 207 people in household
Most advantaged 69 198 ≤$39,999 153 240
Sex 0.69 $40,000-$79,999 121 250
Female 67 199 ≥$80,000 37 154
Male 267 244 Body Mass Index (BMI)f (kg/m2) 0.54
Marital status 0.47 <18.50 (underweight) 6 407
Single 147 244 18.50-24.99 (normal) 123 213
Married/de facto 167 212 ≥25.00 (overweight) 124 203
Divorced/widowed/separated 18 270 ≥30.00 (obese) 79 302
Education skill levelb 0.13 Smoking history 0.02
Bachelor degree and above 60 138 Current smoker 89 394
Certificate and advanced diploma 143 213 Ex-smoker 90 207
Secondary education 118 305 Never smoked 153 199
Pre-primary and primary education 10 229 Self-reported chronic illnesses 0.88
Occupation skill levelb 0.01 Yes 93 215
Managers/administrators/professionals/ 78 125 No 241 240
associate professionals Medication use 0.93
Tradespersons/advanced clerical and 111 229 Yes 67 229
service workers
No 266 237
Intermediate clerical/sale/service 52 396
production/transport workers Recent injury other than crash 0.40
Elementary clerical/sales/service/ 92 340 Yes 16 365
labourers/related workers No 316 231
Work level before injury 0.87 Risk of long term harm due to alcohol 0.92
Full Duties 321 231 consumptiong (standard drinksh/week)
Part Duties 13 213 Low risk - ≤28 male or ≤14 female 311 231
Work hours before injury c
0.14 Risky - 29–42 male or 15–28 female 13 358
Full-time 273 215 High risk - ≥43 male or ≥29 female 9 244
Part-time 57 302 Risk of short term harm due to alcohol 0.14
d consumptiong (yes)
Pre-injury job satisfaction 0.28
Yes 116 297
Satisfied 320 231
No 218 215
Not Satisfied 14 342
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 9 of 13
Table 2 Baseline characteristics and time to RTW of study Table 3 Concordance (c-index), R squared as each variable is
participants (n = 334) (Continued) added into the model
Self-reported at-fault 0.04 Factor R-squared C-index
Yes 125 203 Recovery expectations for usual activities 0.087 0.605
No 208 250 Occupation skill levela 0.104 0.643
Vehicle type 0.02 New injury severity score 0.121 0.656
Motor vehicle 175 276 Self-assessed pre-injury health statusb 0.154 0.662
Motorcycle 140 199 Work hours before injuryc 0.177 0.671
Bicycle 19 182 Smoking history 0.194 0.678
Pre-morbid neck pain in last 6 months 0.70 Education skill level a
0.197 0.679
Yes 15 203 Recovery expectations for work 0.206 0.687
No 319 237
Injury severity score 0.212 0.687
Post-morbid neck pain 0.74
Total yearly household incomed 0.225 0.695
Yes 59 215
Self-reported at fault 0.227 0.695
No 275 231
Language other than English 0.240 0.699
Crash on a public road 0.06
Crash on public road 0.241 0.700
Yes 297 240
Risk of short term harm due to alcohol 0.244 0.700
No 37 156
consumptione
Self-assessed pre-injury health statusij, 0.03
Age 0.251 0.704
Excellent 103 240
Sex 0.252 0.707
Very good 137 199
The blank row indicates the point where the concordance index plateaus
Good 78 250 Factors above this were maintained while factors below this were dropped
a
Measures for occupation and education are from the Australian Standard
Fair-Poor 16 -
Classification of Occupations (ASCO), Cat. No. 1220.0, Australian Bureau of
Claim made by 6 months 0.08 Statistics 1997 and the Australian Standard Classification of Education (ASCED),
Cat. No. 1272.0, Australian Bureau of Statistics 2001
Yes 140 178 b
Self-assessed health status is based on Question 1 from the Short Form 36,
No 91 120 version 2, (SF36v2)
c
Measures for full-time (usually working at least 35 hours per week) and
Legal representation at 6 months 0.007 part-time (usually working 1–35 hours per week) are from the Australian
Yes 95 199 Health Survey: Users' Guide, 2011–13, Cat. No. 4363.0.55.001, Australian Bureau
of Statistics
No 136 122 d
Categories of income are from the Household, Income and Labour Dynamics
a
The Index of Relative Socioeconomic Disadvantage (IRSD) is a summary in Australia (HILDA) Survey Wave 6 Household Questionnaire.
e
measure of economic and social conditions within a particular area/postcode Questions to determine risk of harm were from the Alcohol Use Disorders
(e.g. employment, fluency in English and household size). It is taken from the Identification Test: Self-Report Version (AUDIT-C) were resourced from the
Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA), Drink-less program, The University of
Cat no. 2039.0.55.001: Australian Bureau of Statistics; 2001. A low score is Sydney. http://sydney.edu.au/medicine/addiction/drinkless/resources.php
indicative of greater socioeconomic disadvantage
b
Measures for occupation and education are from the Australian Standard
Classification of Occupations (ASCO), Cat. No. 1220.0, Australian Bureau of significant. However, baseline variables in the Cox re-
Statistics 1997 and the Australian Standard Classification of Education (ASCED),
Cat. No. 1272.0, Australian Bureau of Statistics 2001 gression could be a common cause of legal representa-
c
Measures for full-time (usually working at least 35 hours per week) and part-time tion and time to RTW. In addition, it may be that
(usually working 1–35 hours per week) are from the Australian Health Survey:
Users' Guide, 2011–13, Cat. No. 4363.0.55.001, Australian Bureau of Statistics people seek legal representation because they haven’t
d
Pre-injury job satisfaction is based on the stem question from the Measure of returned to work and/or because of other intervening
Job Satisfaction questionnaire by Traynor, M. and Wade, B. 1993
e
Categories of income are from the Household, Income and Labour Dynamics
factors, or those who seek legal representation are more
in Australia (HILDA) Survey Wave 6 Household Questionnaire likely to take longer to RTW. It is not possible to ad-
f
BMI classification is from the Global Database on Body Mass Index, World dress these issues in this study. Regardless, there was no
Health Organisation
g
Questions to determine risk of harm were from the Alcohol Use Disorders association once baseline variables were taken into
Identification Test: Self-Report Version (AUDIT-C) were resourced from the account.
Drink-less program, The University of Sydney. http://sydney.edu.au/medicine/
addiction/drinkless/resources.php
More recently, within a compensable setting, legal rep-
h
1 standard drink contains 12.5 millilitres or 10 grams of alcohol according to resentation has been linked to socio-economic and
the National Health and Medical Research Council (NHMRC), Australian Alcohol psychosocial factors such as: stressfulness of making a
Guidelines Health Risks and Benefits, October 2001
i
Self-assessed health status is based on Question 1 from the Short Form-36, claim; poorer baseline mental health; higher disability;
Version 2.0, (SF36v2)
j
socio-economic disadvantage; and financial entitlements
No median score for fair-poor self-assessed pre-injury health status, the median
indicates that more than half did not return to work (mean = 529 days) [42–44]. This suggests that people seeking legal repre-
sentation have different characteristics compared to
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 10 of 13
Table 4 Cox proportional hazards regression model for predictors of time (days) to RTW
Factor B SE P value HRR 95 % CI
Age 0.009 0.006 0.10 1.010 0.998–1.021
Sex (Male) −0.038 0.202 0.851 0.963 0.648–1.430
New Injury Severity Score 0.007
Minor - moderate 1-8 - - - - -
Serious 9-15 −0.170 0.202 0.401 0.844 0.568–1.254
Severe - critical 16-75 −0.619 0.216 0.004 0.539 0.353–0.822
a
Occupation skill level 0.05
Managers/administrators/professionals/associate professionals - - - - -
Tradespersons/advanced clerical and service workers −0.355 0.204 0.081 0.701 0.470–1.045
Intermediate clerical/sale/service production/transport workers −0.311 0.249 0.211 0.733 0.450–1.193
Elementary clerical/sales/service/labourers/related workers −0.632 0.227 0.003 0.532 0.341–0.829
Work hours before injury (Full-time)b 0.688 0.232 0.003 1.989 1.261–3.136
Recovery expectations for usual activities (≤90 days) 0.741 0.174 <0.001 2.099 1.494–2.949
Self-assessed pre-injury health statusc 0.005
Excellent - - - - -
Very good 0.343 0.184 0.062 1.409 0.983–2.019
Good −0.111 0.219 0.612 0.895 0.582–1.376
Fair-Poor −1.029 0.476 0.031 0.357 0.141–0.908
0.090
Current smoker - - - - -
Ex-smoker 0.368 0.222 0.097 1.445 0.936–2.232
Never smoked 0.434 0.201 0.031 1.543 1.041–2.288
a
Measures for occupation and education are from the Australian Standard Classification of Occupations (ASCO), Cat. No. 1220.0, Australian Bureau of Statistics 1997
b
Measures for full-time (usually working at least 35 hours per week) and part-time (usually working 1–35 hours per week) are from the Australian Health Survey:
Users' Guide, 2011–13, Cat. No. 4363.0.55.001, Australian Bureau of Statistics
c
Self-assessed health status is based on Question 1 from the Short Form-36, Version 2.0, (SF36v2)
other compensable and non-compensable participants; time periods inclusive of time (days) to RTW, full/part-time
which could be partly or wholly responsible for prolong- hours and full/modified duties. At each period the majority
ing their RTW. were working full-time on full duties but below baseline
Other research shows that people seek legal advice to figures. In similar studies RTW varied from 28-68 % at
help with the adversarial claims processes, communica- 6 months [3, 4]; 42 % at 12 months [4]; and 51 % at
tion and administrative deficits with insurers, perceived 24 months [4]. It is difficult to compare RTW rates due to
illegitimacy of their injury, and accessing reasonable en- heterogeneity between populations and the multi-
titlements [11–13]. It may not be ‘legal representation’ dimensional nature of facilitators and barriers for RTW [9].
per se that is associated with RTW but these other fac- Taking into account the unemployment rate in
tors. In addition, there is a lack of granularity when Australia over the follow up period (4.2 % in 2008 –
classifying exposure to legal representation. For ex- 5.4 % in 2013) [46, 47] and the socio-demographic pro-
ample, the ‘no win, no fee’ legal services in NSW CTP file of the study population, the limited RTW rate is of
and WC schemes provide a financial incentive for concern. Accepting that work is good for health and
plaintiff lawyers to take viable cases where extracting a well-being, the converse is also true and poor health
reasonable fee is more likely (e.g. people with more contributes to lost productivity and lower socio-economic
serious injuries, pre-existing and/or crash related fac- status [1].
tors that could allow access to greater financial entitle-
ments) [45]. Strengths and limitations
This prospective study was a representative cohort of
Characteristics of return to work moderate-severe injuries following motor vehicle related
Measuring RTW is challenging – definitions and durations orthopaedic trauma. Standardised and validated mea-
are diffuse [7]. In our study RTW was measured at three sures were used with repeated follow up.
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 11 of 13
Additional measures at baseline would have been Lastly, taking into account the significance of lower
beneficial including initial pain intensity, baseline mental occupational skill levels, it is crucial to improve RTW
health, and other psychological measures. These have rates, and this is feasible, considering the strong evi-
been associated with poorer outcomes following trauma dence base for vocational rehabilitation. The coordin-
[3, 4, 16, 17]. Further, there appears to be a relationship ation of early work-focused health interventions and
between these factors and having a compensation claim accommodating workplaces with modified duties and
[42, 44]. The inclusion of medical and/or vocational in- hours is essential [53]. In WC jurisdictions this is not
terventions, individual job characteristics/tasks, and unforeseen, but in the CTP arena it remains arduous.
workplace/organisational factors would have been useful. There is often no legal impetus on the employer to re-
These determinants of RTW are often population spe- employ an injured worker. In this instance, it may be
cific and amenable to intervention in a compensation necessary to advocate for legislative change or other
setting [6, 9]. policy initiatives like early identification and referral to
Another limitation was moderate loss to follow up. vocational rehabilitation, or proactive claims manage-
The study population characteristics were a plausible ment involving the employer to provide appropriate
reason for loss to follow up. Participants were predomin- duties in the early post-injury period [53].
antly younger males of lower socioeconomic status who
were in semi-unskilled occupations. They were often Conclusions
contactable (see Fig. 1) but would not return the ques- A longer time to RTW was associated with greater injury
tionnaires. Those lost to follow up were younger, less severity and lower occupational skill levels. A shorter time
likely to be married, and less likely to be currently taking to RTW was associated with recovery expectations for
medication. If they had remained in the study, these dif- usual activities of ≤90 days, full-time pre-injury work
ferences could have influenced time to RTW. Lastly, hours, and very good self-assessed pre-injury health status
these findings require validation in future research with following motor vehicle related orthopaedic trauma. Our
larger cohorts and different study populations. findings reinforce existing research. There is an opportun-
ity to trial interventions that address potentially modifiable
Future research and policy implications factors such as poor recovery expectations. The issues
Predictors of RTW are multidimensional and cover nu- surrounding legal representation are complex and require
merous individual, work, organisational and societal do- further research.
mains, which makes high quality research challenging.
Despite the abundant research to date, much remains in-
conclusive [7]. It is important to focus on factors amen- Ethics approval and consent to participate
able to intervention. Injury severity, pre-injury work The study was approved by the governing human research
hours and health status are relatively static outside the ethics committees (South Western Sydney Local Health
bounds of injury prevention programs. District, South Eastern Sydney Local Health District, and
However, expectations for return to usual activities, ill- The University of Sydney).
ness perception and self-efficacy are more dynamic. Vali-
dated measures are now available to gauge this risk Consent for publication
factor of poor RTW and/or recovery [37, 39, 48]. Invest- Not applicable.
ing in interventions such as education, coaching or
multidisciplinary programs could improve RTW rates by Availability of data and materials
adjusting expectations; thereby reducing the associated Results from the dataset are presented in the paper.
costs of lost productivity [49–51]. The full dataset is available from the first author
There is a need to understand the paradigm of legal upon request.
representation, and whether it is a valid measure. Meas-
urement error can occur when the timing of exposure to a Abbreviations
factor does not occur at baseline and/or there is question- ABS: Australian Bureau of Statistics; AIS: abbreviated injury scale;
ASCED: australian standard classification of education; ASCO: Australian
able quality of the measure [52]. Since legal representation standard classification of occupations; AU: Australian dollar; AUDIT-C: alcohol
was measured at six months, not baseline, these results use disorders identification test: self-report version; BEACH: bettering the
need to be interpreted cautiously. Further scheme specific, evaluation of care and health; BMI: body mass index; CALD: culturally and
linguistically diverse; CI: confidence interval; CTP: compulsory third party;
qualitative and quantitative research – principally of pop- HILDA: household, income and labour dynamics in australia; HRR: hazards
ulations at risk for poor RTW – may assist to tease apart rate ratio; IRSD: index of relative socioeconomic disadvantage; ISS: injury
these complexities and provide researchers with ideas for severity score; NHMRC: National Health and Medical Research Council;
NISS: new injury severity score; NSW: New South Wales; PCS: physical component
RTW initiatives and scheme policy makers with opportun- score; RTW: return to work; SD: standard deviation; SEIFA: socio-economic indexes
ities for legislative or policy change if appropriate. for areas; SF36v2: short form-36 version 2.0; WC: workers compensation.
Murgatroyd et al. BMC Musculoskeletal Disorders (2016) 17:171 Page 12 of 13