Low back pain in compensated Australian workers: a retrospective cohort study

Objectives To describe incidence, duration, and patterns of working time loss claims in compensated Australian workers with low back pain (LBP), and compare this with other musculoskeletal conditions. Methods The National Dataset for Compensation-based Statistics (NDS) was used for this study. Any accepted workers' compensation time loss claims for LBP, limb fracture or limb soft tissue disease occurring between July 2010 and June 2015 were included. Demographic information, occupation, and total cumulative time loss data were extracted. Counts, rates per 10,000 covered workers, the relative risk and median duration of time loss were calculated. Multivariate Cox and quantile regression models were used to determine factors affecting time loss duration and patterns. Results There were 56,102 LBP claims, 42,957 limb fracture claims, and 18,249 limb soft tissue disease claims. The relative risk of a claim for LBP was significantly greater than limb fractures after adjustment for all covariates (RR 1.30, 95%CI 1.29 - 1.32, p < 0.001). LBP claims had longer median time loss (9.39 weeks) than limb fracture claims (9.21 weeks). Quantile regression demonstrated that LBP claims were more likely than limb fracture claims to resolve within seven weeks, and to persist for periods beyond seven weeks. Conclusions There were differential patterns of time loss in LBP claims and limb fracture claims. The interaction between conditions, and policies and practices may contribute to these patterns. The findings should reiterate to workers' compensation stakeholders the importance of returning a worker to work as soon as practicable, to avoid future delays or challenges.


INTRODUCTION
Low back pain (LBP) is a prevalent musculoskeletal symptom and the leading contributor to 67 hypothesised were likely to have different (limb fractures) and similar (muscle and tendon 115 diseases) patterns of functional recovery and occupational outcomes compared to LBP. 22, 23 116 The sample included workers aged > 15 years and < 80 years. Claims were excluded if they 117 contained unlikely weekly working hours prior to a claim (< 1 hour and > 100 hours). Claims 118 with time-loss less than two weeks or greater than 365 weeks were excluded, and censor 119 indicator marked the maximum duration of any one claim at 104 weeks' time loss (i.e., two 120 years). 19,20,24 Filtering claims in this manner creates a standardised cohort across all 121 jurisdictions. For example, some jurisdictions require an employer to pay the first 10 business 122 days wage replacement (i.e., the two week filter), and each jurisdiction has a different 123 maximum wage replacement period (i.e., censoring at 104 weeks, or two years). These 124 eligibility criteria have been applied previously in studies using the NDS. 19,20,24 Application 125 of these eligibility criteria is described in Figure 1. 126

127
The primary outcome variables for this study were (1) the incidence of accepted workers' 128 compensation time loss claims per 10,000 covered workers, and (2) duration of time loss for 129 workers with accepted claims in weeks. These outcomes have been used in previous similar 130 studies. 19,20,24 131  7 at the time of injury was reported as age group (15- 24 years, 25-34 years, 35-44 years, 45-55  138 years, and 55 > years), and sex reported in binary terms (male / female). Australian and New 139 Zealand Classification of Occupations (ANZSCO) major codes were used to define eight 140 major occupation groups. 25 Jurisdiction was defined as the workers' compensation 141 jurisdiction in which the claim was accepted. Socioeconomic status was defined by the Index 142 of Relative Socioeconomic Advantage and Disadvantage (IRSAD). 26,27 As per previous 143 analyses, the middle three quintiles were collapsed, with the most advantaged and most 144 disadvantaged quintiles at either extreme. 24  Descriptive statistics were first used to determine the counts of time loss claims for LBP and 154 the comparator conditions for each covariate. Claim incidence was calculated for LBP and 155 the comparator conditions for sex, age group, occupation and jurisdiction. Poisson regression 156 was used to calculate the unadjusted relative risk of a claim for each condition. The 157 covariates sex, age group, occupation, and jurisdiction were then added to an adjusted model. 158 Other covariates were not included in Poisson regression as denominator data (i.e., total 159 covered workers) were not available. The log of the total number of covered workers was 160 used as an offset in Poisson regression. 19 Results were expressed as unadjusted and adjusted 161 relative risk, with 95% confidence intervals (CI). 162 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint (i.e., a return to work) relative to one another (i.e., they were not proportional). . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint .

189
There were a greater number of claims for LBP (n=56,102) than limb fracture (n=42,957) and 190 limb soft tissue disease (n=18,249) (see Table 1 and Figure 1). The incidence of claims for 191 LBP was also higher than both comparator conditions at 9.37 per 10,000 covered workers 192 (7.17 per 10,000 covered workers for limb fracture, and 3.05 per 10,000 covered workers 193 limb soft tissue disease). The relative risk of a claim for LBP was significantly greater than 194 for limb fractures (RR 1.31, 95% CI 1.29 -1.32, p < 0.001), and remained significant after The incidence of claims was greater for males than females for LBP and limb fracture claims, 229 but there was a relatively small difference between sexes for limb soft tissue disease claims. 230 The incidence of claims in all three condition groups generally increased with age, but the 231 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.02.24.20027540 doi: medRxiv preprint trend was most noticeable in LBP claims. More physically demanding occupations had a 232 significantly greater incidence of claims for all conditions, and increased relative risk in the 233 adjusted model. This was most notable for LBP claims, with Machinery Operators and 234 Drivers and Labourers claiming for LBP at a rate of 23.6 and 22.7 per 10,000 covered 235 workers, respectively. 236 The highest incidence of claims per jurisdiction was for LBP in Queensland (12.2 per 10,000 237 covered workers). However, other jurisdictions had relatively similar rates of LBP claims at 238 8.8 to 11.3 per 10,000 covered workers. The incidence of claims for limb fracture and limb 239 soft tissue disease were lower, with limb soft tissue disease claims as low as 0.6 per 10,000 240 covered workers in Western Australia (WA). 241

Time loss
242 Median time loss for LBP (9.39 weeks, IQR 3.95 -30.2) and limb fractures (9.22 weeks, IQR 243 5.07 -19.30) were similar, despite a more right-skewed 75 th percentile for LBP (see Table 2). 244 The median time loss for limb soft tissue disease claims was greater at 14.40 weeks (IQR 245 5.92 -40.4). A relatively large proportion of claims for LBP had a short duration of time loss 246 (see Figure 2A). However, numerous claims lasted for a greater duration, extending the 75 th 247 percentile. The majority of limb fracture claim time loss appeared to cluster closer to the 248 median, with only a relatively small proportion of outliers. The distribution of time loss for 249 limb soft tissue disease claims did not appear to follow a specific trend. 250 251 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.02.24.20027540 doi: medRxiv preprint There did not appear to be an effect of age on time loss for LBP claims, nor were there 257 substantial differences in time loss between occupations (see Table 2). Median time loss for 258 LBP claims was highly varied between jurisdictions; LBP claims in NSW lasted a median of 259 6.77 weeks, compared to 18.80 weeks in Victoria. This same inter-jurisdictional variability 260 was apparent for limb soft tissue diseases, but less so for limb fractures. There did not appear 261 to be substantial differences in time loss for LBP claims between IRSAD. However, median 262 time loss was up to 3.40 weeks greater for cases with missing socioeconomic status. LBP 263 claims in major cities tended to have greater median time loss than in more remote areas. 264 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.02.24.20027540 doi: medRxiv preprint  CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint .
found.) demonstrated that a larger time loss for LBP claims were more likely to cease than 273 limb fracture claims in approximately the six weeks. However, this trend reverses after 274 approximately six weeks, with time loss for limb fracture claims more likely to cease after 275 this time period. Limb soft tissue disease claims displayed the most shallow survival curve. 276 Quantile regression confirmed the patterns observed in the Cox model (see Table 3

288
This study aimed to determine the incidence and duration of working time loss due to work-289 related LBP in Australia compared with other common work-related musculoskeletal 290 conditions. There was a greater incidence of LBP time loss claims compared to limb fracture 291 and limb soft tissue disease claims. The relative risk of claiming for LBP was greater than 292 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . loss duration may be a combination of factors from both conditions; workers with LBP may 308 be more likely to return to work (i.e., time loss has ceased) as their pain has resolved quickly, 309 with workers with limb fracture more likely to return to work after their fracture has healed 310 sufficiently to resume activities. However, without further information it can only be 311 hypothesised that this is the case. 312 The interaction between each condition and policies and practices may also contribute to the 313 fluctuating time loss duration patterns observed in this analysis. For example, a practice such 314 as diagnostic imaging may contribute to variable outcomes for each condition. Imaging has 315 minimal diagnostic merit in the acute phase of LBP, and may be detrimental to a return to 316 work for compensated workers. 37, 38 In the case of limb fractures however, imaging is an 317 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . essential and useful tool for diagnosis, and may be used to indicate that a worker is capable of 318 returning to work. Other policy features may contribute to the observations made in this study. 319 For example, both LBP and limb soft tissue disease claims observed a notable cluster of 320 distribution of claims lasting approximately 100 weeks. This is unlikely to translate to 321 recovery of large portion of claimants at this point. Instead, it is more likely due to the 322 cessation of benefits in short-tail workers' compensation schemes, such as Victoria. 10 323 The higher rate of LBP in older workers aligns with previous research. 39 However, the 324 incidence of claims for work-related LBP does not necessarily reflect the incidence of work-325 related LBP. For example, lower claim rates in one jurisdiction do not necessarily mean that 326 it is safer, and may instead indicate differences in system eligibility criteria, cultural or local 327 norms, or health literacy. 24 The higher rate of claims for LBP in more physically demanding 328 occupations may indicate a link between physically demanding work and LBP. However, this 329 link has previously been debated elsewhere. 40 It may be that workers are more likely to lodge 330 a claim for LBP if they have a physically demanding occupation because LBP limits their 331 ability to perform their normal duties. Finally, the effect of policy differences on workers' 332 compensation claims rates between workers' compensation schemes across Australia on time 333 loss has previously been reported and is reflected in these findings. 24 334 This study benefited from a large, population-wide sample. The use of robust and novel 335 statistical methods provides new insights into time loss in this population. The analytic 336 techniques used provide new prognostic insight in the return to work setting, by 337 understanding the likelihood of time loss duration associated with each of the three included 338 conditions at different times, rather than simply the pooled likelihood of return to work. 339 There were also several limitations to this study. Firstly, there was missing data for some 340 variables, and denominator data for all variables was not available. 341 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . may contain errors. For example, it was possible that some conditions included in the data 343 were misclassified. Injury classification (such as nature or location) is performed by workers' 344 compensation schemes, and may be subject to data entry errors. Several implausible 345 conditions (e.g., tennis elbow in the hip) were identified and removed during data cleaning. 346 Higher-level TOOCS codes (e.g., soft tissue diseases) were used to absorb this possible 347 misclassification. However, this may have led to a cohort definition that was too clinically 348 broad; an upper limb fracture and lower limb fracture are likely to have different effects on a 349 person's mobility. However, we believe that even with such limitations the analyses captured 350 the overall important trends. 351 Finally, most statistical relationships were significant despite some relatively small effect 352 sizes. This was likely due to the large sample size available for this study, and while not 353 necessarily a limitation, future analyses may benefit from sub-sampling to ensure effects 354 remain significant at even smaller sample sizes. 355 Future research should consider that time loss should not be viewed a single fixed value, but 356 a fluctuating concept. Results presented here demonstrated that the likelihood of return to 357 work may change with time depending on condition, and other recent research has 358 demonstrated that gender has a similar time-dependent effect on working time loss outcomes. 359 As in the general population, LBP is a prevalent musculoskeletal symptom in compensated 360 Australian workers. Time loss due to claims for LBP is relatively similar to claims for limb 361 fractures. However, changes in the likelihood of time loss duration for LBP relative to limb 362 fracture claims varies depending on the duration of the claim. These findings should reiterate 363 to clinicians, claims managers, and workers' compensation schemes the importance of 364 returning a worker to work as soon as practicable, to avoid future delays and challenges. 365 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.