Utilisation of exercise as part of guideline-based care for hip pain in the Australian workers’ compensation environment

2.1 Study design

This study used a retrospective cohort design. Data was extracted via a standardised audit tool from Western Australian, Tasmanian and Northern Territory finalised workers’ compensation case files of workers with hip pain.

2.2 Subjects

Subject data was extracted from the database of a single insurance company (Insurance Australia Group Limited trading as CGU Workers Compensation) operating as an ‘approved’ insurer within the Western Australian, Tasmanian, and North Territory workers’ compensation schemes. All these schemes are considered ‘risk based’, no fault systems where the risk is carried by licensed insurers with state government oversight [22]. Subjects included in this study were described as having a compensable, specific or non-specific hip or gluteal injury as described in the medical certificates and reports issued by the treating medical practitioners. Cases were selected from a closure date range of 1 July 2014 to 31 August 2017. Subjects were excluded if they had suffered catastrophic injuries, had injuries to multiple body areas, had unresolved pre-existing hip injuries (prior to the compensation claim), had a hip fracture, had inflammatory arthropathy or had pain referred from the lumbar or abdominal regions. Additionally, claims that were impacted by events outside the scope of injury, such as significant delay in liability decision or concurrent non-compensable injuries were excluded from the study. This was because the clinical care pathway for these workers would likely deviate from the hip pain clinical practice guidelines. Ethical approval was obtained through Curtin University Human Research Ethics Committee (Reference Number 12845). A data access and confidentiality agreement was established between Curtin University and the insurer. Given that data was retrospective, subject permission could not be obtained for this study. However, the insurer had previously acquired consent to gather and maintain appropriate medical information upon lodgement of all people’s workers’ compensation claims. Physical files were viewed in a secure area within the insurer’s premises and de-identified data was stored on password protected computers to avoid unauthorised dissemination of claim details.

2.3 Procedure

An excel spreadsheet auditing tool was developed and used to extract demographic data, clinical care details and claim outcomes from case files. The tool was based on contemporary models of care for the management of compensable workers coupled with clinical practice guidelines for hip pain management [3, 10]. Five files were randomly selected to test and refine a draft version of the audit tool. The files were audited by two blinded, independent assessors. Where discrepancies or confusion arose in extracting data, modifications were made to the audit tool. Once these were made, a further three files were audited in the same fashion, for an additional round of refinement. These eight files were not included in the final study. Following refinement, the audit tool was used to extract the case file data. Files that had discrepancies between the two assessors were audited by a third, blinded assessor and the most common score of the three audits selected.

2.4 Variables

Demographic variables were sex, age and occupa-tion. Occupation was recorded and categorised acco-rding to the Australian and New Zealand Standard Classification of Occupations [23]. Occupational categories were then further grouped into either sedentary (Managers, Professionals, Clerical and Administrative Workers, Machine Operators and Drivers, Sales Workers) or physical (Technicians and Trades Workers, Community and Personal Service Workers, Labourers) categories. Utilising the diagnosis listed on medical certificates and reports, injuries were categorised as being hip joint (inclusive of labral tears, femoro-acetabular impingement and osteoarthritis), lateral hip (inclusive of bursitis, greater trochanteric pain syndrome and gluteal ten-dinopathy) or non-specific (inclusive of non-specific diagnoses such as pain, sprain or strain of the hip).

A claim was described as having an exercise component if the worker had billing on their file for either physiotherapy or exercise-specific intervention using the billing codes. These codes are utilised by physiotherapists and exercise physiologists predominantly. A decision was made to assume that physiotherapy would include a component of exercise prescription, as exercise is considered a part of standard physiotherapy care for the management of hip pain [24]. Usual practice is that a medical referral is required prior to billing under these codes. Variables relating to exercise included time between injury and commencement of exercise, number of pre-operative and post-operative supervised exercise sessions, and duration of supervised exercise. Number of claims proceeding to surgery and time between injury and surgery were also recorded.

Injury duration was calculated by recording the timeframe (in days) between the date of injury and the date of upgrade to final work capacity. This was further categorised into five claim duration categories: 0–30 days (short duration claims that may often resolve with minimal intervention), 31–90 days (potential transition phase from acute to chronic presentation targeted in early guideline-based recommendations), 91–180 days, 181–360 days, and 361 + days (based on important time frames for claims review by insurers). Final work capacity was used as an outcome variable, based on the final ‘medical certification for work’ at the time of claim closure. This included the categories: fit for pre-injury duties, fit for restricted duties, or unfit. These responses were dichotomised to ‘achieved pre-injury duties’ and ‘did not achieve pre-injury duties’

2.5 Statistical analysis

Descriptive statistics were reported for demographic and outcome variables. For Aim 1, descri-ptive statistics were used to indicate exercise intervention characteristics for the whole cohort, and by the five claim duration categories. For Aim 2, Spearman’s rank correlation was used to analyse the association between commencement of exercise and the duration of the claim, for the whole cohort and for those who had exercise intervention. For Aim 3, descriptive statistics were used to indicate exercise intervention characteristics by the injury type. Comparisons for Aim 3 were performed with Kruskal Wallis test. Analysis was performed using Stata 15.1 for Mac. The significance value was set at 0.05.

3.1 Demographic data

Twenty seven files (Table 1) were for male workers and 17 for females. Mean age at the time the claim was made was 45.3 years (standard deviation 12.7, range 19 to 72). The cases were evenly distributed between sedentary and physical work. The median claim duration was 108 days (inter quartile range (IQR) 294), ranging from 3 days to 1149 days. Eighty four percent achieved a return to pre-injury duties at the finalization of their claim.

Exercise management occurred for 33/44 cases, meaning 25% of workers with hip pain did not receive guideline recommended exercise management. The median time to commence exercise for those cases that had exercise was 14 days post-injury (IQR = 28), with 33% not commencing exercise until beyond 4 weeks.

Table 2 presents a summary of case information relating to exercise commencement, frequency and duration for the whole cohort and claim duration categories. In the 0–30 days category, only 3/10 workers were referred for exercise, with all workers returning to pre-injury duties regardless of if they received exercise or not. The majority of workers with claim duration greater than 30 days had exercise intervention. Earlier referral to exercise was evident in injury categories with a duration of 31–90 days, and 91–180 days (first exercise visit median was 11 and 8 days respectively). Claims within the categories 181–360 days and 361 + days demonstrated progressively later median commencement of exercise intervention (22 days and 97 days respectively). Claims in the 361 + days category demonstrated a considerably higher median number of supervised exercise visits (33 sessions).

All claims of duration less than 180 days returned to their pre-injury duties, regardless of whether they had exercise intervention or not (Table 2). For those with an injury duration greater than 180 days, there was reduced numbers who returned to their pre-injury duties.

Seven cases proceeded to surgery, with the median time to first surgery being 132 (IQR = 150) days. Three of the seven cases proceeded to surgery in less than 12 weeks after injury. Only two of the surgical cases had pre-operative supervised exercise, with the duration of this exercise being 14 days and 35 days respectively.

For the whole cohort, there was a strong [25] correlation (Spearman’s rho = 0.63, p < 0.001) between time to commencement of exercise and claim duration (Fig. 1). For only those that had exercise, there was also a strong correlation (Spearman’s rho = 0.70, p < 0.001) between time to commencement of exercise and claim duration.

OPEN IN VIEWER

Fig. 1

Scatterplot diagram for the time to commencement of exercise with claim duration (n = 44).

Fifteen (34%) subjects were classified as having lateral hip pain, 16 (36%) as having hip joint pain and 13 (30%) as having non-specific hip pain (Table 3). There was a significant difference between injury type and time to commencement of exercise (Chi Square = 9.0, Degrees of Freedom = 2, p = 0.011), with the main finding being a significant delay in referral to exercise for those with a diagnosis of hip joint pain (Table 3). Fewer subjects returned to pre-injury duties in this group. It was also notable that all seven surgical cases had a hip joint pain diagnosis.

4.1 Strengths and limitations

This study utilised real-world, system-level data from recently closed workers’ compensation claims. The sample size is relatively small, particularly in relationship to claim duration categories, and therefore interpretation of the results should be approached with caution. We did not control for confounders with the potential to impact claim costs and return to work outcomes, such as psychological issues, pain severity, employer relationship and worker expectations [26, 27], and acknowledge that recovery from a work injury is complex and multifactorial. An assumption was made that physiotherapy management (indicated by billing codes) would include exercise management. When reports from physiotherapists were available (23/33 cases), exercise prescription was always documented. Furthermore, a survey conducted in Australia which evaluated physiotherapy management of hip osteoarthritis demonstrated exercise as a highly common treatment modality (used in > 80% of all cases) [28]. Therefore, it is likely that any physiotherapy intervention would include exercise as a component of a multimodal approach, but we cannot be sure. Further, we cannot make any judgement on the nature of the exercise intervention provided. Also, we cannot discount that exercise prescription may have occurred via prescription from health care professions other than physiotherapists and exercise physiologists, such as general practi-tioners.

4.2 Commencement of exercise and injury duration

For the cohort as a whole, the median time of com-mencement of exercise was 14 days with an IQR of 28 (excluding those who did not have formal exercise). Globally speaking this indicates that when exercise is prescribed, for 50% this occurs reasonably early, consistent with clinical practice guidelines for the management of hip pain [3, 10]. Interestingly though, for claims that lasted less than 30 days, exercise was only used 30% of the time, but all these workers returned to pre-injury duties regardless of whether the exercise intervention was received (Table 2). While there is limited research in this area, these findings appear to be similar to those of compensable acute low back injuries, which indicate where an injury appears to be self-resolving early exercise did not ultimately affect injury duration or outcome [29]. It may be that other guideline based management such as staying active and use of simply analgesics was used, but this is speculative. Further investigation of care pathways of those who did not receive exercise would be of value. More research is required to investigate the ideal time to commence exercise for hip pain following a work injury, which will help update hip guidelines. One question specifically would be if potential cost savings by not delaying exercise outweigh the cost with providing exercise to those that would have recovery early anyway.

Claims lasting from 31 to 90 and from 91 to 180 days had exercise as a part of first line management. Exercise typically commenced early, at under two weeks from the date of injury, and intensively, with an average of six and eight supervised exercise sessions respectively (Table 2). All claims that fell within this timeframe returned to pre-injury duties, consistent with best-practice guidelines on exercise and returning to full function [3, 10]. This might lend weight to the importance of timeliness in the delivery of exercise intervention [9, 30].

In claims lasting beyond 180 days, use of exercise was more delayed (Table 2). Why this is the case is unclear, but is worthy of additional research.

A correlation between time to commencement of exercise and claim duration does not imply causation. While it might be tempting to assume that starting exercise sooner will reduce the length of a claim, no such conclusion can be drawn from the data here. Because return to work is a complex, multifactorial premises [11] and exercise is only one element of multimodal management (exercise does not occur in isolation), a reasonable path may be to test the effect of implementation of guideline based management [3, 12] as a whole.

4.3 Commencement of exercise and diagnosis

Guidelines for exercise as first-line management of hip injuries are similar across diagnoses [3, 10]. However, in the claims reviewed here, workers with a structural diagnosis aligned to hip joint pathology experienced a delay to commencement of exercise. This suggests that the type of intervention provided to an injured worker is influenced by the diagnosis given, despite the fact that best practice guidelines provide similar recommendations across all hip injury types [1, 10]. Further investigation of the reasons behind this observation would be reasonable. There is a strong possibility this could be driven by health care provider beliefs which are known to influence clinical practice decision making for other types of musculoskeletal pain [31].

4.4 Pre-operative exercise

Of the seven cases who proceeded to surgery (all with a structural diagnoses aligned to hip joint pathology), only two had pre-operative exercise. This conflicts with recommendations for pre-operative exercise intervention [19]. Although the research around surgery on compensable cases is limited, some findings have identified that when surgery is performed on a compensable worker, the risk of a negative outcome is doubled compared to non-compensable patients [13]. Given the poorer surgical outcomes associated with compensable injuries, adhering to clinical practice guidelines would appear even more pertinent.

4.5 Practice recommendations

While acknowledging the limitation of the study, there seem to be two key recommendations linked to the findings. Perhaps the strongest of these is that there is a need for education of health care practitioners regarding current clinical practice guidelines in regard to use of exercise early as first-line management for hip pain, regardless of the diagnostic category. Secondly, given the association identified between time to commencement of exercise with claim duration, and the pattern of referral categorised by claim duration, exercise should commence in the first two weeks of the claim being made. Additional cost associated with sending people for exercise who may have recovered without it (seen for those with claim duration less than 30 days) may be negligible compared to the potential savings in reducing the length of longer term claims. Also, this needs to be countered against a potential risk of over medicalisation of a claim when they are referred for exercise early if they might not have needed it. Of course, whether earlier referral to exercise does in fact reduce claim duration needs to be tested.