Factors associated with sickness certification of injured workers by General Practitioners in Victoria, Australia

The state of Victoria in Australia had a working population of approximately 2.8 million as at June 2011 [20]. Employers in the state are required to maintain workers’ compensation insurance through the WorkSafe Victoria (WSV) unless they are able to self-insure, obtain insurance through the national workers’ compensation scheme, or if they are a sole trader. The WSV system provides coverage for approximately 85 % of the Victorian labour force. All injuries and illnesses that exceed the pecuniary threshold for healthcare expenses or have required more than 10 days work absence are required to be lodged with the WSV via one of six private insurers.

The Victorian workers compensation system requires production of a medical certificate in order to accept a compensation claim. Certificates can be submitted by GPs and physical therapists or by hospital-based medical practitioners. The medical certificate contains information that include the practitioner’s recommendation regarding fitness to work (UFW, ALT, fit for work) and the start and end date of the certificate [16]. There are statutory limits for the duration of UFW certificates defined in the state’s workers compensation regulations. Initial medical certification for a workers compensation claim can be of up to 14 days duration whilst subsequent certificates can be of up to 28 days duration.

This study was a retrospective population-based cohort study, for which the authors accessed the Compensation Research Database (CRD) established at the Institute for Safety Compensation and Recovery Research (ISCRR) at Monash University, Melbourne, Australia. The CRD contains de-identified case-level administrative data received from the WSV between years 1986-2012 [21, 22]. The CRD only contains details of sickness certificates issued for injuries sustained in the workplace, as periods of sick leave caused by pre-existing non-work related health problems are not recorded.

More detailed information on this dataset is provided elsewhere [16].

All data for accepted compensation claims lodged by working age adults (15 - 65 years) with a date of injury between 1 Jan 2003 and 31 Dec 2010 were extracted from the database (n = 217,076). Claims were excluded if:

In this study only the initial sickness certificates were included in the analysis, since in this database information recorded about subsequent certificates may be incorrect or missing. Sickness certificates of all individual claimants were organised into two pre-defined categories: (1) UFW certificates where GPs recommended a complete absence from work (n = 91,726) and (2) certificates where the GP recommended a RTW with ALT duties (n = 28,174).

Following several consultations within the research team, which included GPs, six categories of the most frequent worker conditions (injuries and diseases described by the Type of Occurrence Classification System (TOOCS) Third Edition (http://​www.​safeworkaustrali​a.​gov.​au/​sites/​SWA) to code injury and disease types) for issuing sickness certificates were included in the analysis: (1) fractures, (2) musculoskeletal diseases (MSD), (3) other traumatic injuries, (4) back pains and strains, (5) mental health conditions (MHC) including work-related stress and post-traumatic stress disorders, and (6) other diseases [16]. The TOOCS system is designed to code both injuries and diseases, and identifies the most serious injury or disease reported on the initial claim for workers’ compensation and allocates an appropriate code from the Nature of Injury/Disease Classification. If more than one injury or disease is reported, the most serious injury or disease that is likely to have the most adverse effect on the worker’s life is selected [16].

Both univariate and multivariate logistic regression analysis was performed to assess the impact of a number of factors on the likelihood that an injured worker would receive an ALT certificate from their GP. In the present study, the model predicted ALT (i.e. ALT certificate was set as 1 and UFW as 0). The model consisted of demographic (age group, gender, residential location), occupational (occupation group and employer segment size), industry type, medical (GP caseload of injured workers), injury type and socio-economic factors each with two or more levels (see Table 1). Employer segment size is based on the employer’s annualised remuneration and is grouped into small - < $1 M, medium $1 M - $20 M, large - > $20 M and government.

All factors had statistically significant contributions and were added to the multivariate model (Table 3). For the univariate analyses, all cases were included except for the Socio-Economic Indexes for Areas (SEIFA) [23] variable, which was missing for 241 cases. In the multivariate model these 241 (0.2 %) cases from the SEIFA variable were removed. The final sample for the multivariate model included 91,541 UFW cases and 28,118 ALT cases.

Cox and Snell [24] and Nagelkerke [25] pseudo R² provides an indication of how well the fit of the model is relative to a ‘null’ model with no risk factors. The Nagelkerke R² allows for the R² to potentially reach 1.0, a correction to Cox and Snell that do not allow this [26].

A total of 119,900 claims with initial sickness certificates were included in this study. A descriptive summary of the variables is provided in Table 2 which outlines the number and proportion (%) of sickness certificates within each risk factor. The majority of the injured workers in both ALT and UFW categories were men, mostly between 45-54 years of age. Nearly half (49.9 %) of injured workers with UFW and 36.9 % of ALT certificates suffered from MSD. The most common occupation in the study sample was labourer, the most common industry – manufacturing, and the most common location of all injured workers was the metropolitan area of the state capital city.

Table 3 summarizes the contributions of each risk factor in the univariate and multivariate model. Univariate analysis (step 1) for all nine category variables was conducted to identify significant individual predictors. The nine category variables were then added into the multivariate model (step 2).

The full multivariate model containing all nine category variables (inclusive of the variables within each category) was statistically significant, X ² (52, N = 120,186) = 8636.976, indicating ability to distinguish between injured workers who receive an ALT and UFW certificate. The model explained between 7 % (Cox and Snell R Square) and 10.5 % (Nagelkerke R Square) of the variance in certificate type.

Compared to younger workers aged 15-24, there was a significantly reduced likelihood of workers in the 55-64 age-category receiving an ALT certificate from their GP OR = 0.86 (by 14 %), (95%CI, 0.81 – 0.91). Compared to men, women were at a slightly increased (0.62 %) chance of being issued ALT certificates, OR = 1.06 (95%CI, 1.02 – 1.10).

Taking other variables into account, worker condition was a significant risk factor. Table 3 shows that workers with MSD, OR = 1.89, (95%CI 1.79 – 1.99), and other diseases, OR = 3.32, (95%CI, 3.11 – 3.54), were three times more likely to receive an ALT certificate than those with fractures, whereas workers with MHC, OR = 0.25, (95%CI, 0.22 – 0.27), were less likely to receive an ALT certificate than those with fractures, MSDs and other diseases.

Worker’s area of residence was also an important risk factor. Compared to workers from metropolitan areas, there was a significantly reduced likelihood of injured workers from a rural area, OR = 0.91, (95%CI, 0.87 – 0.94), and interstate, OR = 0.95, (95 % CI, 0.91-1.01) receiving an ALT certificate from their GP.

An analysis of GP caseloads showed that GPs with the highest case load (i.e. 49 and more claims per provider over the eight year period), OR = 1.16, (95%CI, 1.11 – 1.20), were more likely (by ~16 %) to issue an ALT certificate to an injured worker than those GPs who saw less than 13 injured workers over eight years.

In terms of worker occupation, compared to managers, only professionals, OR = 0.83, (95%CI, 0.76 – 0.91) and community and personal service workers, OR = 0.80, (95%CI, 0.73 – 0.87) were significantly less likely (by 17 and 20 %) to receive an ALT certificate.

Employer segment size was a significant risk factor associated with an ALT certificate. Workers from medium (OR = 1.38 by ~38 %), (95%CI, 1.33- 1.43), large (OR = 1.86 (by ~86 %), 95 % CI, 1.78-1.94) and government size organizations (OR = 1.24 (by ~24 %), 95 % CI, 1.14-1.34) were more likely to receive ALT certificates than those from small organizations.

When considering industry, injured workers from mining, OR = 1.53, (95%CI, 1.18 – 1.97), manufacturing, OR = 1.54, (95%CI, 1.39 – 1.71), wholesale trade, OR = 1.49, (95%CI, 1.33 – 1.66), professional scientific and technical services, OR = 1.37, (95%CI, 1.19 –1.58) and other not elsewhere classified industries, OR = 1.34, (95%CI, 1.20–1.51) were significantly more likely (by ~40 % - 50 %) to receive an ALT certificate compared to injured workers from the agriculture, forestry and fishing industry. Taking other variables into account, SEIFA was not associated with ALT certificate at all (Table 3).

The results of the current study clearly indicate that older workers, those with MHCs and those living rurally are more likely to receive UFW certificates than workers with physical injuries, workers living in metropolitan areas and workers visiting GPs with a higher injured worker case load. The latter are more likely to receive an ALT certificate. It is yet unknown why certain factors are associated with ALT certificates; however assumptions can be made based on existing literature, which show that older workers are less likely to RTW because they may have childcare and family responsibilities, are closer to retirement and may recover more slowly from an injury because of age and other existing health issues [27‐29]. Older workers (between the age of 55 and 64 years) also seem to have more difficulty adapting to modified duties [11, 30]. In contrast, younger adults have been shown to have more favourable employment outcomes after injury [4, 12, 31].

We also found that workers suffering from physical injuries and other diseases were more likely to receive ALT certificates than workers with MHCs. It could be that GPs are more inclined to recommend modified duties and earlier RTW to such workers with physical conditions because they are familiar with interventions and type of modified duties available at workplaces that would be appropriate for such conditions [32]. Moreover, there is still a stigma associated with MHC and health professionals may perceive injured workers with mental illness as having poorer health outcomes than they really have [16, 33]. Studies also show that when it comes to MHC claims GPs grapple with issues such as diagnostic uncertainty, conflicting medical opinions, poor communication between professionals and secondary concerns related to pain management, lack of motivation by the injured worker to RTW and sourcing appropriate care services [34‐36]. It is also possible that accommodations for MHC are absent in workplaces and as such GPs may be reluctant to suggest a return to work.

In terms of occupation, manual workers are less likely to receive ALT certificates than managers. This suggests that working on alternate or restricted duty appears to be a viable option mainly in managerial positions, whereas manual labour occupations have been associated with more severe disabilities of longer duration, probably associated with UFW rather with modified duties [37, 38]. On the other hand, research also shows that occupation does not determine the type of sickness certificate [39], and that may be why the odds of receiving ALT certificate across other occupations are very similar (Table 3).

As opposed to the findings reported by Shiel et al [15], demonstrating that GP and general practice factors had no significant impact on likelihood of a ‘may be fit’ note being issued, we found that those workers who see GPs with a higher caseload of injured workers are more likely to receive ALT certificates. This suggests that GPs with higher caseload of injured workers are familiar with the workers’ compensation system, have a positive attitude towards RTW and modified tasks and therefore more likely to recommend ALT duties [40‐42]. This finding also suggests that in order to achieve improved certification (i.e. higher proportion of ALT certificates) systems may want to steer injured workers towards more “experienced” GPs.

Employer segment size stood out as an important risk factor associated with ALT certificate. Injured workers from large enterprises were nearly twice as likely to receive ALT certificates as those who work for small size organizations. This corresponds to previous findings [12, 37] that showed working for larger companies was positively associated with return to work. Larger organisations are able to employ specialists in disability management [43], provide more information about modified duties and RTW and have greater flexibility in allowing workers to return to modified jobs [37]. Larger workplace size has been associated with a shorter duration of absence following a physical work injury because of an increased ability of larger workplaces to offer accommodations or alternate duties [44].

In terms of industry, workers from mining, manufacturing, electricity, gas, water and waste as well as wholesale trade industries are more likely (up to 50 %) to receive ALT certificates than workers from agriculture, forestry and fishing. Literature on industry as a predictor of RTW is scarce; however it is known that being a blue collar worker (i.e. performing manual labour) is associated with longer duration off work when compared to those workers who perform professional jobs [11]. While physically demanding occupations and employment in goods producing industries have been associated with slower RTW for physical injuries [45], studies on mental health claims have reported longer duration off work in government and educational industries compared to other industry sectors [46].

To the best of our knowledge, this is the first study that explores the factors associated with the type of sickness certificate issued by a GP in Australia. In this study we were able to examine almost all the predicting factors previously reported in the literature.

There are several limitations to our analyses. First, in this study we analysed the initial sickness certificates only. Consequently, we could not ascertain for how long UFW certificates were issued and when (and/or if) the changeover to ALT certificates occurred, thus facilitating RTW. Second, we were unable to analyse other important factors, such as comorbidities, a previous history of sickness certification, expectations of sickness absence and motivation as this information was not available from the data collected. The opportunity to include these explanatory variables would have increased the robustness of the model. Finally, data from administrative datasets are subject to entry errors, miscoding and misclassification, which we could not control for.

It is known that extended periods of sickness can negatively affect injured workers, their family, employers and lead to increased compensation schemes. Workers might have poorer health outcomes and require an increased number of health interventions, which are associated with higher compensation costs [47-49]. From a policy perspective, this study suggests that efforts to target specific groups of injured workers (i.e. older age, workers with MHCs in rural areas) and employers (e.g. smaller companies) could increase the awareness of benefits of modified and alternate duties and facilitate RTW for groups that are otherwise less likely to RTW.

Monash University Human Research Ethics Committee, Melbourne, Australia.

Access to the CRD is publicly available for researchers to use, under strict guidelines approved by the compensation authorities and the Monash University Human Research Ethics Committee. Information about the CRD data can be found at http://​www.​iscrr.​com.​au/​evidence-data-and-research/​using-data/​compensation-research-database-crd. For further information on this database, or to request a data extract for research, please review the ISCRR data access policy and email CRD@iscrr.com.au.