Description of Model Structure
The SDM presented in Fig. 1 represents the causal loop structure of the organizational work disability systems that emerged from model building sessions. As expected, the SDM was multidimensional, including attitudinal characteristics of the individual (i.e., motivation to RTW, and preparedness to RTW, and fulfillment of role demands outside of work), health factors (i.e., functional health status, and performance of work tasks), social factors (i.e., perceived workplace support, quality of communication between supervisor and worker, and co-worker and supervisor pressure to RTW), and organizational components (i.e., work disability management policies, and revenue loss). The polarity between related variables was also established during model building and is depicted in the figure using ‘+’ (denotes that components change in the same direction), and ‘−’ (denotes that components change in opposing directions). Consistency between the models obtained in the two organizations enabled the depiction of one common model. Several key features of the model will be described in this section.
First, findings from model building sessions showed that the likelihood of RTW within each organization was influenced by two primary stock (level of outcome) and flow variables (rate of change of outcome). Indicated by their positive polarity, greater levels of functional health status (+) and preparedness to RTW (+), increased RTW likelihood. Stakeholders also identified causal pathways between the level of RTW preparedness, and several factors. Indicated by the negative polarity, increases in supervisor pressure to RTW (−) and role demands outside of work (−) resulted in lower preparedness. Quality of supervisor-frontline worker communication (+), and co-worker social support (+) were components that had an opposing impact, and increased RTW preparedness.
The model building exercise also revealed that quality of communication was increased by greater supervisor positivity (+), frequency of RTW conversations (+), and amount of information shared regarding work injury (+). Additionally, the length of absence was linked to several model components including role demands outside of work (+), coworker adjustment to workplace injury (+), and supervisor pressure on an injured worker to RTW (+).
Despite similarities in terms of components and feedback relationships uncovered through model building, each company implemented unique policies (depicted as red arrows in Fig. 1) to manage work disability-related costs and facilitate RTW. Company A had a financial incentive, offering a $60,000 annual bonus to be divided amongst all workers. When a work injury occurred, money was deducted from the bonus pot to pay for immediate medical care (e.g., emergency room visit, and initial treatment). Within the specific organizational context, the policy was intended to prevent workplace injuries, incentivize safety behaviors, and minimize short-term health care costs. Modeling sessions identified a causal pathway between the bonus pot and pressure frontline co-workers (+) and supervisors (+) placed on injured workers, suggesting that the policy may have had an unintended impact on the workplace social climate.
In comparison, Company B offered light duty. Work disabled employees that were medically cleared for adapted tasks, were found temporary roles that fit their activity limitations. Within the organizational context, light duty aimed at facilitating early work reintegration and minimizing workers compensation costs. Findings from model building uncovered a causal pathway between the presence of light duty and the ability to perform work tasks (+). At the same time, light duty was also related to increased pressure frontline co-workers placed on injured workers to RTW at full duty (+).
Next, using the SDM designed in the participating organizations, simulation scenarios were conducted to determine how their unique work disability-related policies influenced the RTW process. The model simulated system behavior over a time period of 24 weeks to capture both simple and prolonged work disability cases. Simulations were examined with respect to their impact on percentage of RTW preparedness (0 = no RTW preparedness; 100 % = completely prepared to RTW) and percentage of RTW likelihood (0 = no likelihood to RTW; 100 % = completely likely to RTW) which were used as proxies for the overall performance of the RTW process .
In Company A, where a bonus was provided as an incentive to prevent work injuries, simulations were conducted to compare the current (base case) company-wide bonus ($60,000) to increased ($90,000), decreased ($30,000), and no bonus ($0) scenarios. The simulation presented in Fig. 2a showed that RTW likelihood trajectory was initially low (0–4 weeks), followed by a rapid linear increase, and than a plateau (8 weeks). At first (0–6 weeks), few differences existed between the different bonus levels and the likelihood of RTW (range 39–41 %). At 12 weeks, the differences between the scenarios became apparent. In contrast to what was expected, no bonus (66 %) and reduced bonus (64 %) scenarios exhibited higher RTW likelihood when compared to the current policy (59 %) and increased bonus (57 %) scenarios. The simulation conducted in Fig. 2b found that RTW preparedness increased logarithmically over the time period (Fig. 2b). Few differences existed between the different bonus levels and RTW preparedness between 0 and 6 weeks (range 29–31 %). At 12-weeks, no bonus (39 %) and reduced bonus level (37 %) scenarios exhibited higher RTW preparedness compared to the base case policy (34 %) and increased bonus scenario (33 %). Differences in bonus levels on RTW preparedness plateaued at the 12-week time point, and persisted over the course of the simulation.
In Company B, where light duty was provided to facilitate early work reintegration, the availability of full light duty (base case) was compared to partial and no light duty. The simulations showed that RTW likelihood (Fig. 2c) and RTW preparedness (Fig. 2d) increased logarithmically over the time period. Indicating it’s usefulness to RTW, at 6 weeks, full (46 %) and partial light duty (44 %) availability exhibited a greater likelihood of RTW, compared to no light duty availability (37 %). In more prolonged cases (12 weeks), little difference existed between full (54 %), partial (53 %) and no light duty (53 %) on the likelihood of RTW. When examining RTW preparedness in Company B (Fig. 2d), findings showed that at 6 weeks few differences existed between full (21 %), partial (20 %) and no light duty (22 %) scenarios. At 12 weeks, the no light duty scenario exhibited slightly higher RTW preparedness (27 %) compared to partial (23 %) and full light duty availability (25 %). Differences in the provision of light duty and RTW preparedness also plateaued at 12 weeks and persisted over the remaining 12 weeks of the simulation.