Employment sustainability after return to work among Japanese stroke survivors

As far as we know, there are no other workforce-based Japanese stroke survivor studies after RTW that have examined the frequencies of work continuance, RSA, and resignation, or analyzed the predictors of work discontinuation using survival analysis. This study showed that about 60% of stroke survivors continued to work for 5 years (with the help of RTW support systems) after their RTW. The stroke survivors in this study were supported by the occupational health center, e.g., with a part-time work system, work accommodation, and OP interviews. While there have not been any workforce-based studies examining the work continuance rate after RTW among stroke survivors, we speculate that the work continuance rate of stroke survivors would be lower in smaller companies than the figure reported in this study. Employees working in smaller companies seem to have less ‘protection’ in terms of the capacity of their employer to make ‘reasonable adjustments’ for them to keep working, for example, their ability to conduct OP interviews, ensure a gradual RTW, and transfer the employee to other tasks (Moriguchi et al. 2010). As a future task, we have started to collect data about stroke survivors who returned to work in small- or medium-sized companies, to allow us to compare work continuance rates between companies.

The distribution of work discontinuation due to RSA or resignation observed in this study was similar to the distribution of SA, but Hensing reported that SA data were heavily skewed (Hensing 2004). The Kaplan–Meier curve of the current cohort showed that the work continuance rate declined over time after the employees returned to work. Work discontinuation (i.e., RSA or resignation) was most common in the first year, followed by the second year. The survival curve for RSA among the stroke survivors increased gradually, which was different from the curve for RSA due to depression in the same population (Endo et al. 2013). According to the Kaplan–Meier curve obtained in this study, it might be necessary for occupational health professionals to provide stroke survivors with careful support for 5 years.

Our study showed that older age was significantly associated with the time to work discontinuation (RSA or resignation). The results of our study are consistent with those of previous studies (Perk and Alexanderson 2004; Petty et al. 1998; Saeki et al. 1993; Vyas et al. 2016). For example, older age has been found to be associated with an increased risk of stroke recurrence (Petty et al. 1998), and older stroke survivors might resign more frequently than younger survivors because the period until their retirement is shorter. Vyas et al. reported that older age was associated with a lower likelihood of employment in stroke survivors (Vyas et al. 2016). In studies of RTW rates, older age was demonstrated to be a predictor of a reduced likelihood of returning to work (Perk and Alexanderson 2004; Saeki et al. 1993).

On the other hand, there were no significant associations between work discontinuation and other factors, such as sex, stroke subtype, or the duration of SA. As stroke survivors have various symptoms, which also vary in severity, in future studies we need to use larger samples and have access to clinical and work-related information to allow us to adjust for confounding factors.

In this study, the RSA rate at 1 year after the employees’ RTW was 16.3%, which was relatively consistent with the risk of recurrence reported in previous clinical stroke studies (between 7.1 and 21%), while almost half of RSA were caused by recurrent strokes (Mohan et al. 2011). The RSA rate at 5 years was 33.4%, which was consistent with the risk of recurrence in previous studies (between 18.3 and 35.3%) (Mohan et al. 2011). Cabral’s study reported that the overall risk of stroke recurrence was 9%, which was similar to those described in other cohort studies (Cabral et al. 2015).

Two employees experienced RSA due to AMI, which does not seem to be a high number, while previous studies reported that stroke survivors were at higher risk of AMI (Boysen and Truelsen 2000; Talelli and Greenwood 2008).

In this study, 20.9% (18/86) of RSA were caused by mental disorders. Previous studies have reported that mood and emotional disturbances are common symptoms in stroke survivors (Ibeneme et al. 2016; Kim and Choi-Kwon 2000). Mijajlovic reported that strokes worsen cognitive function by inducing post-stroke dementia (Mijajlovic et al. 2017). Balasooriya-Smeekens’s qualitative study demonstrated that most impairments suffered by stroke survivors are “invisible”, e.g., they include fatigue, problems with concentration, memory, and personality changes (Balasooriya-Smeekens et al. 2016).

In this study, 10.5% (9/86) of RSA was caused by fractures, especially fractures caused by incidents involving steps at a train station or the employee’s house. Fracture-induced RSA are more common among stroke survivors than among employees who RTW after depression-induced SA (Endo et al. 2013). Huo reported that bone fractures often occurred in stroke survivors (Huo et al. 2016). Previous stroke survivor studies stated that stroke survivors exhibited lower physical capacities in terms of their ability to walk, use stairs, lift objects, bend, reach, and grasp, and a greater prevalence of activity limitations (Skolarus et al. 2014). We suggest that staggered working hours and home-based teleworking are desirable for reducing the risk of fractures during commuting. It would be better for employers if they ensured that that their workplaces were barrier-free, e.g., if they introduced handrails on both sides of stairways, non-slip floors, and eliminated steps where possible, as this would facilitate the RTW of stroke survivors.

First of all, the strengths of the present study included the fact that it was the first large-scale Japanese RTW study of stroke survivors (it involved approximately 300 Japanese employees who returned to work after their first stroke-induced SA). Second, the follow-up rate after RTW was 100%. This meant that this study was not affected by bias associated with loss to follow-up or subject selection, which might have affected other studies. Third, the registered data were based on objective measurements of SA (OP confirmed the clinical validity of the physicians’ certificates and registered any SA using the ICD-10 system). As previous studies of SA were based on self-reported data, the current data would be expected to have greater validity and reliability. As for ICD-10 codes, while many studies were examining ICD-10 codes I60-69 for registering stroke, McCormick et al. reported that ICD codes should not be used, because this broad group of codes, which includes the codes for non-acute and ill-defined cerebrovascular disease, and the late effects of stroke (McCormick et al. 2015). However we registered only “cerebral infarction (I63)”, “cerebral hemorrhage (I61)”, and “subarachnoid hemorrhage (I60)”; based on a physician’s certificate, for stroke as a primary outcome, that might mean our study had better validity. Fourth, the maximum follow-up period was 12.3 years.

However, this study had some limitations that should be taken into account when interpreting the results. The validity of the diagnostic information was one of the major limitations of this study. While OP confirmed the content of the physicians’ certificates, neither clinical data nor imaging findings were used to validate the stroke diagnoses. It remains unclear how we could validate physician-certified causes of RSA during the follow-up period, such as recurrent strokes (57%), fractures (10.5%), or mental disorders (20.9%), especially as the latter are extremely hard for physicians without any psychiatric expertise to diagnose (Davis et al. 2016). In general, the use of administrative data, such as physicians’ certificates as well as ICD-10 information was problematic and this was, moreover, the case in stroke (Tirschwell and Longstreth 2002).

It was not possible for us to use clinical information, e.g., about the stroke site, the impairments caused by the subjects’ strokes, or the severity of the subjects’ disabilities. Harris reported that while the stroke location does not have a strong influence on employment outcomes, functional disabilities, which can change over time, can have a significant impact on them (Harris 2014; Saeki and Hachisuka 2004). In this study, while we assumed that stroke-induced functional disability might result in a longer duration of SA, the multivariate analysis did not detect a significant relationship between the duration of SA and work discontinuation. The work continuance rate among stroke survivors can be affected by clinical factors, while previous clinical studies reported that the recurrence rate was not influenced by clinical parameters (Elneihoum et al. 1998; Kulesh et al. 2011; Mohan et al. 2011). In future studies, the effects of clinical factors on the work continuance rate of stroke survivors after RTW should be evaluated. Second, as the data of this study were mainly collected from large-scale companies, which raises a question about the representativeness of the stroke survivors in the current sample. Employees working in smaller companies seem to have less ‘protection’ in terms of the capacity of their employer to make ‘reasonable adjustments’ to accommodate their RTW (Moriguchi et al. 2010). Thus, caution must be taken when generalizing these results to other populations or other countries. Third, while resignation after RTW was evaluated as work discontinuation in this study, some subjects stopped working for the company being studied, but continued working elsewhere. Thus, it is possible that the index of work discontinuation has been overestimated. Fourthly, the duration of the follow-up period depended on when the stroke-induced sick leave started. This study included stroke survivors that suffered strokes between 2000 and 2011, and the follow-up period ended in December 2012. Thus, not all of the survivors could be followed for more than 1 year. In fact, only the cohort that suffered strokes between 2000 and 2007 could be followed for 5 years or more.