Using a population-based cohort of patients who were discharged from 10 IPR facilities in Alberta between 2014 and 2016, we found that only 36.2% of the stroke patients used ESD/OPR services during one year after IPR discharge. Overall, use of ESD/OPR services was lower among female and older patients. Patients in the Calgary zone had the lowest number of visits compared with those in Edmonton or other zones. Occupational/physical/physiotherapy was the most used service and there were no differences in types of ESD/OPR services between urban and rural residents, except for visits to social workers. Regression analyses indicated that factors such as age, health zone, acute care LOS of the associated acute stroke episode, BMI, and stroke position, were consistently associated with ESD/OPR service use. The MLR model appeared to be a better fit for our data, with a higher R-squared value than in the MNB model (12% versus 2%, respectively).

The Canadian Stroke Best Practice Recommendations indicate that a patient is a suitable candidate for OPR if the patient’s rehabilitation needs can be met in the community, the patient meets the general inclusion criteria for stroke rehabilitation (as described in the Canadian Stroke Best Practice Recommendations document [4]), the patient is medically stable, ready to participate in rehabilitation, can be accompanied by a caregiver to the therapy sessions if necessary, and can organize transportation to and from the rehabilitation center [4]. While we did not assess factors associated with whether a patient receives ESD/OPR services, patients who did not receive any ESD/OPR services in our study sample may not have been suitable candidates for it based on their medical condition, the patients’ rehabilitation needs could not be met in the community, and/or the patients may not have had the required assistance (if needed) for participation in ESD/OPR.

Studies examining the factors that determine transitions from IPR to community/outpatient stroke rehabilitation among stroke patients, including which patients get referred to ESD/OPR and which referred patients attend the therapy program, are scarce. Sandel et al. 2009, studied the demographic, socioeconomic, and geographical disparities in access to a variety of post-acute stroke rehabilitation services (including inpatient rehabilitation hospital [IRH], skilled nursing facility [SNF], home health care [HH] and outpatient, or no rehabilitation services) during the year after stroke in the United States, but did not specifically examine the factors associated with the receipt/number of visits of outpatient stroke rehabilitation following receipt of IPR services. They found that the percentage of individuals in the SNF and HH categories as the highest utilized post-acute care service categories decreased between 1996 and 2003, while the percentage of individuals in the outpatient services category as the sole post-cate care treatment increased over time [28]. Freburger et al. [29] also found demographic differences in post-acute rehabilitation care (that is, receiving HH versus no HH among those discharged to home, and receiving IRH versus SNF among those discharged to institutions) among patients in selected states in the United States, even after controlling for factors such as illness severity, comorbidities, and supply. More specifically, they found that Blacks, women, older individuals, and lower income individuals were more likely to get discharged to an institution versus home, while Hispanics and the uninsured were less likely to receive institutional care. Conditional on being discharged home, racial minorities, women, older individuals, and lower income individuals were more likely to receive HH than no HH, while the uninsured were less likely to receive it. Chan et al. [30] used the same cohort used by Sandel et al. to examine disparities associated with the number of outpatient rehabilitation visits during the year following discharge from acute care. Similar to our findings, Chan et al. found age to be negatively associated with the number of outpatient visits and the acute care LOS to be positively associated with the number of visits. However, the study did not specifically examine the number of outpatient rehabilitation visits among those who had received IPR services. They also did not include patient comorbidities in their analysis, which were found to be highly correlated with the number of ESD/OPR visits in the present study. Further, we used two different models to examine the factors associated with the number of ESD/OPR visits and found consistent results. It should be noted that studies by Sandel et al., Freburger et al., and Chan et al. used United States data, so there may be variations between the results of these studies and the present study due to the differences between the United States and Canada healthcare systems as reported previously [31].

Janzen et al. performed a retrospective chart review of a cohort of 1,497 stroke patients who were from an IPR facility between 1 January 2009 and 1 March 2016 within the Southwest Local Health Integration Network geographical boundaries in Ontario [13]. Among these patients, 891 were referred to an OPR program, and 721 of these attended the program. Those who were referred were significantly younger, had higher FIM scores at discharge, and had shorter IPR LOS compared with those who were not referred. Also, most of the referred patients were discharged home following IPR. Patients who attended the program (that is, patients who received the OPR therapy) were, again, significantly younger and had higher discharge FIM scores but did not have a significantly different stroke severity compared with those who did not attend the program. In addition, among patients who received OPR therapy, the average number of visits was 32.2 (standard deviation: 26.2). Our study differs from Janzen et al. in that we examined the factors that were associated with the utilization of ESD/OPR services (by those who received any service). We also looked at a wider range of patient factors than Janzen et al., including more detailed patient characteristics (specifically for predicting the frequency of visits). Unlike Janzen et al. who found that getting referred to OPR services and/or receiving any OPR services was significantly associated with the discharge FIM score and IPR LOS, we did not find any of those factors to be significantly associated with the utilization of ESD/OPR services received. Instead, we found the LOS of the associated acute stroke episode to have a positive and significant effect on the number of ESD/OPR visits. However, low R-squared value of 12% suggests that there could be other factors that we did not observe in the data and hence could not control for. Further research and richer data on patients’ clinical and socio-economic characteristics are needed to fill this gap.

Optimal allocation of healthcare resources between acute care and rehabilitation, and among segments of rehabilitation including inpatient rehabilitation, OPR, and ESD is another key challenge for healthcare planners in responding to the increasing demand for provision of care to stroke survivors [32]. Yan et al. used a stroke rehabilitation optimal model, combining discrete event simulation with a genetic algorithm, that changes care capacity across segments of rehabilitation to identify an optimal solution for minimizing wait times in each segment in Alberta. Their model predicted that if ESD and OPR could be provided to additional 138 and 262 stroke survivors, respectively (compared with the status quo), it would result in cost savings of $25.45 million annually [32].

Another challenge in the delivery of post-stroke rehabilitation services is in their delivery to patients residing in rural settings as they have been shown to have decreased access to healthcare, including rehabilitation services, compared with those residing in urban areas [33, 34]. This may partly explain our finding regarding a higher utilization of ESD/OPR services among urban residents than rural residents. Allen et al. suggested providing home-based specialized rehabilitation services for rural residents as a potential solution to this problem and found that providing this service will result in functional gains for rural resident comparable to those living in urban settings [35]. However, successful home-based rehabilitation partly depends on effective communication and collaboration between the caregiver, patient, and therapist [36, 37]. Fisher et al. also suggested that (a) developing strategic networks can help understand the needs of these patients at an organizational level and (b) the existing gap in skill mix and staff establishment among teams providing rehabilitation services is one reason for the unmet needs of patients with more severe disabilities [38].

Although this study contributes to knowledge of the clinical and socio-demographic factors associated with utilization of ESD/OPR services following discharge from IPR among stroke patients, it has limitations. The administrative data sets did not include several clinical data elements of the acute stroke episode [18]. These clinical data, such as the severity of specific impairments (that is, ataxia or aphasia) which may not be fully captured in the FIM score, can be associated with increased LOS at IPR [39]. We expect these factors to contribute to the number of ESD/OPR visits following discharge from IPR as well. Thus, more detailed information about the patients’ clinical characteristics could facilitate better understanding about the association between those factors and the utilization of ESD/OPR services received.