Chapter 6 – Discussion and Conclusions
6.1 Summary
This thesis was motivated by concerns that non-trivial numbers of patients who
experience a stroke in Ontario are not getting the rehabilitation they need, while others may be getting rehabilitation that is inappropriate for their needs. However, the objective was to go beyond a simple demonstration of inequity and to offer new ways of thinking about stroke system evaluation and novel tools to support future system planning. Using the literature identified in Chapters 2 and 3, refined criteria were proposed for identifying candidates for both inpatient and in-home rehabilitation. In addition to being useful to clinicians making decisions about referral to post-stroke rehabilitation, these criteria may also be useful to policy makers and health service providers developing regional plans for stroke rehabilitation systems. To demonstrate this potential, the
subsequent chapters (4 & 5) built on these refined criteria to assess the equity of access to inpatient rehabilitation across Ontario, and to propose novel ways of testing the
relationship between rehabilitation resource availability and discharges to inpatient rehabilitation after stroke. As hypothesized, the results demonstrated significant challenges faced by Ontario’s stroke rehabilitation system.
One particular challenge in planning and evaluating rehabilitation systems is
identification of the need for services within a given population. Unlike many acute conditions, rehabilitation need is difficult to measure objectively and is often seen as non- urgent. Therefore, historical utilization rates do not necessarily correspond with
rehabilitation need. Research in this area has typically focused on professionally defined need by assessing the factors most frequently used by clinicians during patient discharge, which are usually studied through direct survey or indirect observation.1 While this
provides important insight into clinical judgment, studies of these factors generally fail to account for biases in patient selection or the context in which decisions are being made. Clinicians making decisions about referral to rehabilitation may rely on traditional selection criteria that have little to no bearing on outcomes. This was demonstrated in Chapter 2 by the large number of variables that have been frequently explored in multi- variable models without proving to be significant predictors of functional outcome. Furthermore, studies of clinical judgment in environments where inpatient rehabilitation is in short supply may see clinicians refer more severely impaired patients to nursing homes (or long-term care) out of necessity, not because it is best for the patient.2 To overcome these limitations, Chapter 2 attempted to identify scientifically-confirmed need for post-stroke rehabilitation by focusing on variables that have been demonstrated to show an independent association with post-rehabilitation functional independence, one of the primary objectives of post-stroke rehabilitation.
Years of research into predictors of functional outcomes after post-stroke rehabilitation have led to a substantial amount of literature on the topic. Despite the restrictive inclusion criteria used in Chapter 2, a total of 27 studies reporting 63 multilevel models were identified and, in these models, only a few variables were found to frequently predict functional outcomes. Broadly speaking, the most influential variables fell within the following five general categories: age, initial stroke severity, functional level on acute discharge, cognitive function on acute discharge and history of previous stroke. Each of these constructs can be measured in different ways and more research is required to zero in on the most appropriate measures. Still, the results of Chapter 2 should be helpful in refining future work and suggest that any decision-making algorithm (whether for clinical use or system evaluation) should include at least one variable from each of the five
identified categories.
Unfortunately, less research has been devoted to the identification of variables that predict functional outcomes after in-home rehabilitation. While one strength of the methods used in Chapter 3 was the focus on team-coordinated and delivered ESD
programs (which have been shown to be the optimal model for ESD delivery3) the available literature dictated a focus on variables used in selection for ESD rather than those associated with improved outcomes. Although these studies used a large number of diverse selection criteria, some interesting similarities were identified that should be useful to clinicians and policy makers looking to identify ESD candidates in the future. Not surprisingly, programs generally sought patients with mild-to-moderate functional deficits and potential to improve. However, many studies also noted cognitive deficits as an important consideration, for reasons of both safety and potential to participate in rehabilitation. Furthermore, numerous studies cited pragmatic concerns such as
proximity to the hospital and the suitability of the home environment when considering appropriateness for ESD. These findings have considerable policy relevance for large jurisdictions like Ontario and highlight the need for systems of care that account for regional context.
With a better understanding of predictors of functional independence from Chapter 2, Chapter 4 turned to an evaluation of patterns of discharge to inpatient rehabilitation after stroke across Ontario’s LHIN regions. In this chapter, the feasibility of using multi-level modelling for system evaluation was demonstrated and discrepancies in regional access to rehabilitation across Ontario were identified. The adjusted estimates of the proportion of patients discharged to inpatient rehabilitation from multi-level analysis provide an improved method for system evaluation compared to the ecologic data typically used. This is important because region-level demographics and risk factor prevalence can contribute not only to variations in stroke incidence, but also to variation in the type of stroke experienced and the corresponding need for rehabilitation. Factors like older age and female sex are associated with increased stroke severity,4 which means that regions with older populations and more females can anticipate not only more strokes, but more severe strokes requiring more intensive rehabilitation. These factors can have a
considerable impact on the regional demand for rehabilitation resources, which must be accounted for both in system planning and evaluation. As innovations like electronic medical records make patient data easier to collect, health service evaluations should use
multi-level models more frequently and policy decisions should increasingly be based on their results.
In addition to demonstrating the feasibility of multi-level modelling in Ontario, the results presented in Chapter 4 also supported previous assertions that access to inpatient rehabilitation across Ontario is inequitable. In both cohorts (2004/05 and 2008/09), statistically significant variation in the proportion of patients referred to inpatient rehabilitation was demonstrated across LHIN regions, after adjusting for variation in patient-level characteristics. However, modeled data demonstrated mixed results for the relationship between resource availability and referral patterns. Although the estimates of effect in both cohorts suggested a relationship between more beds and better access, this relationship was statistically significant only in 2004/05, prior to LHIN formation. In combination, these results confirm the need for strategies to improve the equity of
rehabilitation access across the province and provide sufficient evidence to warrant pilot study of the role that additional rehabilitation beds may play in addressing inequity. They also suggest opportunities for future research to validate these findings using other data sources and in other jurisdictions.
Finally, Chapter 5 confirmed the suspected association between the availability of in- home rehabilitation resources by LHIN region and the proportion of potentially avoidable admissions of mild stroke patients to inpatient rehabilitation. This result may have the most significant policy-level implications of all. International research has consistently demonstrated that rehabilitation of appropriate patients in the community, rather than in hospital, leads to improved outcomes at reduced cost.3 In addition, caring for appropriate patients in the community can improve access to much needed inpatient rehabilitation beds (which in turn frees up acute care beds), thereby increasing the capacity of emergency departments. Appropriate funding for community-based rehabilitation can play a major role in ensuring that patients have timely access to a level of rehabilitation appropriate to their needs. This could impact Ontario’s healthcare system in many areas beyond stroke care.