A systematic review protocol of timing, efficacy and cost effectiveness of upper limb therapy for motor recovery post-stroke

Up to 80% of stroke survivors have upper limb impairment early after stroke [1‐3], and few demonstrate complete functional recovery at 6 months post-stroke [1, 4]. Upper limb rehabilitation trials designed to improve recovery rates have been largely unsuccessful. As a result, the burden of upper limb impairment after stroke remains high [5]. Therefore, understanding how to improve the potential for recovery of upper limb function remains a major scientific, clinical and patient priority [6, 7].

Why has it proved difficult to improve upper limb recovery after stroke? Upper limb therapy provided clinically is described as too little, too late in the recovery timeline [8]. Recent high-profile clinical trials of upper limb rehabilitation have evaluated relatively low amounts of therapy (13 to 36 h total or ~ 30 min/day), e.g. [9, 10] and/or enrolled patients relatively late post-stroke (i.e. > 6 months post-stroke), e.g. [11]. It remains unclear whether these trials were equivocal because the therapy was ineffective in terms of how much, when or what was delivered. Several syntheses have highlighted the importance of how much therapy; defining 30 min/day as too low [12], and 2 h per day as an emerging therapy threshold [13]. To date, no systematic review has examined the impact of the timing after stroke when rehabilitation is delivered.

Current clinical practice guidelines recommend that stroke patients start therapy as soon as possible post-stroke [14]. The rationale for early upper limb therapy post-stroke is supported by robust preclinical (animal) evidence, where it is well established that there is a sensitive window of heightened plasticity starting around 5 days after ischaemic stroke [15, 16]. In rodents, upper limb reaching therapy that is commenced at the beginning of the sensitive window capitalises on the injury-induced plasticity environment to produce gains in reaching that are quantitatively and qualitatively superior to those achieved when therapy of the same dose and duration is started later (around 14 and 30 days post-stroke) [15]. This underpins the rationale that timing of therapy is critically important. In humans, the post-stroke sensitive window is proposed to begin within the first 10 days post-stroke [17] and to extend out to 3 and even 6 months post-stroke [1, 4, 18]. The wide sensitive window of 6 months post-stroke needs to be better understood to guide recommendations around the critical time window to deliver upper limb therapy. A handful of upper limb trials have commenced training ≤ 10 days post-stroke [10, 19]. None have directly compared the effects of starting training early compared to a later time point. This is a major gap, which can be probed by looking at the functional gains demonstrated in previous upper limb studies that have started therapy and documented outcomes within the potential sensitive window, the first 6 months post-stroke.

The majority of completed upper limb therapy trials have tested what therapy approach to provide. Despite this, the ideal upper limb therapy intervention(s) that can improve recovery after stroke remains largely unknown. Previous Cochrane systematic reviews of specific therapy approaches have demonstrated an evidence gap, which highlights limited and/or insufficient evidence for upper limb therapy approaches that span repetitive task training [20], constraint-induced movement therapy [21], electromechanical and robot-assisted therapy [22] and hands-on-therapy interventions [23]. No systematic review has compared the relative benefit of different treatment approaches tested within clinical trials. While a previous review provided a summary of individual therapy approaches that started early post-stroke [24], the authors did not compare outcomes across therapy approaches. Direct comparison between different therapy approaches is also seldom performed in a single study due to sample size and cost constraints. Completing a review that directly compares different intervention approaches may help to establish the relative efficacy of each approach, as well as a potential hierarchy of treatment approaches to target in future trials.

The amount of upper limb therapy provided after stroke is currently pragmatic (i.e. based on available resources and what can be comfortably delivered within existing health services) rather than what might work (i.e. aspirational to match best scientific evidence). Delivery of more therapy is considered to be cost-ineffective (e.g. staffing, environment). However, consideration of the health economic arguments for optimum upper limb therapy is rarely mounted but are warranted to inform best clinical practice and policy decisions.

To address these gaps, we aim to identify relevant upper limb therapy studies that aim to improve functional recovery post-stroke and include participants of an average stroke onset < 6 months who are undergoing hospital-based therapy. We have designed a systematic review protocol to address three research questions that have been prospectively registered with PROSPERO. Systematic reviews are resource- and time-intensive, using data from a single search strategy to inform multiple questions is one way to mitigate such issues. We plan to address the following study objectives:

The present protocol is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) statement [25‐27] (see PRISMA-P checklist in Additional file 1). This protocol covers three research questions that will be answered from one systematic search strategy. All research questions have been prospectively registered on the International Prospective Register of Systematic Reviews (PROSPERO): timing (registration number, CRD42018019367), therapy efficacy (registration number, CRD42018111629) and cost-effectiveness (registration number, CRD42018111628).

The described protocol represents a novel approach to undertaking a systematic review. By using a broad search strategy, we can address several critical research questions simultaneously. An appropriately designed search strategy is likely to yield large volumes of data and will require considerable time to determine eligibility and extract relevant data. The proposed approach to completing a systematic review will minimise research waste, which is critical given the time taken to complete systematic reviews. Furthermore, the outputs will synthesise past research to generate novel, data-driven hypotheses for future research that aims to define a new therapy pathway(s) to promote upper limb recovery. While each review may yield the answers to our questions, there is the likelihood that they will demonstrate that we do not yet have the appropriate data to allow clinicians to make evidence-based decisions about hospital-based upper limb therapy within the first 6 months post-stroke. This finding is as important as yielding all the answers and will form the foundation for designing studies that can answer these important research questions in stroke rehabilitation and recovery.

The methods and results of our systematic reviews will be reported following the PRISMA statement and their relevant extensions [25‐27]. While this review will be completed with the upmost of care, there are some limitations. It is well documented that a large range of upper limb outcome measures are reported in research [37] which prevents individual patient data analysis. Further, there may be missing data that will impact data collation, comparison across studies and pooled analyses. We acknowledge language restrictions beyond English, German, French and Dutch. We do not intend to hand-search conference proceedings. In addition, the observational nature of the subgroup analyses means they should be interpreted with caution, as it is known that subgroup analyses can be less highly powered than analyses for main effects.