Starting rehabilitation early is a widely accepted principle of care for people affected by stroke. Recommendations that rehabilitation begin ‘as soon as possible’ or ‘early’ are common in clinical guidelines [1, 2]. However, on closer inspection, it is evident that there is little clear consensus about the optimal time to start and only limited evidence to help guide our practice.
It is only recently that researchers have begun to explore questions related to the timing (and amount) of rehabilitation after stroke. Why has it taken so long for research to emerge? First, in the early 1990s, research from animal stroke models pointed to an increase in lesion volume in the brains of animals often forced to exercise for long periods soon after induced stroke . This animal research prompted a number of calls to avoid exercise in the first days after stroke , which appear to have been heeded by many. Second, rehabilitation research is a relatively young field, and research has, not surprisingly, been primarily focused on the postacute phase of care . Finally, it is only in the last 10–15 years, with the development and broad uptake of organized systems of multidisciplinary acute stroke unit care and treatment, including thrombolysis, that the introduction and testing of earlier rehabilitation interventions has become possible. Consequently, there is growing recognition that opportunities exist to improve both the individual and collective contributions of members of the stroke unit team, thus further enhancing the care delivered in these highly effective stroke units.
A number of challenges in the conduct of rehabilitation research early after stroke exist. A current challenge is that the time-to-intervention start is often poorly defined and there are no agreed definitions of what constitutes early rehabilitation. To date, the term has been used to refer to interventions beginning within days, weeks, or even months after stroke onset. A simple step that would significantly aid efforts to build the evidence base in this area is for authors to describe the start of any rehabilitation interventions in days (or hours) from stroke onset, not from hospital admission. Standardizing definitions and outcomes (see Ali et al. in this edition) will allow systematic reviews and meta-analyses to be undertaken in the area, and these are likely to drive future research. More importantly, clinical trials in which start times are clearly defined and detailed information about training regimes is provided  will aid the development of specific practice guidelines and translation into practice.
Rehabilitation is, of course, a broad concept, and there are any numbers of targeted interventions that may enhance recovery if applied in the acute phase of care. Our group elected to focus on early mobility training for a number of reasons. First, ‘early mobilization’, getting people up out of bed and continuing at regular intervals, represents a relatively simple intervention that could be implemented in both developed and developing countries if found to be effective. Furthermore, early mobilization was already a standard part of stroke unit care in some Scandinavian settings, with evidence to support the feasibility and probable safety of the intervention [7-9]. The potential for this early, highly targeted treatment to both prevent complications and the common negative sequels of stroke (e.g. secondary weakness, loss of cardiovascular fitness) and restore function, was an exciting prospect. Our group have shown that commencing mobility training within 24 h of stroke onset is feasible and appears safe , is likely to be associated with reduced costs of care , and hastens recovery of unassisted walking . These findings support the conduct of a large, international phase III trial (A Very Early Rehabilitation Trial) that is currently under way . A number of other researchers have also begun to examine mobility interventions in the first week after stroke. Three additional pilot trials (with a total of 138 patients) have recently been published [14-16], including one trial testing physiological monitoring alone and in combination with mobilization  and one in which transcranial Doppler was used to measure cerebral blood velocity during mobilization . The results from these small studies are mixed, and so far, reliable evidence about the efficacy and safety of mobility interventions is lacking.
As the body of research testing whether rehabilitation interventions can, or should, commence in the first hours or days after stroke has grown, commentary around the topic has emerged. Early mobilization now appears as one of the ‘controversies in acute stroke treatment’ , together with a range of other more medically driven interventions. What makes treatments controversial? ‘Controversial’ treatments are often those that challenge established care models or long-held views. When new approaches challenge established ideas or practices, we tend to require high-level evidence before considering a change in practice, even if the existing care is not strongly underpinned by evidence . Weighing up the potential benefits and harms of any new treatment approach is obviously important, and this should apply to rehabilitation as much as to other aspects of care.
Another early rehabilitation intervention that has been the topic of commentary  is that of constraint-induced movement therapy (CIMT) when commenced within the first two-weeks after stroke. In a small trial of 52 patients with mild stroke [Very Early Constraint-Induced Movement During Stroke Rehabilitation (VECTORS)], Dromerick et al.  found that patients in the high-intensity CIMT training group, who had three-hours of shaping per day and wore a mitten for 90% of waking hours, improved significantly less than patients with lower intensity CIMT training or those who received usual care. These findings were unexpected. In a magnetic resonance imaging substudy of nine patients, the authors found no evidence that CIMT-caused lesion enlargement and postulated why higher intensity training may be less effective, or even detrimental, to the outcome. Although the title of Steven Cramer's commentary in the journal Neurology , ‘The VECTORS study: When too much of a good thing is harmful’ may sound alarmist, Cramer does not propose we cease testing early treatments; instead, he provides a considered and insightful commentary about the field of rehabilitation research. In particular, Cramer cautions that, in cases where patients already receive therapy as usual care, the commonly held view that ‘more is better’ may not always hold. He suggests that dose-finding studies could be just as important in rehabilitation trials as they are in pharmacotherapy trials. We agree. Better understanding of the mechanism of the observed effects of rehabilitation interventions is also highly desirable. Systematic review and meta-analysis of data from studies using animal stroke models can help rehabilitation researchers explore this important data pool. Interestingly, in our most recent review exploring the response of stroke-affected rodents to constraint, we found no significant functional benefits of constraint . Study quality was only modest, and methods varied considerably across the 15 studies. Closer links between basic scientists and rehabilitation clinicians are highly desirable , and animal studies may ultimately help us unpack the effects of ‘timing’ versus ‘intensity’, which are often difficult to separate in clinical trials in this area.
The timing of rehabilitation has recently emerged as a topic of considerable interest in the intensive care literature . A number of clinical trials examining early, active mobility training for patients of mixed etiology (excluding stroke) treated in intensive care units (ICUs) have recently been published [24, 25]. These researchers have challenged the belief that acutely ill patients should be treated with bed rest. The preliminary results are promising. Patients who commence mobility training early required less mechanical ventilation, experienced less delirium, and regained functional independence faster than those who rested in bed [24, 25]. Intensive care models of stroke unit care, in which patients with stroke spend several days in the ICU before being transferred to a general medical ward for the remainder of their care, are common in some parts of the world, although clinical trials examining the efficacy of these stroke care models are few . It is unclear whether bed rest is seen as an important part of the treatment package in these units or is simply a consequence of intensive medical monitoring, which is itself controversial [27, 28]. The growing evidence from trials incorporating early rehabilitation as part of ICU care suggests that bed rest in critically ill patients may not be the best treatment option.
In the acute phase of Stroke care, getting the balance right between diagnosis and workup, rescue therapies, other medical interventions, and starting rehabilitation can be challenging. Although we have clear understanding of when some acute therapies are most effective (e.g. thrombolysis), the optimal time course for many other aspects of acute care is unknown. Improving our understanding of the impact of rehabilitation interventions on brain recovery and functional restoration should be a priority. Rehabilitation has traditionally started when everything else (more acute, important, time critical) is done and with no particular sense of urgency so long as it starts ‘early’. One of the questions we need to address in rehabilitation is: Can we afford to wait?