Access, timing and frequency of very early stroke rehabilitation – insights from the Baden-Wuerttemberg stroke registry

Organized stroke unit care has proven to be more effective than treatment on general wards only [1‐3]. A systematic review of the Stroke Unit Trialists Collaboration demonstrated that professional stroke unit care is particularly superior because it prevents immobility-related complications. Further, physical therapy (PT), occupational therapy (OT) and speech therapy (ST) including are regarded to significantly account for this effect [4]. Dysphagia is a common finding in stroke and associated with a higher risk for pneumonia, malnutrition and death, thus making a swallowing assessment necessary immediately after admission by trained nursing staff and/or speech therapists [5, 6]. Recently a Cochrane review confirmed the general effectiveness of physical rehabilitation on stroke outcome and also provided some evidence for beneficial effects when treatment was initiated early, although a lack of high quality studies was noted [7]. Very early rehabilitation (VER) is thought to prevent inactivity-associated complications of multiple biological systems, i.e. the respiratory system (pneumonia, atelectases), blood circulation (deep vein thrombosis, pulmonary embolism), immunosuppression, bedsores and catabolism/muscle atrophy. Further favourable effects are hypothesized to comprise of stimulated neuronal plasticity and a reduced risk for mood disorders associated with stroke [8‐11]. The benefits of VER are however accompanied by potentially harmful side effects as arterial blood pressure oscillation during exercise and/or out of bed mobilization with a higher risk for hematoma growth in intracerebral hemorrhage (ICH), collapsing collateral blood supply of the penumbra in ischemic stroke (IS), and secondary ICH in successfully recanalized IS patients [12‐14]. However, the evidence for both the precise beneficial and harmful effects of VER remains poor [15]. In April 2015 the phase III randomized controlled “Efficacy and safety of very early mobilization within 24 h of stroke onset (AVERT)” trial was published and demonstrated a lesser chance to achieve functional independence 3 months after stroke in the earlier and more intensively treated stroke group, defined as a modified Rankin scale (mRS) of 0–2. Subgroup analyses indicated that patients with severe stroke and/or ICH might be more susceptible to harmful side effects [16]. Multiple factors might have contributed to these findings. The treatment group was mobilized earlier and received twice as much daily sessions of out of bed therapy, which took three times longer than for the control group. Despite to this uncertainty regarding the harmful factor, it is somewhat unclear to what extend regular stroke care will have to change in light of the AVERT findings. This moreover since there is a lack of description of interdisciplinary therapy content (type, frequency and intensity) and therapy strategy (in-bed therapy vs. out-of bed mobilization) in everyday clinical practice [16]. Based on the authors clinical impressions we hypothesized a positive correlation between stroke symptom severity and VER at higher frequency, which might need reevaluation after publication of the AVERT trial. With a descriptive analysis of the Baden-Wuerttemberg (BW) stroke registry we thus aimed to provide a more detailed insight into the application of PT, OT, and ST in patients with IS and ICH in clinical practice.

We performed a retrospective analysis of patients with acute IS and ICH prospectively registered in a large and consecutive hospital-based stroke registry in central Europe. Permission to analyze the registry was obtained by the governing board of the office for quality assurance in hospitals (Geschäftsstelle für Qualitätssicherung im Krankenhaus, GeQik).

Despite to the fact that presumably most hospitalized stroke patients receive VER in a widely varying amount, surprisingly little is known about its current stage in clinical routine [16].

We report data from the Baden-Wuerttemberg stroke registry on the general access to VER, the timing from admission to first physical, occupational, and speech therapy and the average numbers of therapy sessions. All admitted patients with the ICD10 diagnoses I61 and I63 from a 5 year period were taken into analysis, except for patients with either early discharge within 24 h of admission or mechanical ventilation, thereby making the application of VER unlikely. In our study population approximately 80% of the patients admitted for IS or ICH were treated on stroke units or intensive care units with a correspondingly high level of expertise. About 90% of the study cohorts had access to PT and this number was ≥95% for patients with motor deficits at admission. OT was applied in approximately 60% of the total study cohorts. 26% (IS) and 28% (ICH) of the patients with motor deficits at admission were not attended by an occupational therapist. 70% (IS) and 65% (ICH) of the total study population were evaluated and/or treated by speech therapists. For patients with speech or language disorders at admission access to ST was higher, suggesting a moderate undersupply of 10% (IS) to 14% (ICH). The majority of patients with IS and, to a slightly minor degree, ICH received their first PT session within 24 h and OT or ST session within 48 h of hospitalization.

Based on our data VER might be less frequently applied to patients with ICH compared to IS and also the proportion of patients with ICH and immediate access to therapy was observed to be lower. Differences were small but consistent throughout the analysis. Two possible factors might have contributed to this finding. First, patients with fatal ICH were more frequently denied therapy compared to IS patients with the same outcome. This might be due to earlier decision making for palliative care in ICH than in IS. Second, a survey from 2011 indicated that stroke professionals seem to be more restrained regarding early mobilization in patients with ICH for concerns regarding blood pressure oscillation, which also may account for the observed differences [21]. This opinion is supported by observations from a post-hoc analysis of the INTERACT 2 trial that excessive blood pressure variability as it may occur during out-of bed mobilization seems to be associated with poorer outcome [22]. In our analysis a discharge mRS score of 2 to 5 was associated with the highest application of VER both regarding its general access and the average number of sessions for all three types of therapy. This observation was first reported for stroke patients in the 1970s, albeit the availability of PT, OT and ST was much less and the timing from onset to therapy was reported in weeks rather than days [23]. It might also explain the higher average number of therapy sessions in patients with ICH, with patients being more severely disabled compared to IS. We performed multiple logistic regression analyses to investigate the influence of VER on stroke outcome but the results were fundamentally contradictory, most likely because we were finally unable to statistically compensate biases by individual decision making. We thus believe that stroke registries are not suitable to investigate the influence of VER on stroke outcome.

Although the incidence of ischemic and hemorrhagic strokes is high with millions affected worldwide annually, the optimal dose and timing for very early stroke rehabilitation are entirely unclear [24]. Data from animal stroke models revealed a potentially NMDA-mediated increased cell death rate within the infarct and peri-infarct area under very early and intense training, which was counterbalanced by an improved motor performance when therapy started at the earliest 5 days post-stroke [25‐29]. However, the intensity of training in animal models is much higher than compared to humans [28]. A trial comparable to animal studies used constraint-induced movement therapy and observed a worse functional day 90 outcome when therapy started at day 10 after stroke onset [30]. VER within 24 h of stroke onset was investigated in three randomized controlled trials of limited size which were published between 2008 and 2010. A systematical review in 2015 did not demonstrate superiority of VER compared to standard care [31‐34]. In April 2015 the large phase III AVERT trial was published and to the surprise of the community reported a less favourable 3 month stroke outcome of the earlier and higher-dose treated group. Although the multiple differences between the study groups (earlier mobilization at higher frequency and for a longer period of time) impair the identification of the actually harmful factor, the AVERT trial broke with the widely-accepted assumption that very early and intense mobilization and rehabilitation represents the best medical therapy to promote recovery in all patients with acute cerebrovascular disease [16]. Further stroke subgroup analyses hypothesized that in particular patients with ICH and/or severe disability might benefit from delayed mobilization and rehabilitative treatment at lower dosage. However, even although more than 2000 patients participated in the trial, pre-specified subgroup analyses were still underpowered to draw robust conclusions. Very recent data from AVERT indicates that short and frequent out-of bed mobilization is superior to an increased amount of time spent with out-of bed activity [35]. In this instable phase VER for patients with severe disability might also be applied in a neurosensorial approach, verticalisation by robotic devices, repetitive treatment by motocycle in bed, swallowing and respiratory treatment [36]. While most clinical trials and observational studies focused on mobilization and physical rehabilitation approaches, the impact of ST was less frequently investigated. ST does not require mobilization and thus is not associated with several potentially harmful side effects of out-of bed PT. Two small clinical trials with in total 59 patients suggested that early ST for mild to moderate aphasia within 3 days of stroke onset results in improved outcome after 6 months [37, 38]. Moreover, very early swallowing assessment by trained nurses or speech therapists is of high value to identify patients at risk for aspiration pneumonia and malnutrition [39].

Obviously, interdisciplinary treatment in the acute phase has to be coordinated and adapted to the acute diagnostic investigations and treatments, otherwise VER might not be applied in the adequate moment [36]. It thus seems very likely that specific subgroups concerning stroke etiology, severity of stroke symptoms, comorbidities, and acute stroke therapy like hematoma evacuation, IV thrombolysis, or thrombectomy need to be taken into account for an individualized rehabilitation protocol [40]. An overview on worldwide national stroke guidelines was presented 2015 by Bernhardt et al and demonstrated a widely recommended early mobilization regime (22 out of 30 national guidelines, of these eight recommended mobilization within 24 h after stroke onset). Early rehabilitation is endorsed in 11 national guidelines, i.e. Canada, Croatia, Finland, Germany, Japan, New Zealand, Norway, Scotland, Singapore, South Africa, and the USA [15]. Current US-American guidelines state that VER in the acute care hospital is optional, but all patients should at least undergo rehabilitative assessment prior to discharge to an inpatient rehabilitation care unit [41, 42]. The 2015 updated Canadian stroke rehabilitation guideline was the first to have incorporated findings from the AVERT trial and does no longer recommend frequent out-of bed activity within 24 h after stroke onset [43]. As a supranational committee, the European Stroke Organization guidelines for the management of ischemic stroke from 2008 recommend an early initiation of rehabilitation therapy but do not provide precise time frames [44]. In Germany, treatment on specialized stroke units or intensive care units is remunerated with the procedure code 8–981.x. This includes daily PT, OT, and ST for patients with treatable disabilities and therapy has to be commenced not later than the next day after admission.

Based on the general access to VER and the average numbers of therapy sessions our data from routine hospital care demonstrates that severely disabled stroke patients receive training at high frequency, most likely under the assumption that they will benefit most. We regard our data robust even though the “admission to treatment time” and not “onset to treatment time” was measured, since there is a correlation between stroke severity and lesser onset to hospital admission time [18, 45]. However, the minimal demand of rehabilitation therapy is at least not to be harmful to the treated subject. The results of the AVERT trial should therefore be taken into account for the application of VER in clinical routine to avoid overly intense therapy in severely disabled patients, as done in the current Canadian stroke rehabilitation practice guidelines [43]. One further important lesson from AVERT is the urgent need for further clinical trials which will have to focus on stroke subgroups instead of the entire spectrum of IS and nontraumatic ICH. Otherwise conclusions as from AVERT lead to misunderstanding. The condition to measure the outcome of the rehabilitation treatment needs to include shaping, that means (as shown in the Baden-Wuerttemberg stroke registry) the treatment has to be adapted to the degree of disability of the patient and his rehabilitation potential. It moreover might be best evaluated by individual goal assessment based for instance on the International Classification of Functioning, Disability and Health [46]. In the meantime a temporary international consensus for severely disabled stroke patients and patients with ICH is regarded necessary.

Specific limitations of our study need to be discussed. Firstly, since the data on therapy was collected from a stroke registry, no quantitative (amount of PT, OT, and ST sessions in minutes) or qualitative (in-bed or out of-bed therapy) information was available. This limits the ability to directly compare our results to the AVERT trial. Moreover, in the BW stroke registry the documentation on timing of OT and ST (≤48 h, >48 h) was less precise compared to PT (≤24 h,>24–48 h, >48 h), which limits a detailed comparison for the first 48 h after admission. Secondly, in large stroke registries the general accuracy of results strongly depends on the accuracy of data input at the participating hospitals. In the Baden-Wuerttemberg stroke registry data documentation is usually performed by a physician, diagnosis related groups (DRG) coordinator or trained study nurse, which should ensure a sufficient data quality. The provided data is validated centrally by predefined logic and range checks and annual performance reports including feedback on data quality are provided for each participating hospital. Nevertheless, for patients being discharged within less than 7 days of admission a calculated score had to be extrapolated by the admitting hospital and it is regarded very plausible that the complexity of data input will negatively affect its accuracy. Based on the lower interquartile range of hospitalization time (7 days in ICH and 6 days in IS) approximately 25% of the ICH and IS patients were discharged after less than 7 days of admission. Thus for this fraction the registry has no access to the raw data and data quality cannot be comparably assured. Thirdly, missing data on functional disability at admission might bias the results and thus interpretation of data.

Patients admitted for IS or ICH in Baden-Wuerttemberg have excellent access to PT, while an approximately 12–17% undersupply of ST and more in particular a 30% undersupply of OT in patients with treatable disabilities was observed. When stratified according to the mRS at discharge, we observed that in acute stroke care the staff decision for VER and its frequency seem to reflect attempts to optimize resources, i.e. those with either very good or very unfavorable prognosis received therapy less frequently. Severely disabled stroke patients received VER at the earliest and treatment sessions were applied at highest frequency. Since there are current concerns how to perform rehabilitation therapy in severely disabled ischemic stroke patients and patients with ICH in the very early, vulnerable phase, a temporary consensus for this subgroup is regarded necessary. Future clinical trials are needed to investigate the response of specific stroke subgroups to VER and the methodology of these trials will have to be fundamentally reconsidered