A cross-over, double-blinded intervention pilot study was performed to assess the feasibility of blinding procedure. Secondary, long-term changes in visuospatial neglect (VSN) symptoms were monitored during inpatient rehabilitation to gain information about recovery-rate in patients with first-time ever stroke in rehabilitative context for future trials. This pilot study closely agrees with CONSORT-Guidelines (Additional file 3).
The study was conducted within the neurological rehabilitation ward of the Kliniken Beelitz GmbH in Brandenburg, Germany. All patients were pre-screened for eligibility. Inclusion criteria comprised: ischemic or hemorrhagic stroke within the right hemisphere (confirmed by neuro-imaging), early subacute phase (> 7 and < 56 days after stroke onset), age ≥ 18 years, right-handed (19), and residual VSN symptoms. Major exclusion criteria included: history of stroke, severe cognitive impairment, epilepsy and the presence of a pacemaker (Additional file 2 table S1).
The presence of VSN was tested at screening visit using selected tests from the Behavioral Inattention Test battery (BIT, German version: Star Cancellation, Figure Copying, and Line Bisection (14)). Only patients with impaired performance in at least two of these tests, and confirmed VSN diagnosis by the treating neuropsychologist entered the baseline visit, which was scheduled approximately one week after the screening visit to account for spontaneous recovery of VSN symptoms. Subsequently, an atDCS and stDCS session were applied in randomized order (48 h wash-out period in between) during standard care neuropsychological therapy (30 min, exploration tasks) by the treating therapist. On the last day of the hospital stay patients were re-assessed (follow-up).
The randomization list was generated by a self-written script (Additional file 1) using R-statistical software (random generator). A stimulation protocol for atDCS and stDCS were programmed and performed by the same assessor (TR). Patients and treating therapists were blinded to the stimulation protocol.
tDCS was applied by a StarStim tDCS stimulator (Neuroelectrics, Barcelona, Spain) via electrodes (round electrodes, 25 cm²) mounted over both posterior parietal cortices (P4-anode; P3-cathode, bi-hemispheric protocol) determined with the international 10-20 EEG System with an intensity of 2 mA (current density: 0.8 A/m²). tDCS was delivered for 20 min (atDCS) or 30 s (stDCS) in a ramp-like fashion with a 15 s (fade in/fade out) interval at the beginning and the end of the stimulation.
After each tDCS session patients were ask (by TR): “Do you think you received an active or sham stimulation or are you undecided?” to assess blinding success, and for the sensation of itching, pain, burning, heat, taste of metal, or fatigue during stimulation. Adverse events were monitored throughout the hospital stay and were noted if they could be related to the intervention.
At baseline demographic and clinical data were recorded including impairment caused by stroke using the National Institute of Stroke Scale (NIHSS) (21). Global cognitive functioning were assessed by the Montreal Cognitive Assessment (MoCA) (22). VSN symptoms were assessed at baseline and follow-up (Star-, Letter-, and Line Cancellation, Line Bisection, Figure Copying and Text Reading ).
To demonstrate the feasibility of blinding procedures 12 patients were included. Around 6 % of all stroke patients admitted to the clinic were eligible patients for this study.
Each patient evaluated stimulation mode twice for stDCS and atDCS. Guessing answers of stimulation mode were coded as: a) sham, b) indifferent, c) active. Binary logistic mixed models were applied to estimate if guessing of the stimulation condition was associated with atDCS-stimulation condition by accounting for the clustered data structure (repeated measures, random intercept model) (melogit command in stata). Patients judgements were included as independent (nominal), the actual stimulation condition as dependent variable (coded: atDCS: 1, stDCS: 0).
All analyses were performed in an exploratory framework with descriptive statistics presenting mean (SD) or median [IQR] depending on the distribution of the data. Sensations were aggregated into a new dichotomous variable “any sensation” (present /absent). Proportions of patients who perceived any sensation under atDCS and stDCS were compaired by non-parametric McNemar test. Changes between performance in baseline and follow-up were analysed by Wilcoxon signed rank test. Cohen’s d with confidence intervals (CI) are reported as effect size. Analyses were not corrected for multiple testing.
All programming and analyses were done using R-Statistical Software Version 3.4.4 (23) or Stata Statistical Software, Release 15 (24).