This study was a single-center, prospective, randomized, double-blind, sham stimulation-controlled trial. Stroke patients who were treated from June 2013 to April 2015 in the Department of Rehabilitation Medicine at Asan Medical Center were enrolled. All study subjects were screened for cognitive dysfunction using the Korean version of the Mini-Mental Status Examination (MMSE). Our study comprised patients who had experienced a first-ever unilateral middle cerebral artery infarction. The duration from stroke onset ranged from 1 week to 2 months, but all of the study patients had experienced their first-ever stroke. A skillful speech therapist evaluated cognitive function and speech function using evaluation tools such as the Korean version of the Western Aphasia Battery, and patients who had aphasia, apraxia of speech, cognitive impairment (MMSE<20), poor mental status, vocal cord palsy, history of epilepsy, or bilateral infarction were excluded. We estimated the target number of participants with reference to a previous study [13]. All participants provided written informed consent, and our local Ethics Committee approved the study protocol.

To determine the resting motor threshold and stimulation area, we recorded motor-evoked potentials from the orbicularis oris muscles on each patient's non-affected side using transcranial magnetic stimulation. Focal transcranial magnetic stimulation was applied using a Magstim Rapid magnetic stimulator (Magstim Company Ltd., Dyfed, UK). Briefly, a Magstim circular coil (external diameter, 90 mm) was placed onto each subject's contralateral motor cortex to identify the hotspot, defined as the area that produced the largest amplitude of motor-evoked potentials. The resting motor threshold was defined as the stimulus intensity required to produce motor-evoked potentials >100 kV at a peak-to-peak amplitude during three of five consecutive trials on the orbicularis oris.

We randomly sorted our enrolled stroke patients into two groups: an rTMS group and a sham stimulation group. Randomization was done according to a table of random numbers; odd numbers went to the rTMS group and even numbers went to the sham stimulation group. According to previous studies, LF rTMS on the non-affected side is more effective in producing recovery from weakness than is HF rTMS on the affected side [12], and we thus decided to use only LF rTMS. While seated in a comfortable chair with foam ear plugs, each patient was treated with 10 consecutive sessions (five times per week for 2 weeks) of rTMS, performed by a physiatrist who used a 70-mm, aircooled, figure-of-eight Y-shaped coil.

We performed rTMS at a low frequency (1 Hz), at 90% amplitude of evoked motor threshold, and with 1,500 stimulations/day on the hotspot. We applied the same type of LF protocol that was described in the study by Takeuchi et al. [14]. The sham stimulation occurred using the same protocol as that for the LF stimulation, except that the angle of the coil was perpendicular to the skull rather that tangential to it. Thus, the magnetic field could not penetrate the brain, although the subjects could hear the sound that was produced. This protocol has been used in other studies [14,15,16,17]. The subjects in both study groups received speech therapy for 30 minutes, 5 days a week from a skilled speech therapist who was blind to the nature of the study during the 2-week intervention period. The patients were also unaware of which group they had been assigned to.

Dysarthria was evaluated by a single skilled speech therapist who was blind to the study protocol. This clinician performed the Urimal Test of Articulation and Phonology (U-TAP), and measured alternative motion rates (AMR), sequential motion rates (SMR), and the maximal phonation time (MPT) as outcome measures before and after the rTMS sessions. These tests were verified for evaluating dysarthria in a number of studies and they were designed to measure the velocity of movement and coordination of articulators including the tongue and lips [13]. MPT was determined by measuring the duration of the /a/ vowel produced at a comfortable amplitude and pitch level after maximum inspiration. Following appropriate instructions, this procedure was repeated three times by the subject, and the recording with the longest duration was considered to be the MPT. Subjects were then asked to repeat the phonemes /pǝ/, /tǝ/, and /kǝ/ as rapidly as possible and for as long as possible at a comfortable pitch and loudness. This test is used for the clinical assessment of speech motor function [18]. Diadochokinesis (DDK) can be used to measure syllable timing and rhythm in speech [19]. Irregular syllable timing during DDK is strongly associated with ataxic dysarthria and can be associated with spastic dysarthria as well [20]. DDK measures are divided into two types: AMR and SMR. AMR is measured as the rapid repetition of a single syllable /pǝ/ and SMR is measured as the rapid repetition of a sequence of syllables (/pǝ/, /tǝ/, and /kǝ/).

We evaluated sex, age, duration from stroke, and location of the stroke as the baseline characteristics of our study patients. We also evaluated their Modified Barthel Index (MBI) and Motricity index scores before beginning the rTMS sessions.

Statistical analyses were performed with SPSS ver. 18 (SPSS Inc., Chicago, IL, USA). Data are presented as mean±standard deviation. The general characteristics of the two study groups were analyzed with the Mann-Whitney U test and Fisher exact test. Paired comparisons within the groups (before and immediately after rTMS) were analyzed using the Wilcoxon signed-rank test. A p-value ≤0.05 was considered to indicate statistical significance. Comparisons of interval changes between the two groups were analyzed with the Mann-Whitney U test.