Stroke survivors commonly have respiratory muscle weakness, swallowing disturbance, general weakness, reduced the daily living activities and short of motivation for rehabilitation. Respiratory muscle training (RMT) has been reported to improve the pulmonary function, respiratory muscle strength, exercise capacity, sensation of dyspnea and quality of life in several diseases, but rare in stroke patients. Reviewing previous reports, the protocol, intensity and duration of respiratory muscle training is still variable.

Purpose: To investigate the feasibility and efficacy of respiratory muscle training on cardio-pulmonary function, swallowing function, cough function and reduction of incidence of pneumonia in post-stroke patients.

Methods: A prospective, single blinded, randomized study. Consecutive patients with diagnosis of stroke will be proved by magnetic resonance image or computerized tomography. Stroke patients, aged 35-80 years old, with inspiratory muscle weakness or swallowing disturbance will be enrolled and randomly divided into control group (usual rehabilitation alone) and experimental group [inspiratory muscle training (IMT) group for patients with inspiratory muscle weakness and expiratory muscle strengthening training (EMT) for patients with swallowing disturbance]. Each patients will receive usual rehabilitation.

The investigator expect that RMT will be practical for the restoration of respiratory muscle, swallowing function, cough function and voice quality, thereby reduction of the incidence of pneumonia.

Stroke patients has been observed to have a higher position of diaphragm in thoracic radiographs on the affected side after 1 day to 2 years of acute event, reduced movement of whole hemi-thorax and excursion of diaphragm on the affected side during deep voluntary breathing, and a tendency for predominance of rib cage contributions during tidal breathing, reduced maximal inspiratory pressure (MIP) and maximal expiratory pressure (MEP). Their cardiorespiratory fitness could be markedly impaired within 7 weeks after a stroke. Neves, et al. reported that the recent infarction group had a positive correlation between peak oxygen uptake and both MIP and maximal incremental pressure.

Inspiratory muscle training (IMT). Inspiratory loading can improve limb blood flow, and oxygen uptake efficiency. Threshold inspiratory loading has been recommended to improve inspiratory muscle strength in chronic obstructive pulmonary disease (COPD), bronchiectasis, congestive heart failure (CHF), heart failure, CABG and neuromuscular disease et al. Sutbeyaz et al. ever reported that a statistically significant increase in MIP and MEP of the breathing retraining group in subacute stroke patients. The training intensity of IMT varied from 30% to 60% of MIP, duration of 3 to 7 times per week, the duration of each session from 10-30 minutes for a period of training from 6 to12 weeks.

Expiratory muscle training (EMT): EMT has been shown to improve the dyspnea perception and walking distance within a time unit in COPD persons, aerodynamic changes in cough flow in patients with multiple sclerosis and Parkinson disease, and swallowing safety in PD, and voice aerodynamic. But till now, the protocol, intensity and duration of EMT is still variable, from twice daily at the user's own frequency and tidal volume for 15 min for 4 week, or a week for 1/2 hour, start 15% MEP max, increased by 5-10% each session to 60% of MEP.

Method: Forty-six stroke patients with swallowing disturbance or respiratory muscle weakness, age between 35 to 80 years, will be enrolled. The participants will be randomly assigned by a computer random number generator into two groups, including the experimental group (RMT group plus regular rehabilitation) and the control group (regular rehabilitation alone). All allocations will be concealed in the opaque envelopes. The research assistant will enroll the participants, generate the allocation sequence, and assign participants to their groups after obtaining the informed consent. The research assistant who are responsible for measuring outcomes will be blind to the allocation.

Experimental group will be randomized into 2 groups:

Group I: EMT+IMT for patients with inspiratory muscle weakness and swallowing disturbance (MIP less than 70% of normal range).

Group II: EMT for patients with swallowing disturbance.

Control group will receive regular rehabilitation.

Each subject with respiratory muscle weakness or swallowing disturbance, baseline characteristics, including height, weight, body mass index, duration of the disease, neurological level (Brunnstrom's stage), spirometry, peak cough flow, resting heart rate, systolic and diastolic blood pressure, resting oxyhemoglobin saturation (SpO2), MIP, MEP , Borg's scale (0.5 to 10), 6-minute walking test, cough function, fatigue assessment scale.

Training protocol Group I: IMT+EMT:

IMT will commence from 30% to 60 % of MIP and then adjust one level of training loading according to the tolerance of continuously breathing through a respiratory trainer for one set of 30 breaths or 6 sets of 5 repetitions with one or two minute of rest between sets, twice per day, 5 days per week. Training resistance will be adjusted as tolerated. The loading will be performed with the previous resistance setting or even lower if training load is not tolerated or not completed.

During IMT, patients will be instructed to place their lips around breathing trainer in a sitting position with a nose-clip, inhale with enough force to open the valve (inhale deeply and forcefully), exhale through the mouthpiece (exhale slowly and gently), and then continue inhaling and exhaling without removing the device from their mouths. The training load and training program will be instructed by an experienced respiratory technician.

Training protocol Group II: EMT for patients with swallowing disturbance EMT will commence from 15% to 75% of threshold load of an individual's MEP, 5 sets, 5 repetition with one or two minute of rest between sets, twice per day, 5 days per week. And training resistance will be adjusted accordingly. The loading will be performed with the previous resistance setting or even lower if training load is not tolerated or not completed.

Each patients with swallowing disturbance will receive swallowing screen test, Functional Oral Intake Scale, voice quality analysis and surface electromyography by an experience physician and experienced speech therapist to evaluate the functional level of oral intake of food and liquid, voice quality and measurement of submental muscle strength respectively. And swallowing training program will be conducted by an experienced speech therapist.

Procedures for surface EMG (sEMG): The sEMG activities of swallowing-related muscles while swallowing and doing expiratory breathing through a breathing trainer will be measured with the Nicolet Viking Four Electrodiagnostic system, Multiple Modality Program.

Procedures for voice analysis: Voice quality will be assessed with Computerized Speech Lab (CSL™)，Model 4500 (Mutlti-Dimensional Voice). Patients will sustain "ah" for at least 3 seconds at the most comfortable speaking pitch and loudness, as well as at the lowest pitch and highest pitch increasing and decreasing loudness with audio recording.

Each patient will be assessed again at 6 weeks later.

Supervision: During the first session, the investigator will decrease one level of threshold load imposed if a modest decrease in SpO2, which implies that it most likely reflects the onset of hypoventilation, particularly when the load approaches the maximum; otherwise the investigators will use continuous use EKG to monitor cardiac activity throughout the RMT session.

Regular rehabilitation program All participants will receive usual rehabilitation care including body positioning instruction, postural correction, breathing control, cough maneuver, respiratory muscle stretch, chest wall mobility exercise, fatigue management.

For checking the compliance of RMT at home, patients will be monitored by making a phone call to them once a week.

Inclusion Criteria:

• Patients identified as stroke and capable of performing voluntary respiratory maneuvers.

Exclusion Criteria:

• Increased intracranial pressure
• Uncontrolled hypertension
• Complicated arrhythmia
• Decompensated heart failure
• Unstable angina
• Myocardial infarction in the preceding 3 months
• Pneumothorax, bullae/blebs or infection.