Key points are not available for this paper at this time.
Background: Post-extubation dysphagia (PED) is a common complication of critical illness and is associated with aspiration pneumonia, prolonged hospitalisation, and prolonged recovery. Expiratory muscle strength training (EMST) has been shown to improve swallowing function in selected populations, but its role in survivors of critical illness remains unclear. Objective: To systematically review the evidence evaluating the effects of EMST on swallowing outcomes in survivors of critical illness with PED. Methods: This systematic review was registered with PROSPERO (CRD42023444479) and conducted in accordance with PRISMA guidelines. Searches of MEDLINE, PubMed, EMBASE, CINAHL, and Cochrane databases were performed from 2000 to March 2026. Eligible studies included adult patients requiring >48 h of mechanical ventilation who underwent volitional EMST following extubation. Due to anticipated limited evidence, all quantitative study designs were included. Risk of bias was assessed using the Newcastle–Ottawa scale and the Joanna Briggs Institute checklist. Meta-analysis was not performed due to heterogeneity. Results: Three studies met inclusion criteria, including two from the same research group: one case series ( n = 2) and two single-group cohort studies ( n = 50 and n = 13). All studies reported improvements in swallowing measures, alongside increases in respiratory muscle strength and cough flow. However, EMST was either combined with inspiratory muscle training or other therapies, and no randomised controlled trials were identified. Methodological quality was low, with high risk of bias and absence of comparator groups. Conclusions: Evidence supporting EMST for PED in survivors of critical illness is extremely limited and of low methodological quality. Whilst mechanistic plausibility and data from other populations suggest potential benefit, robust randomised controlled trials with standardised training protocols and outcome measures are required before clinical implementation can be recommended.
Skurok et al. (Wed,) studied this question.