Predicting feeding-tube dependence in patients following endotracheal extubation: a two-item swallowing screen

Abstract

Abstract Background To meet the surging demands for intubation and invasive ventilation as more COVID-19 patients begin their recovery, clinicians are challenged to find an ultra-brief and minimally invasive screen for postextubation dysphagia predicting feeding-tube dependence persisting for 72 h after extubation. Methods This study examined the predictive validity of a two-item swallowing screen on feeding-tube dependence over 72 h in patients following endotracheal extubation. Intensive-care-unit (ICU) patients (≥ 20 years) successfully extubated after ≥ 48 h endotracheal intubation were screened by trained nurses using the swallowing screen (comprising oral stereognosis and cough-reflex tests) 24 h postextubation. Feeding-tube dependence persisting for 72 h postextubation was abstracted from the medical record by an independent rater. To verify the results and cross-check whether the screen predicted penetration and/or aspiration during fiberoptic endoscopic evaluation of swallowing (FEES), participants agreeing to receive FEES were analyzed within 30 min of screening. Results The results showed that 95/123 participants (77.2%) failed the screen, which predicted ICU patients’ prolonged (> 72 h) feeding-tube dependence, yielding sensitivity of 0.83, specificity of 0.35, and accuracy of 0.68. Failed-screen participants had 2.96-fold higher odds of feeding-tube dependence (95% CI, 1.13–7.76). For the 38 participants receiving FEES, the swallowing screen had 0.89 sensitivity to detect feeding-tube dependence and 0.86 sensitivity to predict penetration/aspiration, although specificity had room for improvement (0.36 and 0.21, respectively). Conclusion This ultra-brief swallowing screen is sufficiently sensitive to identify high-risk patients for feeding-tube dependence persisting over 72 h after extubation. Once identified, a further assessment and care are indicated to ensure the prompt return of patients’ oral feeding. Trial registration NCT03284892, registered on September 15, 2017.