Dysphagia commonly affects patients in intensive care units (ICU), particularly those on mechanical ventilation, and is associated with high risk of mortality. This systematic review and meta-analysis aimed to identify predictors for dysphagia in ventilated ICU patients by summarising existing predictive models and considering their risk of bias. Following searches of five databases according to stated inclusion criteria, the team identified nine prediction models for post-extubation dysphagia (PED) for data extraction. Sample sizes ranged from 87 to 933, with the prevalence of PED ranging from 12.4% to 46.8%. Assessment of PED took place between three to 48 hours following extubation and the main outcome measure for six of the nine studies was standardised swallowing assessment. Predictive models included 19 different predictors. The most identified risk predictors were duration of in situ endotracheal tube (≥72hrs), APACHE II score, age ≥65 years and duration of enteral feeding (≥72hrs). However, there was high risk of bias for all studies when assessed using the PROBAST tool. This was due to lack of external validation, limited pre-screening of variables and inappropriate analysis of predictor variables based on whether they were continuous or categorical. The review recommends that further research into predictor variables for PED requires methodological rigour during development and a focus on validating existing models with a range of populations and different patient characteristics. This will improve their clinical effectiveness, reliability and accuracy.
Can prediction models help identify dysphagia in ventilated patients?
Reviewed by Gemma Clunie
Prediction Models for Dysphagia in Intensive Care Unit after Mechanical Ventilation: A Systematic Review and Meta-analysis.
CONTRIBUTOR
Gemma Clunie
BA (Hon), MSc, PhD, MRCSLT, Imperial College Healthcare NHS Trust; Honorary Research Fellow, Imperial College London, UK.
View Full Profile
