Prediction of respiratory failure and prolonged mechanical ventilation in Guillain-Barré syndrome: A prospective cohort study in Bangladesh.

Abstract

Background and aims: The aim of this study is to validate and perform a region-specific adjustment of the Erasmus GBS Respiratory Insufficiency Score (EGRIS) and identify potential predictors of prolonged mechanical ventilation (PMV) among Guillain-Barré syndrome (GBS) patients from Bangladesh.

Methods: We enrolled GBS patients from four prospective observational cohort studies conducted in Bangladesh. Accuracy of EGRIS to predict the requirement of MV in <7 days of study entry was evaluated. Model performance was assessed by discrimination (ability of the model to differentiate between patients who needed MV or not) and calibration (accuracy of absolute risk estimates). PMV was defined as duration of MV >14 days. Potential predictors for PMV were evaluated by Cox regression.

Results: A total of 594 GBS patients aged ≥6 years old were enrolled; of whom 541 patients had complete EGRIS data prior to MV and were included in validation analysis. EGRIS correctly distinguished between patients requiring MV or not in 81% pairs (AUC = 0.81). EGRIS overestimated the probability of MV than the observed probability (41% vs. 20%) which was resolved by updating of the model intercept. Inability to flex hip at day 7 of start of MV was the strongest predictor for PMV with predicted probabilities of 82%.

Interpretation: EGRIS accurately predicts the need for MV in GBS patients from Bangladesh. This study developed a region-specific version of EGRIS and identified predictors of PMV. These findings can assist clinicians to identify patients at high risk of developing respiratory failure and requiring PMV to ensure timely intubation and tracheostomy of the patients in low resource settings.