Published in:
01-12-2021 | Guillain-Barré Syndrome | Original work
Predicting Mechanical Ventilation Using the EGRIS in Guillain–Barré Syndrome in a Latin American Country
Published in: Neurocritical Care | Issue 3/2021
Login to get accessAbstract
Background
Up to one fifth of patients with Guillain–Barré syndrome (GBS) require mechanical ventilation (MV). The Erasmus GBS Respiratory Insufficiency Score (EGRIS) is a clinical predictive model developed in Europe to predict MV requirements among patients with GBS. However, there are significant differences between the Latin American and European population, especially in the distribution of GBS subtypes. Therefore, determining if the EGRIS is able to predict MV in a Latin American population is of clinical significance.
Methods
We retrospectively analyzed clinical and laboratory data of 177 patients with GBS in three Peruvian hospitals. We performed a multivariate logistic regression of the factors making up the EGRIS. Finally, we evaluated the EGRIS discrimination through a receiver operating characteristic curve and determined its calibration through a calibration curve and a Hosmer–Lemeshow test, a test used to determine the goodness of fit.
Results
We found that 14.1% of our patients required MV. One predictive factor of a patient’s need for early MV was the number of days between the onset of motor symptoms and hospitalization. The Medical Research Council sum score did not alter the likelihood of early MV. Bulbar weakness increased the likelihood without showing statistical significance. In contrast, facial weakness was a protective factor of it. The EGRIS was significantly higher in patients who required early MV than in those who did not (P = 0.018). It showed an area under the curve (AUC) of 0.63, with an insignificant Hosmer–Lemeshow test result.
Conclusions
Although the EGRIS was higher in patients who required early MV than in those who did not, it only showed a moderate discrimination capacity (AUC = 0.63). Facial weakness, an item of the EGRIS, was not found to be a predictive factor in our population. We suggest assessing whether these findings are due to subtype predominance and whether a modified version of the EGRIS could improve performance.