Abstract Details

Title Modified EGRIS Proposal For Improved Prediction Of Mechanical Ventilation In The Guillain-Barre Syndrome For The Peruvian Population

Topic Neuro Trauma, Critical Care, and Sports Neurology

Presentation(s) Neurocritical Care Posters (7:00 AM-5:00 PM)

Poster/Presentation
Number 026

Objective

To evaluate the predictive capacity of EGRIS and compare it to a modified version for the Peruvian population.

Background The Erasmus GBS Insufficiency Score (EGRIS) is a predictive clinical model developed in Europe for early mechanical ventilation (MV) in Guillain-Barré syndrome (GBS). However, the main subtype in Latin America, the axonal subtype, is not frequent in Europe. Differences in clinical findings and disease progression have been described between both subtypes, including the frequency of facial weakness, an item of the score. This could limit its applicability, therefore, we aimed to evaluate the predictive capacity of EGRIS and compare it to a modified version for the Peruvian population.

Design/Methods We retrospectively analyzed GBS cases from three hospitals in Lima-Peru between 2007-2018. We determined the OR of the EGRIS items using multivariate logistic regression. We elaborated a modified EGRIS eliminating the non-associated elements to compare the discrimination and calibration of both in predicting MV using received operator curves (ROC), and the Hosmer-Lemeshow test in the STATA 16 software. The research protocol was approved by all IRBs.

Results We identified 179 cases, of these 14.1% required MV. In the multivariate regression, facial weakness was not predictive in the Peruvian population (OR: 0.13, IC95%: 0.02-0.97, p=0.01). We tested that the modified version of the EGRIS presented an area under the curve (AUC) of 0.66, significantly higher than the original score (AUC: 0.63, p <0.000). Both scores showed good calibration in the Hosmer-Lemeshow test.

Conclusions Ourmodified EGRIS presented better discrimination in the Peruvian population compared to the European version of EGRIS in predicting MV, despite remaining suboptimal. These results justify a larger prospective study to test the correct performance of both scores.

Authors/Disclosures
Marco M. Malaga, MD (University of California in San Francisco) PRESENTER	Mr. Malaga has nothing to disclose.
	No disclosure on file
Diego A. Bustamante (Universidad de San Martin de Porres)	Mr. Bustamante has nothing to disclose.
	No disclosure on file
	No disclosure on file
Carlos Alva-Diaz	Carlos Alva-Diaz has nothing to disclose.

�鶹��ýӳ��

�鶹��ýӳ��

Abstract Details