Citas bibligráficas
Delgado, C., (2024). Modelos predictivos con índice de ROX y variables basales para el uso de ventilación mecánica invasiva en residentes a elevada altitud utilizando una cohorte de pacientes fallecidos por COVID-19 [Universidad Peruana Cayetano Heredia]. https://hdl.handle.net/20.500.12866/16092
Delgado, C., Modelos predictivos con índice de ROX y variables basales para el uso de ventilación mecánica invasiva en residentes a elevada altitud utilizando una cohorte de pacientes fallecidos por COVID-19 []. PE: Universidad Peruana Cayetano Heredia; 2024. https://hdl.handle.net/20.500.12866/16092
@mastersthesis{renati/911559,
title = "Modelos predictivos con índice de ROX y variables basales para el uso de ventilación mecánica invasiva en residentes a elevada altitud utilizando una cohorte de pacientes fallecidos por COVID-19",
author = "Delgado Flores, Carolina Jaqueline",
publisher = "Universidad Peruana Cayetano Heredia",
year = "2024"
}
The purpose of this secondary study was to build predictive models for the use of invasive mechanical ventilation (IMV) using the index of ROX (iROX) and baseline variables in high-altitude residents (1500 to 3500 masl) using a cohort of patients who died from COVID-19. The iROX has been described as useful for predicting multiple negative outcomes in low-altitude geographic settings. However, its predictive value for IMV, in conjunction with other variables, has not been previously characterized in high-altitude patients. Using data collected between 2020-2021, a secondary analysis was performed to build predictive models using data from a retrospective cohort. Specifically, information from 767 patients who died from COVID-19 in three hospitals in Cusco, Peru was included. Three methods; theoretical or based on previous studies (model A), LASSO (model B), and backward selection (model C) were used in the modeling. In the training subsample (70%), multiple parameters are reported, including ROC curves and areas under the curve (AUC). In the validation sample (30%), internal calibration was performed. Models B (AUC=0.71 [95%CI: 0.66-0.77]; 11 variables) and C (AUC=0.71 [95%CI: 0.65-0.76]; 6 variables) were selected for having better performance and better balance between sensitivity and specificity. The common baseline predictors in both models were age, sex, muscle pain, chest pain, arthralgia, and iROX. Therefore, it is feasible to conclude that both models had a fair performance in predicting IMV use in high-altitude residents. However, both models showed poor performance in internal calibration. In conclusion, this study provides a first approximation of the evaluation of low-cost baseline variables as potential predictors of IMV use. In the future, we recommend the validation of our models.
Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons