The MAGENTA Model for Individual Prediction of In-Hospital Mortality in Chronic Obstructive Pulmonary Disease With...

The MAGENTA score identifies acute exacerbation of chronic obstructive pulmonary disease (AECOPD) patients at high risk of in-hospital mortality to guide monitoring and treatment, yet its generalizability requires confirmation. This study aimed to externally validate the performance of the MAGENTA model.

We conducted a temporal external validation using retrospective data from 938 admission records of patients hospitalized in general wards and intensive care units at a tertiary center in Thailand (2018-2019). The model, which utilizes mean arterial pressure, age, blood urea nitrogen, endotracheal intubation, sodium, temperature, and albumin, was evaluated regarding all-cause in-hospital mortality.

The validation cohort had an 11.2% mortality rate with moderate case-mix differences compared to the development set. The model demonstrated acceptable discrimination with an area under the curve (AUC) of 0.75 (95% confidence interval (CI), 0.70-0.80), though lower than the AUC in the original derivation. Calibration analysis revealed systematic overprediction (expected-to-observed (E:O) ratio of 1.335) and overfitting (slope of 0.536), particularly when the predicted risk exceeded 20%. Importantly, recalibration of the intercept and slope substantially improved the agreement between predicted and observed risks.

While the MAGENTA model offers acceptable discriminative ability for stratifying AECOPD mortality risk, local recalibration is recommended to address overestimation in high-risk patients.