Browsing by Author "Marques, Daniela Andrade"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
- Traditional versus intentionally created severity scores for the prognosis of COVID-19 patients in PortugalPublication . Marques, Daniela Andrade; Pinto, Iola Maria Silvério; Bento, LuísAbstract Context - The COVID-19 pandemic motivated the development of prognostic tools for critically ill patients. The calculation and analysis of traditional severity scores and scores created specifically for COVID-19 allow evaluating their discriminatory capacity in decision-making in a clinical context and predicting mortality in patients with COVID-19 allocated to intensive care units (ICU). Purpose - The goal was to compare the performance of traditional ICU severity scores (APACHE II, SAPS II, SAPS 3, SOFA) with scores developed specifically for COVID-19 (Shang-COVID, SEIMC, BURDEN, and inflammation-based), in predicting four mortality outcomes in ICU patients during the first and second waves of the pandemic in Portugal. Methods - Data from adult patients with COVID-19 admitted to the ICU at the Centro Hospitalar Universitário Lisboa Central in Lisbon, Portugal, were analysed. Eight severity scores were calculated for each patient, and four outcomes were considered: hospital mortality, ICU mortality, early ICU mortality (death within 7 days of admission), and late ICU mortality (death after 7 days of admission). The discriminative ability of the scores was evaluated through ROC curve area estimates, their respective confidence intervals, and p-values. Results - For hospital mortality, SEIMC performed best with AUCs of 0.810 (first wave) and 0.723 (second wave). APACHE II and SAPS 3 also showed good AUCs (>0.7), while BURDEN and Inflammation-base had AUCs below 0.6. For ICU mortality, SEIMC stood out with an AUC of 0.808 in the first wave. SAPS 3 and APACHE II also demonstrated good discriminative ability (>0.7). In the second wave, SEIMC had an AUC of 0.705. For early ICU death, SAPS 3 was the most effective in the second wave (AUC of 0.828), followed by SEIMC, while BURDEN and INFLAMMATION-BASED showed low discriminative ability. For late ICU mortality, SEIMC and SAPS 3 stood out in the first wave, with discriminative ability decreasing in the second wave. Conclusion - While traditional scores are relevant for predicting outcomes in COVID-19 patients, the newly developed SEIMC score demonstrated more consistent discriminative ability across all four outcomes. The good performance of traditional scores indicates that traditional scores can still be reliable in this context, emphasizing the importance of continuously validating severity scores to improve patient care and resource allocation.