Evaluating BLOOMY and SOFA scores in hospitalised patients – Authors' reply

• Published in: The Lancet infectious diseases Vol. 22; no. 5; pp. 592 - 593
• Main Authors: Gladstone, Beryl P, Göpel, Siri, Kern, Winfried V, Tacconelli, Evelina
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.05.2022; Elsevier Limited
• Subjects: Cancer; Hospital Mortality; Humans; Infections; Infectious diseases; Intensive Care Units; Mortality; Organ Dysfunction Scores; Patients; Performance prediction; Prognosis; Retrospective Studies; ROC Curve; Subpopulations; United States > US
• ISSN: 1473-3099; 1474-4457
• DOI: 10.1016/S1473-3099(22)00229-8

The predictive performance assessed with the area under the receiver operating characteristic curve (AUROC) was slightly lower in patients with cancer than in patients without cancer for the BLOOMY 14-day mortality score (p=0·28; appendix), thus resembling the results obtained in the US cohort reported by Benzoni and colleagues. The prognostic performance of the quick SOFA (qSOFA) score was explored outside the intensive care unit (ICU) through a meta-analysis.2 When predicting in-hospital mortality, qSOFA had a pooled sensitivity of 51% (95% CI 39–62) and a pooled specificity of 83% (74–89). The discrimination for in-hospital mortality had an AUROC of 0·74 (95% CI 0·70–0·78).2 Similar findings were observed in patients with cancer in the ICU, for whom the AUROC did not exceed 0·76 for ICU mortality and 0·69 for hospital mortality.3 Another cohort study in a similar patient population found that qSOFA had a lower AUROC than SOFA (0·66 [95% CI 0·56–0·75] vs 0·79 [0·72–0·87]) in predicting 30-day mortality.4 We agree with Benzoni and colleagues on the need for mortality scores to be validated in different subpopulations and especially in those who are immunocompromised.