Article Text
Abstract
Aims, Objectives and Background Reliable risk assessment in patients presenting to emergency departments (ED) with suspected infection is of utmost importance to support clinical decisions. Vital sign-based scoring systems such as NEWS2 or qSOFA enable a rapid first assessment of patient urgency at triage. However, their inherent high sensitivity might drive over-utilization of healthcare resources. Our aim was to evaluate if adding the result of a transcriptomic severity classifier can synergistically improve current score-based risk assessment in the ED.
Method and Design We performed a secondary analysis of a patient cohort (n=312) enrolled in the Charité University hospital ED (Berlin, Germany) with suspected infection and at least one vital sign alteration. The expression of 29-host mRNAs in PAXgene-stabilized whole blood was quantified using NanoString nCounter® SPRINT. The proprietary machine learning classifier IMX-SEV-3 was applied to calculate a score that falls into pre-defined interpretation bands: low/moderate/high severity. NEWS2 and qSOFA were documented on admission and combined with the classifier results to analyze the incidence of two clinical endpoints: ‘need for critical care’ (composite of need for ventilation, dialysis, and/or vasopressors) within 7d and ‘28d mortality’.
Results and Conclusion Among enrolled patients, 22 (7.1%) died and 66 (21.1%) required ICU-level care. Of patients with a high NEWS2 (≥5 points; n=184), there was a stepwise increase in mortality among the low (0%; n=0/47), medium (10.1%; n=12/119) and high (44.4% n=8/18) IMX-SEV-3 severity subgroups. A similar stratification was achieved across the low (17%), moderate (31%), and high (61%) IMX-SEV-3 subgroups for prediction of critical care. More granular risk stratification could also be confirmed when using IMX-SEV-3 in combination with high qSOFA (≥2 points; n=76): 0/10.6/50% mortality and 23.5/40.4/66% need for critical care in the low/moderate/high subgroups, respectively.
In summary, the combined use of immune-based IMX-SEV-3 results for ED patients with high clinical scores allows improved prediction of mortality and the need for critical care.