Early risk assessment of severe outcomes in patients with suspected infection in the emergency department using clinical scores in a multicenter international validation

BACKGROUND: Early identification of patients with suspected infection who are at risk of clinical deterioration in the Emergency Department (ED) is challenging, yet crucial for timely initiation of appropriate care and prevention of clinical deterioration. This multicenter international study evaluates the performance of established scores (SIRS, LODS, qSOFA, SOFA, NEWS) in predicting severe outcomes in ED patients with suspected infection. METHODS: We combined data from five different ED cohort studies (n = 4044 adult patients) from the Netherlands and Sweden. Suspected infection was defined as obtaining a microbiological culture combined with starting antibiotics. The scores were evaluated for their accuracy in predicting the composite outcome of intensive care unit (ICU) admission and in-hospital mortality. FINDINGS: Of 4044 patients, 655 (16 %) experienced a severe outcome, defined as ICU admission (n = 429, 11 %) or in-hospital mortality (n = 327, 8 %). SOFA (AUC 0.75 [95 %CI:0.73-0.77]) and LODS (AUC 0.73 [95 %CI:0.71-0.75]) showed the highest predictive accuracy. NEWS ≥ 5 yielded fair sensitivity (78 %) but low specificity (40 %), whereas NEWS ≥ 7 (sensitivity 71 %, specificity 59 %) performed similarly to SOFA ≥ 2 (69 %, 68 %) and LODS ≥ 2 (70 %, 66 %). In contrast, qSOFA (AUC 0.66 [95 %CI:0.64-0.69]) and SIRS (AUC 0.54 [95 %CI:0.52-0.57]) performed poorly for ED risk stratification. INTERPRETATION: SOFA and LODS showed the highest accuracy but are limited by their complexity and the requirement for laboratory data. The NEWS score, simpler and more accessible, is a practical tool for rapid ED screening with comparable accuracy as SOFA and LODS.