Improving the accuracy of cardiovascular component of the sequential organ failure assessment score

Abstract

Objectives: The Sequential Organ Failure Assessment score is an attractive risk prediction model because of its simplicity and graded assessment of morbidity and mortality. Due to changes in clinical practice over time, the cardiovascular component of the Sequential Organ Failure Assessment score no longer accurately reflects current clinical practice. To address this limitation, we developed and validated a modified cardiovascular component of the Sequential Organ Failure Assessment score that takes into account all vasoactive agents used in current clinical practice, uses shock index as a substitute for mean arterial pressure, and incorporates serum lactate as a biomarker for shock states. Design: Retrospective cohort. Setting: Mayo Clinic, Rochester, MN. Patients: Adult patients admitted to one of six ICUs. Interventions: None. Measurements and Main Results: Score performance was assessed via area under the receiver operator characteristic curve. A total of 16,386 ICU admissions were included: 9,204 in the derivation cohort and 7,182 in the validation cohort. area under the receiver operator characteristic curve was significantly higher for modified cardiovascular component of the Sequential Organ Failure Assessment score than for cardiovascular component of the Sequential Organ Failure Assessment for in-ICU mortality (0.801 vs 0.718; difference = 0.083; p < 0.001), in-hospital mortality (0.783 vs 0.651; difference = 0.132; p < 0.001), and 28-day mortality (0.737 vs 0.655; difference = 0.082; p < 0.001). When modified cardiovascular component of the Sequential Organ Failure Assessment score was added to the remaining Sequential Organ Failure Assessment components, the modified Sequential Organ Failure Assessment score again outperformed the existing Sequential Organ Failure Assessment score: in-ICU mortality (0.836 vs 0.822; difference = 0.014; p < 0.001), in-hospital mortality (0.799 vs 0.784; difference = 0.015; p < 0.001), and 28-day mortality (0.798 vs 0.783; difference = 0.015; p < 0.001). Similar results were seen in the validation cohort. Conclusions: The modified cardiovascular component of the Sequential Organ Failure Assessment score outperforms the existing cardiovascular component of the Sequential Organ Failure Assessment score in predicting patient outcomes and improves the overall performance of the Sequential Organ Failure Assessment model. This score is easily calculated, includes serum lactate as a biomarker for shock states, and incorporates all vasopressors used in current clinical practice.