In Search of the Ideal Risk Score in Sepsis

Abstract

To the Editor: We read with great interest the recent article by Machado and colleagues (1) revealing low sensitivity of the quick Sequential Organ Failure Assessment (qSOFA) score ≥2 in predicting mortality among emergency department and ward patients with suspected infection or sepsis and that using qSOFA ≥1 and qSOFA ≥1 together with lactate improved sensitivity. Being from a middle- to upper-income country comparable with Brazil, we performed an observational retrospective cohort study in a tertiary public university hospital in Turkey to evaluate and compare the predictive roles of qSOFA and SOFA scores, systemic inflammatory response syndrome (SIRS) criteria, and Modified Early Warning Score (MEWS) (2, 3) obtained during the 48 hours before ICU admission for hospital mortality. A total of 120 patients admitted to the medical ICU from the emergency department or wards between January 1 and May 31, 2018, with suspected infection were included. The hospital mortality rate was 33%. Sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC) (95% confidence interval) of qSOFA ≥2 were 72.7% (54.2–86.0), 47.1 (36.4–58.0), and 0.60 (0.49–0.71), respectively. The corresponding values for SOFA ≥2 were 97.0 (82.4–99.8), 37.2 (22.7–43.1), and 0.65 (0.54–0.75), respectively; for SIRS ≥2, they were 87.8 (70.8–96.0), 12.6 (6.7–21.9), and 0.50 (0.39–0.62), respectively; and for MEWS ≥4, they were 84.8 (67.3–94.2), 42.5 (32.1–53.5), and 0.64 (0.53–0.74), respectively. In this study, the sensitivity of qSOFA with the standard cutoff value of 2 was the lowest among all scores; therefore, its use as a screening tool and mortality predictor might not be sufficient. qSOFA was introduced as a mortality prediction tool on the basis of North American and European cohorts with an area under the curve of 0.81 for patients outside the ICU (4). However, in a large study in patients admitted to the ICU in Australia and New Zealand (5), in which investigators used the scores calculated within the first 24 hours of ICU admission, SOFA had the greatest prognostic accuracy (AUROC, 0.75), with qSOFA and SIRS having AUROCs of 0.61 and 0.59, respectively. Early warning scores could also be more accurate than qSOFA scores for predicting mortality and ICU transfer. In a recent study by Churpek and colleagues (6), qSOFA was found to be less accurate than early warning scores for predicting in-hospital mortality in non-ICU patients with suspicion of infection. qSOFA score greater than or equal to 2 had a sensitivity of 68.7%, specificity of 63.5%, and AUROC of 0.69 (0.67–0.70), whereas the AUROC was 0.77 (0.76–0.79) for the National Early Warning Score and 0.73 (0.71–0.74) for MEWS. Though the authors conducted a single-center study, together with the other studies, the accuracy of the qSOFA score as a risk score remains questionable. SOFA and early warning scores seem to be better mortality predictors.