A targeted real-time early warning score (TREWScore) for septic shock

Abstract

Sepsis is a leading cause of death in the United States, with mortality highest among patients who develop septic shock. Early aggressive treatment decreases morbidity and mortality. Although automated screening tools can detect patients currently experiencing severe sepsis and septic shock, none predict those at greatest risk of developing shock. We analyzed routinely available physiological and laboratory data from intensive care unit patients and developed "TREWScore," a targeted real-time early warning score that predicts which patients will develop septic shock. TREWScore identified patients before the onset of septic shock with an area under the ROC (receiver operating characteristic) curve (AUC) of 0.83 [95% confidence interval (CI), 0.81 to 0.85]. At a specificity of 0.67, TREWScore achieved a sensitivity of 0.85 and identified patients a median of 28.2 [interquartile range (IQR), 10.6 to 94.2] hours before onset. Of those identified, two-thirds were identified before any sepsis-related organ dysfunction. In comparison, the Modified Early Warning Score, which has been used clinically for septic shock prediction, achieved a lower AUC of 0.73 (95% CI, 0.71 to 0.76). A routine screening protocol based on the presence of two of the systemic inflammatory response syndrome criteria, suspicion of infection, and either hypotension or hyperlactatemia achieved a lower sensitivity of 0.74 at a comparable specificity of 0.64. Continuous sampling of data from the electronic health records and calculation of TREWScore may allow clinicians to identify patients at risk for septic shock and provide earlier interventions that would prevent or mitigate the associated morbidity and mortality.

Prediction using patient comparison vs. modeling: A case study for mortality prediction

“…To accomplish this, we iteratively remove time frames of 4 hours starting from the moment of death and work backwards. This is in line with [3]. We obtain an AUC on the set aside test set of 0.78 from the Cox model, substantially lower than the performance of the logistic regression model.…”

“…Such methods have also been used to estimate how long in advance certain predictions can be made (cf. [3]). Another category of approaches has focused on defining patient similarity, driving more instance-based learning (e.g., k-nearest neighbor) approaches.…”

Surviving sepsis campaign: research priorities for sepsis and septic shock

“…Recent advance in omics (genomics, epigenomics, transcriptomics, proteomics, metabolomics, pharmacogenomics, microbiomics) have the potential to revolutionize care by assaying the state of an individual [78,79]. Individual insights need not be confined to "omics"-based data, however, as important insights can be drawn from easily interpretable clinical information and by use of big data approaches that allow insight from information accessible within the ICU that might not be able to be processed by a bedside provider [80].…”

Flexible, cluster-based analysis of the electronic medical record of sepsis with composite mixture models

“…Other approaches have been able to recover clinically relevant phenotypes with striking prediction capabilities for a wide range of medical conditions [14,18]. Particularly in sepsis, supervised approaches have demonstrated remarkable performance in early identification of patients at risk of entering septic shock [17,16]. In essence, these models characterize the effects of a fixed set of predictor variables or features on an outcome of interest ( supervised learning; a conditional model), without directly modeling the features themselves.…”