JMIR Publications

ABSTRACT

Background:

Considering morbidity, mortality, and annual treatment costs, the dramatically increasing incidents of sepsis and septic shock among ICU admissions in the United States hospitals are an increasing concern. The recent changes in sepsis definition (sepsis-3) motivate the medical informatics research community around the world to investigate score recalculation and information retrieval and study the intersection between Sepsis-3 and Sepsis-2.

Objective:

The objective of this study is threefold. First, we aim to unpack the most prevalent criterion for sepsis (for both sepsis-3 and sepsis-2 predictors). Second, we intend to determine the most prevalent sepsis scenario in the ICU (among four possible scenarios for sepsis-3 and eleven possible scenarios for sepsis-2). Third, we investigate the multicollinearity or dichotomy among sepsis-3 predictors and sepsis-2 predictors.

Methods:

We conducted this observational study using MIMIC-III: the critical care database of MIT. We took the qSOFA (sepsis-3) and SIRS (sepsis-2) parameters into account for patients admitted to the critical care units (2001-2012) in Beth Israel Deaconess Medical Center in Boston to understand the prevalence and underlying relation between these parameters among the patients undergone sepsis screening. We adopted a multi blind Delphi method to seek a rationale for decisions in several stages of research design regarding handling missing data and outlier values, statistical imputations and biases, and generalizability of the study.

Results:

This study reveals that the altered mental status in Glasgow Coma Scale (59.09%) is the most prevalent Sepsis-3 (qSOFA) criterion, and the white blood cell count (57.11%) is the most prevalent Sepsis-2 (SIRS) criterion confronted in the ICU. Besides, the two-factored sepsis of high respiratory rate (RR ≥ 22 bpm) and altered mental status (43.39%) is the most prevalent sepsis-3 (qSOFA) scenario, and the two-factored sepsis of tachypnea and high white blood cell count (27.83%) is the most prevalent sepsis-2 (SIRS) scenario in the ICU. Apart from that, the insignificant absolute Pearson correlation coefficients nullify the likelihood of any linear correlation among the critical parameters, thereby assuring that multicollinearity does not exist between the parameters. It, further, bolsters the dichotomy among them. However, the absence of multicollinearity cannot guarantee that two random variables are statistically independent.

Conclusions:

Quantifying the prevalence of the qSOFA criteria of sepsis-3 (so as SIRS of sepsis-2) and understanding the underlying dichotomy among the parameters have significant inferences for sepsis treatment initiatives in ICU and informing hospital resource allocation. Besides, these data-driven results offer design implications for multi-parameter intelligent sepsis prediction in the ICU.