• Toshifumi Asada, Yuta Aoki, Takehiro Sugiyama, Miyuki Yamamoto, Takeshi Ishii, Youichi Kitsuta, Susumu Nakajima, Naoki Yahagi, and Kent Doi.
• 1Department of Emergency and Critical Care Medicine, the University of Tokyo, Tokyo, Japan.2Department of Neuropsychiatry, School of Medicine, the University of Tokyo, Tokyo, Japan.3Child Study Center, Langone Medical Center, New York University, New York, NY.4Department of Clinical Study and Informatics, Center for Clinical Sciences, National Center for Global Health and Medicine, Tokyo, Japan.5Department of Public Health/Health Policy, the University of Tokyo, Tokyo, Japan.
• Crit. Care Med. 2016 Jan 1; 44 (1): 83-90.

ObjectivesAs interactions of each organ system have been conceptually known to play an important role during life-threatening conditions, we quantitatively evaluated the organ system interactions in critically ill patients and examined the difference in the organ system network structure between the survivors and the nonsurvivors.DesignProspective observational study.SettingsAn ICU of a university hospital.PatientsTwo hundred and eighty-two patients who were admitted to the ICU.InterventionsBlood samples were obtained at ICU admission.Measurements And Main ResultsWe analyzed the associations among nine representative laboratory variables of each organ system using network analysis. We compared the network structure of the variables in the 40 nonsurvivors with that in the 40 survivors. Their baseline characteristics, including the degree of organ dysfunction, were matched using propensity score matching method. Network structure was quantitatively evaluated using edge (significant correlation among variables evaluated by the p value), weight (connective strength of edge evaluated by coefficient), and cluster (group with tight connection evaluated by edge betweenness). The number of edges among the nine variables was significantly fewer for the nonsurvivors than for the severity-matched survivors (3 vs 12; p = 0.035). The mean weight of edges was significantly smaller for the nonsurvivors (0.055 vs 0.119; p = 0.007). The nine laboratory variables for the nonsurvivors were divided into a significantly larger number of clusters (7 vs 2; p = 0.001). Statistical conclusions were preserved with Bonferroni multiple comparison procedure. These findings were consistently observed in comparison of the 40 nonsurvivors with all the survivors.ConclusionsThis study, as a preliminary proof-of-concept, quantitatively demonstrated a more disrupted network structure of organ systems in the nonsurvivors compared with that in the survivors. These observations suggest the necessity of assessment for organ system interactions to evaluate critically ill patients.

12 keywords...

hide…