BACKGROUND: Many patients in intensive care units (ICUs) are clinically unstable during the first several days of ICU care. The goal of this study was to describe how the association between selected clinical variables and in-hospital mortality varied across short, sequential time periods during the early phase of ICU care. METHODS: Retrospective analysis of 19,439 ICU encounters among 17,769 patients, aged 18 to 102, who were treated in three ICUs at a single academic medical center during the years 2018 to 2023. Each encounter was segmented into brief intervals of care defined by time between complete blood count results. We analyzed 144,157 time intervals representing the early period of care up to the first 20 consecutive time intervals. For each time interval, we performed multivariable logistic regression to estimate the odds ratios of association between nine selected clinical variables and in-hospital death. Variables included patient demographics, laboratory results, and treatment modalities previously found to be associated with ICU survival in published studies. Both current and previous exposure to life-support treatments (mechanical ventilation, pressors, or dialysis) and interactions among life-support treatments were also investigated as variables. RESULTS: The mean duration of the analyzed time intervals of care was 512 ±215 minutes (median 574, IQR 326–720 minutes). One third of in-hospital mortality among ICU patients occurred during the first 20 time intervals of care. The odds ratios for death for age, sex, platelet count, hemoglobin, and transfusion ranged from 1.01 to 1.35 with small variations across periods of care, whereas the odds ratios for ventilator use, dialysis, pressors, and bilirubin varied more widely from 1.43 to 4.75. CONCLUSION: The strength of association between selected clinical variables and in-hospital mortality varied with time during early periods of ICU care.
Roth et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: