Hierarchical endpoint analysis using win ratio in critical care: An exploration using the balanced solutions in intensive care study (BaSICS).

Abstract

To reanalyze the results of the Balanced Solutions in Intensive Care Study (BaSICS) through hierarchical endpoint analysis with win ratio. All patients with full data in BaSICS trial were elected for the analysis. BaSICS compared balanced solutions (Plasma Lye 148) versus 0.9% saline in critically ill patients requiring fluid challenge. The win ratio was defined as a hierarchical endpoint of 90-day mortality, recepit of kidney replacement therapy, hospital length-of-stay (LOS), and intensive care unit (ICU) LOS. Both unstratified and stratified (by admission type: planned admission, unplanned admission with sepsis, and unplanned admission without sepsis) approaches were used. A subgroup analysis was performed in patients with traumatic brain injury. A total of 10,490 patients were included in the analysis, resulting in 27,587,566 unique combinations for unstratified WR. Unstratified Win ratio was 1.02 (95% confidence interval 0.97; 1.07), which was similar to stratified WR. No stratum in the stratified analysis resulted in significant results. Subgroup analysis confirmed the possible harm of balanced solutions in traumatic brain injury patients (WR 0.80; 95% confidence interval 0.64; 0.99). In this reanalysis of BaSICS, a win ratio analysis largely replicated the results of the main trial, yielding neutral results except for the subgroup of patients with traumatic brain injury where a signal of harm was found.