Derivation and performance of an end-of-life practice score aimed at interpreting worldwide treatment-limiting decisions in the critically ill

Abstract Background Limitations of life-sustaining interventions in intensive care units (ICUs) exhibit substantial changes over time, and large, contemporary variation across world regions. We sought to determine whether a weighted end-of-life practice score can explain a large, contemporary, worldwide variation in limitation decisions. Methods The 2015–2016 (Ethicus-2) vs. 1999–2000 (Ethicus-1) comparison study was a two-period, prospective observational study assessing the frequency of limitation decisions in 4952 patients from 22 European ICUs. The worldwide Ethicus-2 study was a single-period prospective observational study assessing the frequency of limitation decisions in 12,200 patients from 199 ICUs situated in 8 world regions. Binary end-of-life practice variable data (1 = presence; 0 = absence) were collected post hoc (comparison study, 22/22 ICUs, n = 4592; worldwide study, 186/199 ICUs, n = 11,574) for family meetings, daily deliberation for appropriate level of care, end-of-life discussions during weekly meetings, written triggers for limitations, written ICU end-of-life guidelines and protocols, palliative care and ethics consultations, ICU-staff taking communication or bioethics courses, and national end-of-life guidelines and legislation. Regarding the comparison study, generalized estimating equations (GEE) analysis was used to determine associations between the 12 end-of-life practice variables and treatment limitations. The weighted end-of-life practice score was then calculated using GEE-derived coefficients of the end-of-life practice variables. Subsequently, the weighted end-of-life practice score was validated in GEE analysis using the worldwide study dataset. Results In comparison study GEE analyses, end-of-life discussions during weekly meetings [odds ratio (OR) 0.55, 95% confidence interval (CI) 0.30–0.99], end-of-life guidelines [OR 0.52, (0.31–0.87)] and protocols [OR 15.08, (3.88–58.59)], palliative care consultations [OR 2.63, (1.23–5.60)] and end-of-life legislation [OR 3.24, 1.60–6.55)] were significantly associated with limitation decisions (all P