The handling oxygenation targets in the intensive care unit (HOT‐ICU) trial: Detailed statistical analysis plan

Background No solid evidence exists on optimal oxygenation targets in intensive care patients. The handling oxygenation targets in the intensive care unit (HOT‐ICU) trial assesses the effects of a targeted arterial oxygen tension of 8 vs 12 kPa on 90‐day mortality in acutely admitted adult patients with hypoxaemic respiratory failure. This article describes the detailed statistical analysis plan for the predefined outcomes and supplementary analyses in the HOT‐ICU trial. Methods The trial will include 2928 patients to be able to detect or reject a true 20% relative risk reduction in the primary outcome of 90‐day all‐cause mortality with an α of 5% and a β of 10%. Analyses of the primary and secondary outcomes will be conducted according to the intention‐to‐treat principle and adjusted for stratification variables. The primary outcome and dichotomous secondary outcomes will be analysed using a generalised linear model with a log‐link and binomial error distribution. For the primary outcome, a 95% confidence interval (CI) not including 1.00 for the risk ratio will be considered statistically significant. Continuous secondary outcomes will be analysed using a generalised linear model or nonparametric test. CIs adjusted for the multiple secondary outcomes not including the null effect will be considered statistically significant. One planned interim analysis has been conducted. Conclusions The HOT‐ICU trial and the pre‐planned statistical analyses are designed to minimise bias and produce high quality data on the effects of a lower vs a higher oxygenation target throughout ICU admission in acutely admitted adult patients with hypoxaemic respiratory failure. Registration ClinicalTrials.gov identifier: NCT03174002, date of registration: June 2, 2017. European clinical trials database, EudraCT number 2017‐000632‐34