Time course of energy expenditure in persistent critical illness: a prospective multicentre study

Background: Metabolic alterations are a fundamental part of critical illness, but changes during prolonged ICU stay are inadequately understood. This study aimed to describe longitudinal trends in energy expenditure and substrate utilisation in persistent critical illness, defined as an ICU stay of ≥ 10 days. Methods: This prospective, observational, multicentre study was conducted from 2022 to 2024 at five European and two Australian ICUs. The primary outcome was the change in energy expenditure over time. Adult patients with ≥ 1 indirect calorimetry and length of stay ≥ 10 days were included. Clinical parameters, markers of inflammation and protein catabolism were collected at each measurement. Longitudinal trends were analysed using mixed-effects models with restricted cubic splines. Latent class analysis was performed with identical covariates. Results: 433 patients with 1194 measurements were included. The mean age was 56 years, and 70% were male. An initial increase in energy expenditure, peaking around day 10, and subsequent decline were found (p < 0.001), remaining significant after adjustment for sex, age, CRP, FiO2, presence of fever, BMI, renal replacement therapy, administered protein as fixed effects and patient and site as random effects (p = 0.001, conditional R² = 0.76). The association between the respiratory quotient and time was non-significant (p = 0.067). The urea: creatinine ratio increased over the first 10 days (p < 0.001). Latent class analysis identified three trajectories of energy expenditure: hypo-, normo-, and hypermetabolism (entropy 0.63). Conclusions: Mean energy expenditure follows a biphasic pattern during prolonged ICU stay, with the inflection point coinciding with the empirical onset of persistent critical illness. Further research is required to validate potential metabolic subgroups and explore their biological correlates in this population. Trial registration: The study, including the statistical analysis plan, was prospectively registered at clinicaltrials.gov (NCT05124860, registered 2021-11-15).