Did COVID-19 ICU patient mortality risk increase as Colorado hospitals filled? A retrospective cohort study.

Objectives: To assess whether increasing levels of hospital stress-measured by intensive care unit (ICU) bed occupancy (primary), ventilators in use and emergency department (ED) overflow-were associated with decreasing COVID-19 ICU patient survival in Colorado ICUs during the pre-Delta, Delta and Omicron variant eras.

Design: A retrospective cohort study using discrete-time survival models, fit with generalised estimating equations.

Setting: 34 hospital systems in Colorado, USA, with the highest patient volume ICUs during the COVID-19 pandemic.

Participants: 9196 non-paediatric SARS-CoV-2 patients in Colorado hospitals admitted once to an ICU between 1 August 2020 and 1 March 2022 and followed for 28 days.

Outcome Measures: Death or discharge to hospice.

Results: For Delta-era COVID-19 ICU patients in Colorado, the odds of death were estimated to be 26% greater for patients exposed every day of their ICU admission to a facility experiencing its all-era 75th percentile ICU fullness or above, versus patients exposed for none of their days (OR: 1.26; 95% CI: 1.04 to 1.54; p=0.0102), adjusting for age, sex, length of ICU stay, vaccination status and hospital quality rating. For both Delta-era and Omicron-era patients, we also detected significantly increased mortality hazard associated with high ventilator utilisation rates and (in a subset of facilities) states of ED overflow. For pre-Delta-era patients, we estimated relatively null or even protective effects for the same fullness exposures, something which provides a meaningful contrast to previous studies that found increased hazards but were limited to pre-Delta study windows.

Conclusions: Overall, and especially during the Delta era (when most Colorado facilities were at their fullest), increasing exposure to a fuller hospital was associated with an increasing mortality hazard for COVID-19 ICU patients.

Competing Interests: Competing interests: None declared.

(© Author(s) (or their employer(s)) 2024. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.)