Persistent critical illness and long-term outcomes in patients with COVID-19: A multicenter retrospective cohort study.

Abstract

OBJECTIVES:

A nontrivial number of patients in ICUs experience persistent critical illness (PerCI), a phenomenon in which features of the ICU course more consistently predict mortality than the initial indication for admission. We aimed to describe PerCI among patients with critical illness caused by COVID-19, and these patients’ short- and long-term outcomes. DESIGN:

Multicenter retrospective cohort study. SETTING:

Australian and New Zealand Intensive Care Society Adult Patient Database of 114 Australian ICUs between January 1, 2020, and March 31, 2022. PATIENTS:

Patients 16 years old or older with COVID-19, and a documented ICU length of stay. EXPOSURE:

The presence of PerCI, defined as an ICU length of stay greater than or equal to 10 days. MEASUREMENTS:

We compared the survival time up to 2 years from ICU admission using time-varying robust-variance estimated Cox proportional hazards models. We further investigated the impact of PerCI in subgroups of patients, stratifying based on whether they survived their initial hospitalization. MAIN RESULTS:

We included 4961 patients in the final analysis, and 882 patients (17.8%) had PerCI. ICU mortality was 23.4% in patients with PerCI and 6.5% in those without PerCI. Patients with PerCI had lower 2-year (70.9% [95% CI, 67.9–73.9%] vs. 86.1% [95% CI, 85.0–87.1%]; p < 0.001) survival rates compared with patients without PerCI. Patients with PerCI had higher mortality (adjusted hazards ratio: 1.734; 95% CI, 1.388–2.168); this was consistent across several sensitivity analyses. When analyzed as a nonlinear predictor, the hazards of mortality were inconsistent up until 10 days, before plateauing. CONCLUSIONS:

In this multicenter retrospective observational study patients with PerCI tended to have poorer short-term and long-term outcomes. However, the hazards of mortality plateaued beyond the first 10 days of ICU stay. Further studies should investigate predictors of developing PerCI, to better prognosticate long-term outcomes.