Persistent critical illness (PerCI) denotes patients who survive the initial acute phase of critical illness but remain dependent on ICU-level care because recovery is delayed or fails. PerCI has been linked to persistent inflammation, immunosuppression, and catabolism and has previously been reported to affect 5–20% of general ICU populations while consuming 40–60% of ICU costs. One-year mortality in prior PerCI cohorts has been high (estimated 40–67%), and fewer than 12% of survivors return to full independence. Existing literature largely addresses general ICU cohorts; disease-specific estimates are limited.
Patients with acute hypoxic respiratory failure (AHRF) requiring invasive mechanical ventilation represent a clinically coherent high-risk subgroup in which PerCI may be particularly common. Mechanical ventilation and hypoxemia introduce additional risks for ICU-associated complications and multi-organ dysfunction that can prolong ICU dependence beyond resolution of the original respiratory failure. Prior work on prolonged ventilation does not fully capture the PerCI trajectory, because many patients remain ICU-dependent due to downstream complications even after extubation.
The objectives of this study were to quantify PerCI prevalence in mechanically ventilated AHRF patients, identify clinical factors associated with developing PerCI while accounting for competing early mortality, and describe one-year survival in this population.
Study design and setting
This was a retrospective cohort study including adult patients (age ≥18 years) admitted to ICUs across a 12-hospital health system (two academic and ten community hospitals) from January 2011 to December 2022. We included only the index ICU admission for each patient. The institutional review board at the University of Minnesota approved the study (STUDY00014815).
Patient selection and PerCI definition
Eligible patients required invasive mechanical ventilation within 72 hours of ICU admission and had a PaO2/FiO2 (P/F) ratio <300 to enrich for intubations primarily for hypoxemia. The P/F ratio used the lowest daily value, matching PaO2 to the closest FiO2 within a ±1-hour window when available. PerCI was defined a priori as ICU length of stay ≥10 days, consistent with prior literature and prior validation in the authors’ general ICU population. The authors note that some patients with severe ARDS may still be in their acute phase at day 10; to mitigate misclassification, they used a dual-model approach (described below) to distinguish factors linked to prolonged dependence from those reflecting acute severity.
Data sources and variables
Data were abstracted from electronic health records and included age, sex, race/ethnicity, admission code status, primary service (medicine vs surgical), ICU source (ED vs ward), admission diagnoses (ICD-10), Elixhauser comorbidities, admission LAPS2 score (worst), tobacco use, worst daily P/F ratio, duration of invasive mechanical ventilation, and discharge disposition. Mortality beyond hospitalization was captured monthly via Department of Health records.
Statistical analysis
Descriptive statistics used means and SDs for normally distributed continuous variables, medians and IQRs for skewed variables, and frequencies and percentages for categorical variables. Group comparisons used ANOVA, Kruskal-Wallis, or Chi-square tests as appropriate. Kaplan–Meier curves and log-rank tests compared unadjusted survival between PerCI and non-PerCI patients. Missing data were minimal (<5% for all covariates) and handled with complete-case analysis. All analyses were performed in SAS 9.4; p-values <0.05 were considered statistically significant.
Approach to identifying risk factors
To address competing risks from early mortality, the study implemented a dual-model logistic approach: one logistic regression compared PerCI patients to those who died early (≤10 days) and another compared PerCI patients to those discharged from the ICU early. Variables significantly associated with PerCI in both models were considered robust predictors of developing persistent ICU dependence, distinguishing them from factors primarily reflecting acute illness severity.
Primary outcome
The primary outcome was one-year mortality. A Cox proportional hazards model treated PerCI as a time-varying exposure to account for the timing of PerCI onset relative to mortality risk.
Cohort and PerCI prevalence
A total of 10,626 patients met inclusion criteria. Of these, 3,107 patients (29.2%) met the study definition of PerCI (ICU length of stay ≥10 days). This prevalence substantially exceeds previously reported PerCI rates in general ICU populations.
One-year survival and adjusted mortality risk
Observed one-year survival was 64.2% in the PerCI group compared with 73.8% in the non-PerCI group. In the Cox model with PerCI as a time-varying exposure, onset of PerCI was associated with a threefold increase in mortality risk (hazard ratio 3.01; 95% confidence interval 2.72–3.33).
Risk factors associated with PerCI
Using the dual-model approach to account for competing early deaths and early discharges, the study identified several clinical factors consistently associated with the development of PerCI. These included ARDS, pneumonia, aspiration pneumonitis, heart failure, septic shock, documented weight loss, surgical admission, and postprocedural respiratory and circulatory failure. Factors that were significant in only one model were treated with more caution; the authors prioritized variables that were significant across both comparisons as robust predictors.
Additional findings and data handling
Duration of invasive mechanical ventilation and worst P/F ratios were included in data abstraction, and discharge dispositions were recorded. Missing covariate data were under 5% and managed with complete-case analysis. The study reports that mortality capture included deaths occurring after hospital discharge via Department of Health linkage.
In this multicenter retrospective cohort of mechanically ventilated AHRF patients, nearly one-third developed persistent critical illness, a rate notably higher than estimates from general ICU cohorts and from prior disease-specific samples reported in the literature. Development of PerCI was associated with substantially worse one-year survival, with an adjusted threefold increase in mortality risk after PerCI onset.
The identified risk factors—particularly ARDS, pneumonia, aspiration pneumonitis, septic shock, heart failure, weight loss, surgical admission, and postprocedural respiratory/circulatory failure—can help clinicians identify patients at elevated risk for prolonged ICU dependence. Because PerCI consumes a disproportionate share of ICU resources and has poor long-term outcomes, earlier recognition of high-risk patients may support timely goals-of-care conversations, targeted rehabilitation planning, and allocation of post-ICU resources.
The authors acknowledge limitations inherent to retrospective design and potential misclassification of PerCI for some patients whose primary acute illness may not have resolved by day 10. They sought to mitigate this concern using a validated 10-day threshold and a dual-model analytic strategy to separate markers of ongoing acute severity from predictors of persistent dependence.
Among 10,626 mechanically ventilated patients with AHRF, 29.2% developed PerCI (ICU length of stay ≥10 days). PerCI was associated with lower one-year survival (64.2% vs 73.8%) and a threefold increase in mortality risk after onset (HR 3.01; 95% CI 2.72–3.33). Several identifiable clinical factors were robustly associated with PerCI development. The authors propose that these findings could inform early goals-of-care discussions and post-ICU care planning for this high-risk population.