# Characterizing ICU-like profiles of very old patients hospitalized in medical intermediate care units in France: A clustering analysis of a nationwide population-based study

## TL;DR

This study identifies patterns of care for very old patients in intermediate care units in France and finds that age and comorbidities strongly influence admission decisions.

## Contribution

The study introduces a clustering approach to characterize ICU-like profiles in very old patients in intermediate care units and compares them to ICU patients.

## Key findings

- Seven distinct phenotypes of very old patients hospitalized in IMCUs were identified.
- Age and comorbidity status were the strongest predictors for IMCU admission over ICU.
- An 'ICU-like' group accounted for 11.6% of the cohort with high mortality and organ support.

## Abstract

The population aging has a growing impact on intensive care. The admission policies for critically ill very old patients in intermediate care units (IMCUs) are unclear. This study aimed to identify clusters of very old patients hospitalized in IMCUs in France, and to compare “ICU-like” profiles among these patients with matched very old patients admitted to ICU patients, in order to better understand the factors influencing admission to an IMCU rather than an ICU.

We conducted a nationwide study and examined all hospitalisations in IMCUs in France over a two-year period. We studied the utilization of medical IMCU for all critically ill patients ≥80 y.o and reported patient characteristics, including the Charlson comorbidity index and Hospital Frailty Risk Score. Latent class analysis identified distinct IMCU phenotypes. Clusters with high-use organ support (≥5%) and high mortality (≥40%) were considered “ICU-like.” Logistic regression compared “ICU-like” IMCU patients with propensity-matched ICU patients to identify factors associated with IMCU admission.

Among the 202,976 very old individuals hospitalized in IMCU, seven phenotypes were identified. The "ICU-like" population accounted for 11.6% of the cohort (n = 23,508). After propensity score matching, the multivariate logistic regression identified age as the strongest determinant to IMCU admission, with very old patients being significantly more likely to be admitted to the IMCU over the ICU. Comorbidity status was also an independent predictor for admission to IMCU. Frailty status showed only a marginal effect, and no association was detectable for sex.

A substantial number of very old patients were admitted to IMCU despite meeting ICU-level criteria. Clearer admission criteria and objectives for IMCU care in older critically ill patients are urgently needed.