Feasibility of In-Hospital Administration of a Tool to Predict Persistent Post-ICU Functional Impairment Among Older ICU Survivors

Background

A recent international consensus conference called for the development of risk prediction models to identify ICU survivors at increased risk of each of the post-ICU syndrome domains. We previously developed and validated a risk prediction tool for functional impairment after ICU admission among older adults.

Research Question

In this pilot study, we assessed the feasibility of administering the risk prediction tool in the hospital to older adults who had just survived critical illness. An exploratory objective was to evaluate whether augmentation of the model with additional hospital-related factors improved discrimination.

Study Design and Methods

Between January and October 2020, 50 adults aged 65 years and older underwent in-hospital administration of the risk prediction tool. Survivors were called monthly for 6 months after discharge. Feasibility was defined as completion of all tool components by ≥ 70% of enrolled participants. Persistent functional impairment was defined as failure to return to the functional baseline from before the ICU stay at the 6-month interview based on seven daily activities. The model was sequentially refit after adding three in-hospital factors as predictors, one at a time and then all together. Model discrimination was assessed with receiver operating characteristic curves.

Results

The tool met the a priori feasibility threshold, with 92.0% of enrolled participants completing all eight components. In the exploratory analysis, the addition of Acute Physiology and Chronic Health Evaluation II score, presence of delirium, and maximum in-hospital mobility resulted in a 5% gain in discrimination that did not achieve statistical significance (area under the receiver operating characteristic curve, 0.75; 95% CI, 0.68-0.82; P = .09).

Interpretation

Our results indicate that the risk prediction tool is feasible for use in the hospital setting, enabling the identification of ICU survivors at high risk of persistent functional impairment at 6 months after discharge. Augmentation with hospital-related factors improved model discrimination, but did not achieve statistical significance in this pilot study. Future studies should evaluate the augmented model in larger cohorts.