Responsiveness of Critically Ill Adults With Multimorbidity to Rehabilitation Interventions: A Patient-Level Meta-Analysis Using Individual Pooled Data From Four Randomized Trials

Abstract

Objective: To explore if patient characteristics (pre-existing comorbidity, age, sex, and illness severity) modify the effect of physical rehabilitation (intervention vs control) for the coprimary outcomes health-related quality of life (HRQoL) and objective physical performance using pooled individual patient data from randomized controlled trials (RCTs).

Data sources: Data of individual patients from four critical care physical rehabilitation RCTs.

Study selection: Eligible trials were identified from a published systematic review.

Data extraction: Data sharing agreements were executed permitting transfer of anonymized data of individual patients from four trials to form one large, combined dataset. The pooled trial data were analyzed with linear mixed models fitted with fixed effects for treatment group, time, and trial.

Data synthesis: Four trials contributed data resulting in a combined total of 810 patients (intervention n = 403, control n = 407). After receiving trial rehabilitation interventions, patients with two or more comorbidities had HRQoL scores that were significantly higher and exceeded the minimal important difference at 3 and 6 months compared with the similarly comorbid control group (based on the Physical Component Summary score (Wald test p = 0.041). Patients with one or no comorbidities who received intervention had no HRQoL outcome differences at 3 and 6 months when compared with similarly comorbid control patients. No patient characteristic modified the physical performance outcome in patients who received physical rehabilitation.

Conclusions: The identification of a target group with two or more comorbidities who derived benefits from the trial interventions is an important finding and provides direction for future investigations into the effect of rehabilitation. The multimorbid post-ICU population may be a select population for future prospective investigations into the effect of physical rehabilitation.