This study aimed to examine the association of worsening frailty patterns at hospital discharge relative to premorbid status with long-term neurological recovery in a large cohort of cardiac arrest (CA) survivors.  

Frailty, a marker of impaired resiliency to stressors, is increasingly recognized as a useful prognosticator for mortality and disability in critical illness survivors.

This prospectively collected data on 246 consecutive survivors to hospital discharge after CA included patients resuscitated after in-or-out-of-hospital CA between 3/1/2015 and 2/28/2018 in a single tertiary care center. The 9-point Clinical Frailty Scale was used to retrospectively estimate frailty before admission and at hospital discharge. Poor neurological recovery at 1-year was defined as Cerebral Performance Category score (CPC) >2. We used logistic regression to estimate the association of change in frailty scores from admission to hospital discharge with poor neurological recovery at 1-year.

Of 724 admitted CA patients, 246 (34%) survived to hospital discharge and 237 (mean age 57±16 years) with 1-year CPC were analyzed. Median frailty scores changed from 4 to 6, i.e. from mostly independent premorbid to mostly dependent by hospital discharge. Poor outcomes were seen in 110 (46%) patients. Patients with worsening frailty patterns were more likely to have poor 1-year neurological recovery (Odds Ratios (OR):1.9 (95% confidence interval (CI) 1.5-2.3) after adjusting for premorbid frailty status. The associations remained significant (OR 1.5; CI 1.1-2) after adjusting for age, sex, race, insurance status, discharge CPC>2, Charlson index, initial cardiac rhythm, use of bystander CPR, defibrillation, therapeutic hypothermia, hospital length of stay, and discharge disposition.

Increase in clinical frailty scores from pre-hospitalization to hospital discharge after CA is independently associated with worse 1-year neurological recovery. Future studies confirming these findings utilizing inflammatory and hormonal biomarkers along with physiological measures such as gait speed, handgrip, and physical activity to describe various phenotypes are warranted.