Efficacy outcome selection in the therapeutic hypothermia after pediatric cardiac arrest trials.

OBJECTIVES: The Therapeutic Hypothermia After Pediatric Cardiac Arrest trials will determine whether therapeutic hypothermia improves survival with good neurobehavioral outcome, as assessed by the Vineland Adaptive Behavior Scales Second Edition, in children resuscitated after cardiac arrest in the in-hospital and out-of-hospital settings. We describe the innovative efficacy outcome selection process during Therapeutic Hypothermia After Pediatric Cardiac Arrest protocol development. DESIGN/
SETTING: Consensus assessment of potential outcomes and evaluation timepoints.
INTERVENTIONS: None.
MEASUREMENTS AND MAIN RESULTS: We evaluated practical and technical advantages of several follow-up timepoints and continuous/categorical outcome variants. Simulations estimated power assuming varying hypothermia benefit on mortality and on neurobehavioral function among survivors. Twelve months after arrest was selected as the optimal assessment timepoint for pragmatic and clinical reasons. Change in Vineland Adaptive Behavior Scales Second Edition from prearrest level, measured as quasicontinuous with death and vegetative status being worst-possible levels, yielded optimal statistical power. However, clinicians preferred simpler multicategorical or binary outcomes because of easier interpretability and favored outcomes based solely on postarrest status because of concerns about accurate parental assessment of prearrest status and differing clinical impact of a given Vineland Adaptive Behavior Scales Second Edition change depending on prearrest status. Simulations found only modest power loss from categorizing or dichotomizing quasicontinuous outcomes because of high expected mortality. The primary outcome selected was survival with 12-month Vineland Adaptive Behavior Scales Second Edition no less than two SD below a reference population mean (70 points), necessarily evaluated only among children with prearrest Vineland Adaptive Behavior Scales Second Edition greater than or equal to 70. Two secondary efficacy outcomes, 12-month survival and quasicontinuous Vineland Adaptive Behavior Scales Second Edition change from prearrest level, will be evaluated among all randomized children, including those with compromised function prearrest.
CONCLUSIONS: Extensive discussion of optimal efficacy assessment timing, and of the advantages versus drawbacks of incorporating prearrest status and using quasicontinuous versus simpler outcomes, was highly beneficial to the final Therapeutic Hypothermia After Pediatric Cardiac Arrest design. A relatively simple, binary primary outcome evaluated at 12 months was selected, with two secondary outcomes that address the potential disadvantages of primary outcome.