Individualized Treatment Effects of Therapeutic Hypothermia in Children Postcardiac Arrest: A Reanalysis of Two Randomized Clinical Trials.

OBJECTIVES: To use patient characteristics to estimate individualized treatment effects (ITE) of hypothermia vs. normothermia after pediatric cardiac arrest. DESIGN: Secondary, exploratory analysis of two pediatric randomized controlled trials (RCTs), Therapeutic Hypothermia after Pediatric Cardiac Arrest Out-of-Hospital (THAPCA-OH; NCT00878644) and Therapeutic Hypothermia after Pediatric Cardiac Arrest Out-of-Hospital In-Hospital (THAPCA-IH; NCT00880087), using a causal forest machine learning model to estimate ITEs within each trial. SETTING: THAPCA-OH was conducted at 38 children's hospitals across the United States and Canada. THAPCA-IH was conducted at 37 children's hospitals across the United States, Canada, and the United Kingdom. PATIENTS: Pediatric patients aged 48 hours to 18 years who remained comatose within 6 hours after return of circulation following cardiac arrest and were randomized in THAPCA-OH and THAPCA-IH to normothermia or therapeutic hypothermia for 48 hours. Patients with a baseline Vineland Adaptive Behavior Scales, Second Edition (VABS-II) score less than 70 were excluded. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: The final cohorts included 260 patients in THAPCA-OH and 257 in THAPCA-IH. The primary outcome was survival at 1 year with VABS-II greater than or equal to 70 (favorable outcome). In THAPCA-OH, estimated ITEs (calculated as the individualized absolute risk difference [iARD] between hypothermia and normothermia, positive favoring hypothermia) ranged from -0.01 to 0.16. Patients were grouped into tertiles of estimated ITE within each trial. In THAPCA-OH, the tertile with the greatest estimated benefit from hypothermia had an observed absolute risk difference (ARD; hypothermia minus normothermia) of 0.18 (95% CI, 0.02-0.34). In THAPCA-IH, estimated ITEs ranged from -0.17 to 0.13. The tertile estimated to benefit most from hypothermia had an ARD of 0.27 (95% CI, 0.07-0.48), whereas the tertile estimated to benefit from normothermia had an ARD of -0.20 (95% CI, -0.40 to -0.01). CONCLUSIONS: These analyses suggest heterogeneity of treatment effect may exist in postcardiac arrest temperature management warranting further study.