Abstract 20931: Therapeutic Hypothermia in In Hospital Cardiac Arrest

    loading  Checking for direct PDF access through Ovid


Introduction: Role of therapeutic hypothermia in out of hospital cardiac arrest is well studied but data is sparse for in-hospital cardiac arrest (IHCA).

Methods: Nationwide Inpatient Sample (NIS) database from 2011-2014 was utilized to identify ICHA using the ICD-9-CM procedural codes 99.60(CPR) and 99.63(closed cardiac massage) in any procedural fields. Therapeutic hypothermia was identified by 99.81 in any procedural field. Patient with age<18 years and missing information on age, gender and mortality was removed. Primary outcome was hospitalization survival. We used Greedy’s algorithm for propensity score match (1:20) to adjust for extensive list of clinically important confounders. We used simple logistic regression to generate odds ratio on matched cohort.

Results: 86,593 patients (Age>65 years: 58.48%, female: 44.12%) with In hospital cardiac arrest was identified,of which 1777 (2.05 %) received therapeutic hypothermia. Overall survival was 30%- with or without hypothermia. After propensity score matching (1:20) we identified 1537 matched pairs (32,277 patients) without any statically significant difference in baseline characteristics between two groups. C- index for the model was 0.714. Worse survival was observed in patients with hypothermia in matched cohort [Survival: (31.82% without hypothermia vs. 28.1% with hypothermia with p-value- 0.002) (OR:0.84, 95% CI:0.75-0.94, p-0.002)]. About a quarter of in-hospital cardiac arrest were Vfib/Vtach arrest and Subgroup propensity score match analysis in Vfib/Vtach arrest showed better survival without hypothermia, similar results were observed in PEA/asystole cardiac arrest.

Conclusion: Worst survival was observed with hypothermia in hospital cardiac arrest patients. Risk of hypothermia complications outweighs the benefit of preserving brain in patients who received almost immediate resuscitation. Future clinical trials and prospective studies will be needed to concur our results.

Related Topics

    loading  Loading Related Articles