The data were obtained from the National (Nationwide) Inpatient Sample, part of the Healthcare Cost and Utilization Project (HCUP), sponsored by the Agency for Healthcare Research and Quality. This is the largest publicly available all-payer inpatient care database in the United States, containing data from more than seven million hospital stays each year (7). The National Inpatient Sample database randomly samples 20% of the discharges from participating hospitals in 47 US states and the District of Columbia. The sampling method provides a geographically distributed sample that represents all inpatient admissions in the nation. The use of data from approved public datasets is not considered human subject research; the study was granted exempt status from the Cleveland Clinic Institutional Review Board.
The study population was identified using the International Classification of Diseases, Ninth and Tenth Revisions, Clinical Modification (ICD-9&10-CM). Data were queried from the years 2000–2017. Pediatric patients (≤ 18 years of age) who had cardiac arrest during their in-hospital stay between January 1, 2000, and December 31, 2017, were included. Cardiac arrest was defined as a primary or secondary diagnosis of cardiac arrest (ICD codes 427.5, I46.2, I46.8, I46.9) or ventricular fibrillation (ICD codes 427.41, I49.01). To distinguish between in-hospital and out-of-hospital CA, we excluded patients with no cardiopulmonary resuscitation (CPR) procedure during the hospitalization. Patients were divided into two groups, survivors and non-survivors. We collected patient demographics, hospital-level data (location of the hospital, type of hospital, bed size of the hospital, the region of the hospital), type of admission (elective or non-elective), and day of admission (weekday versus weekend).
Outcomes measured were in-hospital mortality, length of hospital stay (LOS), and cost of hospitalization. Medical cardiac conditions compiled into one category included myocarditis, pulmonary heart disease, heart failure, acute myocardial infarction, hypertrophic cardiomyopathy, cardiogenic shock, cardiac tamponade, Marfan syndrome, and coronary artery anomalies. The arrhythmias group included heart block, sinus node dysfunction, long QT dysfunction, Wolff-Parkinson-White syndrome, and conduction disorders. To identify patients with congenital heart surgery, we used ICD-9 and ICD-10 procedural codes for congenital heart surgery operations, patients were included in the congenital heart disease (CHD) group if they received the procedure during the hospitalization. We also examined survival of inpatient cardiac arrest in the CHD group over the years of the study.
We also included secondary analysis of the use of ECMO at the time of CPR. For this analysis, patients were divided into two groups, CPR with ECMO (ECPR) and CPR without ECMO. For the multivariable regression analysis, we first compiled a list of clinically important potential predictors of in-hospital survival after cardiac arrest, which included both patient and hospital characteristics as demographics, clinical diagnoses, ECMO utilization and hospital teaching status. A univariate analysis was initially performed, after that, variables with significance (p < 0.2) were incorporated in a multivariable analysis using a logistic regression model. Because of the limited information we can gather from this database, this study did not have details about the location and duration of cardiopulmonary resuscitation provided, medications administered during CPR and hemodynamic data.
Continuous variables were described using median and interquartile range (IQR). Categorical variables were described using frequencies and percentages. Demographics, clinical characteristics, and outcomes were compared using Mann-Whitney U test for continuous variables and Chi-square or Fisher’s exact tests for categorical variables. The non-parametric Jonckheere-Terpstra test was used to study the trends of in-hospital cardiac arrest survival, ECPR, and survival in congenital heart disease surgical patients with in-hospital cardiac arrest. Statistical significance was set at p < 0.05. The analysis was performed by SPSS software, version 25 (SPSS Inc., Chicago, IL) was used for statistical analysis.