Mortality and Cause of Death Following Pediatric Cardiac Surgery for Congenital Heart Defects

Abstract

Congenital heart defects (CHD) affect almost 1% of births. The primary method for managing these defects is surgery. These analyses used data from the Pediatric Cardiac Care Consortium (PCCC), a large, US-based registry of pediatric interventions for CHDs. The PCCC was previously linked to the National Death Index and was linked to the American Medical Association Physician Masterfile as part of this dissertation. The first two analyses examined the associations between surgeon and center characteristics and post-surgical mortality using multilevel modeling. These analyses examined procedure-specific volume at the surgeon and center levels as well as training center status at the center level and years since graduation from medical school at the surgeon level. In the third analysis, multiple cause of death data were examined to determine the burden of contributing causes of death. Standardized mortality ratios and competing risk Cox regression compared these results with those calculated using underlying cause of death. The first analysis found that after adjusting for known patient-level risk factors, center factors including procedure-specific volume were not associated with early post-discharge mortality. The second analysis of short and medium term mortality demonstrated a consistent center-level association between procedure-specific volume and mortality among several complex repairs. No association was observed among patients with relatively simple ventricular septal defect repairs. Finally, we found that standardized mortality ratios based on underlying cause of death underestimated the burden of death associated with injury as well as perinatal, infectious, endocrine, genitourinary, and circulatory diseases. These differences varied by age and defect severity. Perinatal and endocrine disease were highest among those with severe defects. Differences in mortality due to infection showed a bimodal association with age at the time of death. The combination of multi-level modeling and multiple cause of death methods leveraged in these analyses advances the understanding of the roles of healthcare systems and multiple causes of death.