Limitations:
Administrative databases provide the benefit of including large groups
of patients across the nation and can be helpful in obtaining large
samples sizes for rare diseases or rare outcomes. However, they lack the
clinical details and are susceptible to coding errors. This study is
based on assumptions formed by diagnostic codes and hence the
conclusions may be prone to errors. This study also represents the
hospitalizations and not individual patients. However, in-hospital
deaths would represent individual patients. We cannot determine the
events that lead to the in-hospital death or the actual cause of death.
We are also not able to discern the type of ICD (single chamber vs dual
chamber) and whether the ICD was utilized, i.e. delivered an appropriate
shock, during the admission. It is also difficult to determine the
specific forms of congenital heart disease (for example patients with
single ventricular physiology vs biventricular repairs), precluding us
from stratifying patients into risk categories. Lastly, since only
hospital data is available, outcomes such as death after discharge or
readmission rates cannot be determined.