Surgical treatment of congenital heart disease represents a major cause of pediatric hospitalization and healthcare resource use. Larger centers may provide more efficient care with resulting shorter length of postoperative hospitalization (LOH). Data from 46 centers over 25 years were used to evaluate whether surgical volume was an important determinant of LOH using a competing risk regression strategy that concurrently accounted for deaths, transfers, and discharges with some time interactions. Earlier discharge was more likely for infants and older children compared to neonates [subhazard ratios at postoperative day 6 of 1.64 (99 % confidence interval (CI) 1.57, 1.72) and 2.67 (99 % CI 2.53, 2.80), respectively], but less likely for patients undergoing operations in Risk Adjustment for Congenital Heart Surgery categories 2, 3, 4, and 5/6 compared to category 1 [subhazard ratios at postoperative day 6 of 0.66 (99 % CI 0.64, 0.68), 0.34 (95 % CI 0.33, 0.35), 0.28 (99 % CI 0.27, 0.30), and 0.10 (99 % CI 0.09, 0.11), respectively]. There was no difference by sex [non-time-dependent subhazard ratio 1.019 (99 % CI 0.995, 1.040)]. For every 100-operation increase in center annual surgical volume, the non-time-dependent subhazard for discharge was 1.035 (99 % CI 1.006, 1.064) times greater, and center-specific exponentiated random effects ranged from 0.70 to 1.42 with a variance of 0.023. The conditional discharge rate increased with increasing age and later era. No sex-specific difference was found. Centers performing more operations discharged patients sooner than lower volume centers, but this difference appears to be too small to be of clinical significance. Interestingly, unmeasured institutional characteristics estimated by the center random effects were variable, suggesting that these played an important role in LOH and merit further investigation.
- Competing risk regression
- Congenital heart disease
- Length of hospitalization
- Pediatric Cardiac Care Consortium