Edward J. Hickey1, Christopher A. Caldarone1, Eugene H. Blackstone2, Gary K. Lofland3, Thomas Yeh4, Christo I. Tchervenkov5, Christian Pizarro6, Frank A. Pigula7, Brian W. McCrindle1; 1The Hospital for Sick Children, Toronto, ON, Canada; 2The Cleveland Clinic Foundation, Cleveland, OH; 3Children's Mercy Hospital, Kansas City, MO; 4University of Texas, Southwestern Medical Center, Dallas, TX; 5Montreal Children's Hospital, Montreal, QC, Canada; 6Alfred DuPont Hospital for Children, Wilmington, DE; 7Children's Hospital Boston, Boston, MA
Objective: Newborns with critical aortic stenosis (AVS) are frequently committed to uni- (UVR) or biventricular (BVR) repair pathways within the first days of life. We aimed to determine the impact of patient characteristics and initial treatment decisions on survival.Methods: Neonates with critical AVS (n=362) were prospectively enrolled with the CHSS from 1994-2001 and had an initial procedure indicating either an intended UVR (n=223, 84 deaths) or BVR (n=139, 39 deaths). Parametric models of UVR- and BVR- survival were used to identify risk factors for time-related death. Decision making was then scrutinized according to the predicted 5-year survival advantage for UVR.
Results: Incremental factors for increased risk of time-related death for UVR were: presence of moderate/severe tricuspid regurgitation (p<.01), smaller mitral annulus z-score (p<.001), smaller indexed length of the dominant ventricle (p=.02) and presence of a large VSD (p=.01); and for BVR: indexed minimum LV outflow tract diameter (p=.001), higher grade of endocardial fibroelastosis (EFE) (p<.01), presence of LV dysfunction (p=.02), and smaller indexed diameter of the mid-aortic arch (p=.05). The above variables all formed a model predicting the 5-year survival advantage for UVR (all p<.0001, R2=.92). When the management path chosen was that which was subsequently associated with optimal predicted survival, the decision was coded as "correct". Actual and predicted deaths closely matched when the chosen pathway had been "correct." "Incorrect" chosen pathways were more common with BVR than UVR (56% v 21%, p<.01). "Incorrect" BVRs had disproportionately more actual vs predicted deaths (BVR 30 v 14, p<.001; UVR 20 v 13, p=.02). Although BVRs were associated with larger dimensions of left-sided structures than UVR patients, presence of LV dysfunction, higher grade of EFE and the absence of a large VSD were features that specifically identified the "incorrect" BVRs.
Conclusion: Pursuing BVR when predicted survival favored UVR was the more frequent pitfall than the converse for this cohort, perhaps due to larger yet functionally inadequate left-sided structures. The potential gain in survivorship was greatest for this "incorrect" BVR group. "Correct" management predicted 26 fewer deaths within this CHSS cohort. Whether advantages in survivorship reflect advantages in functional outcomes is not known.
Back to Annual Meeting
Back to Program Outline
