Impact of Variability in Echocardiographic Interpretation on Assessment of Adequacy of Repair Following Congenital Heart Surgery: A Pilot Study.

ABSTRACT

Technical Performance Score (TPS) is based largely on the presence and magnitude of residual lesions on postoperative echocardiograms; this score correlates with outcomes following repair of congenital heart defects. We evaluated reader variability for echocardiographic components of TPS for complete repair of tetralogy of Fallot (TOF) and arterial switch operation (ASO) in two centers and measured its effect on TPS. Postoperative echocardiograms were evaluated in 67 children (39 TOF and 28 ASO). Two readers (one per center) interpreted each echocardiogram. Reader variability in image quality assessments and measurements was compared using weighted kappa (κ), percent agreement, and intra-class correlation. TPS class (1 optimal-no residua, 2 adequate-minor residua, 3 inadequate-major residua) was assigned for each echocardiographic review by an independent investigator. The effect of reader interpretation variability on TPS classification was measured. There was strong agreement for TPS between the two readers (κ = 0.88). The readers were concordant for TPS classes for 57 children (85%) and discordant for classes 2 (minor residua) versus 3 (major residua) in six (9%). Coronary arteries and branch pulmonary arteries were frequently suboptimally visualized. Although inter-reader agreement for TPS was strong, inter-reader variation in echocardiographic interpretations had a small, but important effect on TPS for TOF and ASO, particularly for the distinction between minor and major residua. Further studies of generalizability and reproducibility of TPS and refinement of scoring modules may be needed before it can be used as a tool to assess pediatric cardiac surgical performance and outcomes.