Updating An Empirically Based Tool For Analyzing Congenital Heart Surgery Mortality
Marshall L. Jacobs, MD1, Jeffrey P. Jacobs, MD2, Dylan Thibault, MS3, Kevin D. Hill, MD, MS4, Brett R. Anderson, MD, MBA, MS5, Pirooz Eghtesady, MD, PhD6, Tara Karamlou, MD7, S Ram Kumar, MD, PhD8, John E. Mayer, MD9, Carlos Mery, MD10, Meena Nathan, MD9, David M. Overman, MD11, Sara K. Pasquali, MD, MHS12, James D. St Louis, MD13, David Shahian, MD9, Sean O'Brien, PhD14.
1Johns Hopkins School of Medicine, Baltimore, MD, USA, 2Johns Hopkins School of Medicine, St Petersburg, FL, USA, 3Dule Clinical Research Institute, Durham, NC, USA, 4Duke University, Durham, NC, USA, 5Columbia University, New York, NY, USA, 6Washington UNiversity, Saint Louis, MO, USA, 7Cleveland Clinic, Cleveland, OH, USA, 8University of Southern California, Los Angeles, CA, USA, 9Harvard Medical School, Boston, MA, USA, 10University of Texas at Austin, Austin, TX, USA, 11Childrens Hospitals and Clinics of Minnesota, Minneapolis, MN, USA, 12University of Michigan, Ann Arbor, MI, USA, 13University of Missouri-Kansas City School of Medicine, Kansas City, MO, USA, 14Duke Clinical Research Institute, Durham, NC, USA.
Objectives: STAT Mortality Categories (developed 2009) stratify congenital heart surgery procedures into groups of increasing mortality risk to account for case mix in outcomes analysis. The updated metric should be empirically based for all procedures and should reflect contemporary outcomes. Methods: STS-CHSD operations (2010-2017) including one or more cardiovascular surgical procedures were analyzed. In the new model, multiple component operations (MCO) with N>5 were eligible for unique STAT Scores when the statistically estimated mortality risk differed from that of the highest risk component procedure based upon: mortality difference ratio ≥1.25 relative to highest risk component with probability ≥90%. For eligible procedures (including MCO), mortality risk was estimated from all operations which included the procedure. Bayesian modeling adjusted for small denominators. Risk estimates were re-scaled to STAT Scores between 0.1 and 5.0. STAT Category assignment was designed to minimize within-category variation and maximize between-category variation Results: Among 161,351 operations (19,090 distinct procedure combinations), 234 single or MCO received unique STAT Scores. The figure shows model-based estimates of procedural mortality risk, with new STAT Category stratification. With the update, numerous procedures changed STAT Category relative to 2009 metrics.Conclusions: The number of procedure codes with empirically derived Scores has grown by almost 60%, reflecting contemporary practice and accounting for risk associated with MCOs. Updating STAT metrics based on contemporary outcomes facilitates accurate assessment of case mix.
Back to 2020 Abstracts