Reduction In Stroke And Pump Thrombosis For Pediatric Ventricular Assist Devices: Impact Of Action (a Collaborative Learning Network)
David Sutcliffe, MD1, Angela Lorts, MD2, Jenna Murray3, Beth Hawkins, RN4, Mary Mehegan, MD5, Farhan Zafar2, Christina VanderPluym, MD4, David Rosenthal, MD3, David L. S. Morales, MD2.
1UT Southwestern, Dallas, TX, USA, 2CCHMC, Cincinnati, OH, USA, 3Lucile Packard Children's Hospital, Palo Alto, CA, USA, 4Boston Children's Hospital, Boston, MA, USA, 5St. Louis Children's Hospital, St. Louis, MO, USA.
Anticoagulation management in children on VAD is evolving rapidly, with significant variation in practice and outcomes across centers. A novel collaboration, Advanced Cardiac Therapies Improving Outcomes Network (ACTION), focused on harmonization of VAD management across centers with bivalirudin protocol being the first protocol. This study evaluates trends in pump exchanges for suspected thrombus and in stroke as it relates to the timing of ACTION initiatives.
Berlin Heart EXCOR database was reviewed for pump-exchanges secondary to suspected thrombus from years 2015-2018 for ACTION sites. Sites report all thrombus-related pump-exchanges to Berlin Heart through their ordering system. Additionally, ACTION registry was used to evaluate %-of-patients with stroke in post ACTION era (2017-onwards) compared to stroke risk from the third annual Pedimacs report.
Patients were similar in age, weight, support duration and diagnosis during all years studied. A significant reduction in pump-exchanges (per100 device days) was noted in 2017 (0.44 [0.26-0.69]) and 2018 (0.29 [0.16-0.48]) compared to prior years: 2015:(0.97 [0.65-1.42]) and 2016:(0.91 [0.56-1.39];p<0.001). There was a reduction in stroke that was also noted in the post ACTION era [5%(3/56) vs 14%(59/423); p=0.07], where 100% of ACTION sites are utilizing Bivalirudin.
There is >60% reduction in pump exchanges and >50% reduction in stroke in post ACTION era. Timing of ACTION collaboration initiatives, especially bivalirudin harmonization protocol correlates with these reductions. This highlights the capability of learning networks to improve outcomes for a unique and relatively small cohort of patients being cared for at many different sites.