A dynamic model to estimate evolving risk of major bleeding after percutaneous coronary intervention

BackgroundBleeding is a complication of percutaneous coronary intervention (PCI), leading to significant morbidity, mortality, and cost. Existing risk models produce a single estimate of bleeding risk anchored at a single point in time and do not update estimates as clinical information emerges, despite the dynamic nature of risk. ObjectiveWe sought to develop models that update estimates of bleeding risk over time, incorporating evolving clinical information, and to demonstrate updated predictive performance. MethodsUsing data available from the National Cardiovascular Data Registry (NCDR) CathPCI, we trained 6 different tree-based machine learning models to estimate the risk of bleeding at key decision points: 1) choice of access site, 2) prescription of medication prior to PCI, and 3) the choice of closure device. ResultsWe included 2,868,808 PCIs; 2,314,446 (80.7%) prior to 2014 for training and 554,362 (19.3%) remaining for validation. Discrimination improved from an AUROC of 0.812 (95% Confidence Interval: 0.812-0.812) using only presentation variables to 0.845 (0.845-0.845) using all variables. Among 123,712 patients classified as low risk by the initial model, 14,441 were reclassified as moderate risk (1.4% experienced bleeds), while 723 were reclassified as high risk (12.5% experienced bleeds). Among 160,165 patients classified as high risk by the initial model, 40 were reclassified to low risk (0% experienced bleeds), and 43,265 to moderate risk (2.5% experienced bleeds). ConclusionAccounting for the time-varying nature of data and capturing the association between treatment decisions and changes in risk provide up-to-date information that may guide individualized care throughout a hospitalization. Condensed AbstractExisting risk models for bleeding with PCI produce a single estimate anchored at a single point in time. We developed models that update estimates of bleeding risk over time, incorporating evolving clinical information, using data available from the National Cardiovascular Data Registry (NCDR) CathPCI. We trained 6 different machine learning models to estimate the risk of bleeding at key decision points, improving discrimination from an AUROC of 0.812 to 0.845, over time. Accounting for the time-varying nature of data and capturing association between treatments and changes in risk provide up-to-date information that may guide individualized care throughout a hospitalization.