Evaluation of an adaptive, rule-based dosing algorithm to maintain therapeutic anticoagulation during atrial fibrillation ablation

Figures

• 

  Figure 1. Description and development of the heparin dosing algorithm. A: The Digital Intern (Integrated Vital Medical Dynamics, Madison, WI) procedure–based heparin dosing calculator, ie, “dosing algorithm,” is an adaptive, rule-based system that mathematically generates recommendations based on the distance from current activated clotting time (ACT) to the central target. To initiate the algorithm, initial bolus and infusion recommendations are required. Bolus1 is determined based on patient weight, type of anticoagulant, and distance from goal ACT to ACT0. The individual patient response to Bolus1 scales the coefficients that generate all future bolus and infusion recommendations. From ACT2 onward, the previous bolus, change in infusion rate, (Goal ACT - ACTn), and (ACTn−ACTn−1) continue to refine the coefficients until steady state is achieved. Text in black boxes represent initial user inputs, those in circles are subsequent user inputs, and text in rhombi represent actions of the dosing algorithm. B: The algorithm was developed in 3 phases. Baseline performance was established using recommendations from consensus documents and institutional experience to initiate the algorithm (blue…

• 

  Figure 2. Heparin dosing algorithm initiation and use. A: The heparin dosing algorithm was opened by staff directly from the electronic medical record. B. Staff then input required data. Patient characteristics input by staff and used in the algorithm calculations included medication used for thromboembolic prophylaxis, weight, and baseline ACT (highlighted in figure). Additional characteristics input included time since last dose of anticoagulant, sex, creatinine, AST, ALT, and total bilirubin. An initial bolus dose was then suggested. Staff entered subsequent activated clotting time results and received recommendations on bolus and infusion rates. Actual doses delivered were tabulated.

• 

  Figure 3. Intraprocedural anticoagulation outcomes improved with the dosing algorithm. A: The time to reach goal activated clotting time (ACT) dropped from 33.3 ± 23.6 minutes to 17.6 ± 11.1 minutes, beta coefficient for linear regression -15.8, confidence interval (CI): -20.8 to -10.8, P < .001. In this box plot, the box covers the interquartile range while the whiskers are restricted to 1.5 times the interquartile range; outliers are defined as greater than 1.5 times the interquartile range. B: The percentage of patients experiencing ACT <300 seconds while operating in the left atrium dropped from 54% (27/50 historical controls) to 19% (26/139 dosing algorithm), odds ratio (OR) 0.20, CI: 0.10–0.39, P < .001. C: The percentage of patients with any ACT >400 seconds decreased from 20% (10/50 historical controls) to 5% (7/139 dosing algorithm), OR 0.21, CI: 0.07–0.59, P = .003.