Characterizing time to conversion to sinus rhythm under digoxin and placebo in acute atrial fibrillation
Stefanie Hennig, Lena E. Friberg, Mats O. Karlsson
Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: The underlying trial was a randomized, double-blind comparison of intravenously administered digoxin and placebo in patients with acute atrial fibrillation. The primary end point was conversion to sinus rhythm within 16 h after randomization, while effect on heart rate (HR) was a secondary end point. Sparse plasma samples and 5 HR measurements were taken. The primary publication  concluded: "There was no clear correlation between serum concentration of digoxin and changes in ECG or HR". A later analysis  reported on the population PK of digoxin and a HR model. The aim now was to evaluate if any predictor of conversion to sinus rhythm can be found using a Time-to-event (TTE) analysis.
Methods: Data from the studies were analyzed using the Laplace method in NONMEM VI. The F_Flag functionality was used to simultaneously predict the continuous data (PK and HR observations) and estimate probabilities for the categorical data. Since the PK and HR analyses were performed previously, PPP&D was used for the combined PKPD and TTE analysis. The final model was also rerun fitting the HR and the conversion data simultaneously. The assessment of statistical significance of additional parameters was based on the difference between the OFV at 5% significance. The predicted performance of the TTE analysis was assessed using Kaplan-Meier VPC graphics.
Results: The main influence on having conversion was ‘time' followed by ‘age' and ‘sex'. Time was included as exponential, step or spline functions. The final model included time as a step function with an estimated break point at 4.2h. The chance of having a conversion declined exponentially with age. The hazard of having conversion was estimated to be 5.2%.h-1 for the typical male (62 year) and 8.7%.h-1 for the typical female (71 year). After 4.2 h the chance of having a conversion to sinus rhythm was reduced by 52% for males and 60% for females.
Even though HR decreases with time, including HR or the change of HR from baseline had no explanatory value in the presence of a time effect. Similarly, models including digoxin concentrations were not significant. The simultaneous fit of HR and TTE data provided very similar estimates to the PPP&D method.
Conclusions: The chance of having a conversion was higher for the first 4.2h. Overall, the model estimated a 30-40% higher chance for females to convert compared to males. No difference could be found between the digoxin and the placebo group.
 Hornestam B, Held P, Edvardsson N. Clinical cardiology 1999; 22: 96-102..
 Hornestam B, Jerling M, Karlsson MO, Held P. Eur J Clin Pharmacol 2003; 58: 747-55.
 Wade JR, Karlsson MO. PAGE 1999, Saintes, France: Abstr 139 [http://www.page-meeting.org/?abstract=39%5d].