Frequency Domain and Time Complex Analyses Manifest Low Correlation and Temporal Variability When Calculating Activation Rates in Atrial Fibrillation Patients

BACKGROUND: Atrial fibrillation (AF) activation rates have been calculated using both frequency domain and time complex analyses. Direct comparisons of these methods are limited. We report: (1) their correlation when measuring AF activation rates, (2) comparisons of recording durations required to minimize variability, and (3) differences in the temporal reproducibility. METHODS: AF activation rates were calculated using domain frequency (DF) (via fast Fourier transform) and time complex (TC) (via beat-to-beat activation measurements) analyses. We compared: (1) AF frequencies derived from each method; (2) successively longer subinterval durations to their 16-second reference intervals, and (3) the correlation between consecutively collected 8-second segments and segments collected 10 minutes apart. RESULTS: There was low intraclass correlation coefficient (ICC = 0.234) when comparing AF activation rates derived using DF versus TC analysis. There was no difference in the frequencies between any of the subintervals compared to their 16-second reference intervals, but variability of measurements was higher for intervals <8 seconds (P < 0.01). Correlations between successive segments and segments taken 10 minutes apart were 0.92 and 0.75 using DF analysis (P < 0.001), and 0.72 and 0.49 using TC analysis (P < 0.001). CONCLUSIONS: There is low correlation between the DF and TC methods of analyzing AF activation rates. While AF rates do not differ between subintervals and 16-second reference electrograms, the variability of measurements is dependent upon the subinterval duration, and increases for durations less than 8 seconds. AF rates were prone to change over a 10-minute time period. These results point out existing clinical limitations of measuring atrial activation rates in AF patients.