Statistical Analysis
Statistical analysis was performed using SPSS v 25 (IBM corporation, Armonk, NY). All missing data was excluded and outliers were included. Categorical variables were compared using Chi-squared test or Fisher exact test while continuous variables were compared using student’s t-test or Mann-Whitney test for normally or non-normally distributions respectively. Univariate regression was used to compare the primary endpoint with baseline variables. Multivariate analysis was performed using multi nominal regression analysis, with step-wise inclusion into the model. Two tailed P values < 0.05 were considered statistically significant. An exploratory analysis of the best discriminators of AF-free HF events was performed using a receiver-operator curve.
To determine sample size for the prospective cohort, we assumed a doubling of the primary endpoint compared to the retrospective cohort given the prospective ascertainment. Additionally, based on previous data that acute hemodynamic changes affect clinical HF, we assumed a left atrial pressure difference of 5 mmHg in those with or without early incident HF (14+/- 6 mmHg v. 9 mmHg), and with alpha = 0.05 and beta = 0.80, a total of 104 patients was recommended (https://clincalc.com/stats/samplesize.aspx). (21) Because of the greater than expected rate of HF, the study was stopped at 100 subjects.
RESULTS
For the retrospective analysis, there were 391 patients who were included, with a primary endpoint reached in 13 patients (3.3%). When the group of early incident HF patients was compared to the control group, univariate correlates to the primary endpoint included longer procedure time (5:47 v. 4:41 hr, p = 0.002), greater fluid administration (2406 v 1887 mL, p = 0.015), and preexisting diagnosis of HF (69% v 39%, p = 0.03). Persistent AF, redo procedure, gender, age, body mass index, chronic kidney disease, serum creatinine, diabetes mellitus, and coronary disease were not different in the two groups. Significantly associated variables were included in a multivariable model, but no factors remained independently associated with the primary outcome.