Author + information
- Satoru Kodama and
- Hirohito Sone⁎ ()
- ↵⁎University of Tsukuba, Institute of Clinical Medicine, Internal Medicine, 3-2-7 Miyamachi, Mito, Ibaraki 310-0015, Japan
We appreciate Drs. Conen and Albert pointing out data that were mistakenly extracted from their study (1) in our meta-analysis of the relationship between alcohol consumption and risk of atrial fibrillation (AF) (2). We apologize. As a result, we carefully rechecked all data from all papers included in our meta-analysis. In addition, we found that, more precisely, AF was ascertained by electrocardiogram rather than from medical records in the study by Cohen et al. (3). Also, studies initially cited as conducted in Sweden by Ruigomez et al. (4,5) were performed in the United Kingdom.
After correcting these data, we reconducted the stratified analyses of pooled relative risk of AF for highest alcohol intake versus lowest alcohol intake according to study characteristics. Nonsignificant differences in AF risk were observed by geographic region (1.38 [95% confidence interval (CI): 1.22 to 1.56] in North America and 1.51 [95% CI: 1.23 to 1.86] in Europe; p = 0.85) or by the dominant AF type (1.42 [95% CI: 1.22 to 1.66] for chronic AF and 1.78 [95% CI: 1.43 to 2.22] for paroxysmal AF; p = 0.11). When AF was ascertained by medical records, the AF risk was larger (2.05 [95% CI: 1.20 to 3.53]) than by other methods (1.48 [95% CI: 1.29 to 1.70]), but was without statistical significance (p = 0.34). In total, the influences of the corrections were very slight.
We agree that it would be desirable to investigate the dose-response relationship between alcohol consumption and AF risk with stratification by sex, considering that the level of alcohol intake at which AF risk elevates may be substantially lower among women than among men (1). However, the number of eligible studies for a dose-response analysis is too few (n = 9) for such an investigation. This point might have been addressed as a study limitation. Actually, we had performed multivariate regression analyses adjusting for sex of study participants but did not report the data. Significant relationships between alcohol intake and AF risk were found in both linear and spline dose-response curves (p < 0.001) but differences between the fits of the model were nonsignificant (R2 0.47 and 0.48, respectively; p = 0.93). The coefficient for the linear term was 8.0 ± 1.4 × 10−3, meaning that the incremental increase in relative risk of AF per 10 g alcohol consumption per day was: e10×8.0 ± 1.4×10−3 = 1.08 (95% CI: 1.05 to 1.11). Therefore, we can conclude that the relationship between daily alcohol consumption with the risk of AF is explained as linear, independent of sex.
- American College of Cardiology Foundation