Combining Data from Subjects' Histories for the Prediction of Imminent Atrial Fibrillation is Useful for the Elderly Male Population

Abstract

OBJECTIVE: (1) To estimate the prevalence of atrial fibrillation (Af) in the general population of Saitama Prefecture, (2) to identify useful information obtained from subjects' history and laboratory findings for predicting imminent occurrence of Af.
DESIGN: Cross sectional study and retrospective cohort study.
SETTING: Annual health examination of the general population.
PARTICIPANTS: Subjects were 5, 375 men and 8, 419 women with a mean age±standard deviation of 47.2±9.6 and an age range of 18 to 88 years old.
MEASUREMENTS AND MAIN RESULTS: Among the 13, 794 participants who underwent conventional 12-lead electrocardiography (ECG) every year from 1994 to 1997, the prevalence of Af was 0.3%. The cross sectional data showed significant differences between male subjects with and without Af regarding age (P<0.0001), current treatment for hypertension (P<0.0001), arrhythmia (P<0.0001), angina pectoris (P<0.05), history of myocardial infarction (P<0.05), perception of palpitation (P<0.0001), perception of irregular pulse (P<0.0001), shortness of breath (P<0.0001), diastolic blood pressure (P<0.05), total cholesterol (P<0.0001), and casual glucose level (P<0.05) . Significant differences were also found between female subjects with and without Af with respect to creatinine (P<0.0001), current treatment of arrhythmia (P<0.0001), perception of palpitation (P<0.0001), perception of irregular pulse (P<0.0001), and HbAlc level (P<0.05) .
The retrospective cohort data identified 22 subjects who had persistent Af during the study period, and 20 (0.15%) who had newly developed Af during this period on the basis of consecutive ECG recordings of the 13, 772 participants in 1994-1996. Logistic regression analysis revealed that there were significant differences in perception of irregular pulse (P=0.0004), history of myocardial infarction (P=0.0134), fatigability (P=0.0243), aging (P=0.0039) and total cholesterol level (P=0.025) for men, and in history of arrhythmia (P=0.0007) for women between the group with and without Af.
The likelihood ratios and the respective 95% confidence intervals (C.I.) for ECG findings to identify the subjects who would develop Af were as follows: ST depression, 14.6 (C.I., 5.1-42) ; ventricular arrhythmia, 14.9 (3.9-56) ; incomplete RBBB, 9.2 (2.4-34) ; supraventricular arrhythmia, 8.9 (1.3-61) ; second-degree atrioventricular block, 342 (32-3624) ; abnormal Q wave, 16.3 (2.4-112) ; left atrial enlargement, 52.7 (7.2-383) .
CONCLUSIONS: Combining data from subjects' histories for the prediction of imminent Af among subjects in the general population is likely to be useful, especially for the elderly male population. Aging in men, in particular, is an important factor when combined with abnormal findings on conventional 12-lead ECG.