Experimental and clinical trial data have suggested an association between fish oil intake and atrial fibrillation (AF). However, previous observational studies have reported conflicting results regarding this association. Thus, we sought to compare the association between dietary fish intake and incident AF in a large sample of older, postmenopausal women. We included 44,720 participants from the Women’s Health Initiative clinical trials who were not enrolled in the dietary modification intervention arm and without AF at baseline. The dietary intake of nonfried fish and omega-3 fatty acid intake was estimated from a Food Frequency Questionnaire at study entry. Incident AF was determined by follow-up electrocardiography at years 3 and 6. The baseline characteristics and rates of incident AF were compared across the quartiles of fish intake. Adjusted logistic regression models were used to evaluate the association between dietary nonfried fish intake and incident AF. A total of 378 incident cases of AF occurred during the follow-up period. In the age-adjusted models, no association was found between dietary nonfried fish intake and incident AF (odds ratio 1.17, 95% confidence interval 0.88 to 1.57 for quartile 4 vs quartile 1 of dietary fish intake). Similar findings were observed in the multivariate models and in the subgroup analyses. In conclusion, in a large cohort of healthy women, we found no evidence of an association between fish or omega-3 fatty acid intake and incident AF.
Atrial fibrillation (AF) represents the most common sustained dysrhythmia in humans and is particularly prevalent in older ages, with a population prevalence of 10% in the ninth decade and a lifetime risk of 25%. Because AF is associated with greater rates of mortality, heart failure, and stroke, the high prevalence of AF in the population represents a significant public health problem. Previous observational studies have shown several pharmacologic therapies and lifestyle patterns to be associated with AF in the general population. For example, observational, clinical trial, and experimental data have suggested that omega-3 fatty acids from dietary fish and/or supplements might decrease the risk of AF. However, 2 additional observational studies have reported no apparent association between dietary fish intake and incident AF. Because of these conflicting findings and the ongoing uncertainty regarding the role of dietary fish intake and incident AF, we sought to compare the association between dietary fish intake and incident AF in a large sample of older, postmenopausal women from the Women’s Health Initiative.
Methods
The details of the eligibility criteria and recruitment methods, randomization, follow-up, data and safety monitoring, and quality assurance for the Women’s Health Initiative have been previously published. In brief, 68,131 postmenopausal women aged 50 to 79 years were recruited by population-based mailing campaigns from 1993 to 1998 to enroll in one or more clinical trials: dietary modification, calcium/vitamin D, and/or hormone replacement therapy. We excluded women in the dietary modification intervention arm (n = 19,541), because the low-fat dietary change program was designed to alter women’s usual dietary habits. We also sequentially excluded women with AF according to the electrocardiographic (ECG) findings at baseline (n = 135) or self-report (n = 1,752) and women providing incomplete data regarding dietary fish intake (n = 1,984). Therefore, we included 44,720 participants for the present analysis. The baseline characteristics were measured at study entry in accordance with standard protocols as reported previously. Participants were queried about a history of AF or other cardiovascular disease diagnoses using a standard questionnaire. In addition, participants were asked to report their ethnicity using a self-report questionnaire. The institutional review boards of all participating institutions approved the protocol and consent form, and all participants provided written informed consent.
The dietary intake of fish was measured at study entry using the Food Frequency Questionnaire (FFQ). In brief, the participants are asked about the frequency, amount, and preparation of 122 distinct foods or food groups. The frequencies of food portion intake ranged from “never” or “less than once per month” to “2+ servings per day” during the past 3 months. The portion sizes were small, medium, and large in reference to a stated medium portion size (3 oz of fish or 1/2 cup of tuna). The “small” and “large” portion sizes were described as “1/2 the medium serving size” and “1 and 1/2 times the medium serving size or more,” respectively. The measurement characteristics of the Women’s Health Initiative FFQ were evaluated by comparison with a 24-hour dietary recall and 4-day food record.
Nonfried fish intake was categorized as the frequency of eating 3 types of fish taken from the FFQ: (1) canned tuna, tuna salad, and tuna casserole; (2) white fish (broiled or baked), such as sole, snapper, or cod; and (3) dark fish (broiled or baked) such as salmon, mackerel, or bluefish. The quartiles of medium portion sizes per week were created according to the portion size and frequency of the meals. Nonfried fish intake was used as the primary exposure variable because of previous reports demonstrating the importance of food preparation on AF and other cardiovascular end points. In subsequent analyses, fish intake was restricted to only tuna and dark fish to create quartiles of tuna/dark fish intake. In addition, secondary analyses were also performed for fried fish. The estimates of dietary omega-3 fatty acid (grams per day) were created using analysis software that incorporates 19 adjustment questions to the FFQ to estimate the specific nutrient intake from food, as described previously. No data were available on the use of oral omega-3 fatty acid supplements.
Standard 12-lead electrocardiograms were recorded at baseline and at the annual visit at years 3 and 6 with the patient in the resting supine position using standardized procedures, as previously described. All electrocardiograms received at the central ECG laboratory (EPICORE Center, University of Alberta, Alberta, Edmonton, Canada, and later by EPICARE, Wake Forest University, Winston-Salem, North Carolina) were inspected visually to detect technical errors; those with inadequate quality were rejected from the ECG data files. The ECG data were stored electronically and transmitted daily to the Electrocardiographic Reading Center for analysis using the Novacode criteria measurement and classification system. Incident AF was defined as the presence of AF or atrial flutter on electrocardiograms obtained at year 3 or 6 among participants free of AF at baseline.
The unadjusted incidence rate of AF by year 6 was calculated as the number of events over the number of participants in each quartile of nonfried fish intake. Logistic regression analysis was used to compare the rates of incident AF across the quartiles of nonfried fish intake, with quartile 1 as the referent group. Three models are presented: model 1, age-adjusted; model 2, multivariate model adjusted for age, body mass index, ethnicity, education, diabetes history, systolic blood pressure, treated hypertension, previous cardiovascular disease, and smoking); and model 3, model 2 covariates plus alcohol use, total energy intake (kilocalories per day), fruit/vegetable intake (medium servings per day), and fiber intake (grams per day).
In the secondary analyses, we used logistic regression analysis to test for the association between quartiles of dark fish/tuna intake and incident AF. Similarly, we also tested for associations across quartiles of omega-3 fatty acid intake with incident AF. We conducted additional analyses separately in women <65 and ≥65 years old and blacks and whites. To test for a potential threshold effect, we conducted additional analyses among women with high levels of fish intake (≥5 servings/wk). All statistical analyses were performed using Statistical Analysis Systems, version 9.1 (SAS Institute, Cary, North Carolina). All tests were 2-sided, and p values <0.05 were considered statistically significant.
Results
The average dietary fish intake in the Women’s Health Initiative was 1.5 medium servings/wk, which corresponded to 0.12 g/day of omega-3 fatty acid intake. As listed in Table 1 , women who consumed more servings of nonfried fish were slightly younger, with greater educational attainment and a lower representation of ethnic minorities. In addition, these women were less likely to smoke and more likely to drink alcohol. Greater nonfried fish intake was also associated with other healthy dietary patterns, including greater dietary fiber and fruits and vegetables. As expected, the dietary omega-3 fatty acid levels from fish were greater among women with more nonfried fish intake.
| Quartile (Servings/wk) | 1 (<1/2) | 2 (1/2−1) | 3 (1–2) | 4 (≥2) |
|---|---|---|---|---|
| Patients (n) | 13,002 | 9,816 | 11,126 | 10,776 |
| Age (years) | 63.1 ± 7.1 | 63.0 ± 7.1 | 62.8 ± 6.9 | 62.4 ± 6.9 |
| Race | ||||
| White | 78.5% | 84.4% | 86.2% | 85.1% |
| Black | 12.1% | 9.1% | 8.1% | 9.4% |
| Hispanic | 6.9% | 4.2% | 3.2% | 2.7% |
| Other/unknown | 2.5% | 2.3% | 2.5% | 2.7% |
| College degree or higher | 26.1% | 33.0% | 40.0% | 45.2% |
| Never smoked | 53.8% | 51.9% | 49.4% | 47.2% |
| Current smoker | 9.7% | 8.8% | 7.9% | 7.4% |
| Alcohol consumption | ||||
| None | 15.8% | 10.2% | 7.9% | 6.6% |
| ≥1 drinks/wk | 6.8% | 9.6% | 12.9% | 15.8% |
| Hypertension | 25.8% | 25.8% | 25.6% | 25.4% |
| Diabetes mellitus | 5.4% | 4.5% | 4.2% | 4.7% |
| Previous cardiovascular disease | 7.0% | 6.6% | 6.5% | 6.5% |
| Systolic blood pressure (mm Hg) | 128.6 ± 17.5 | 127.8 ± 17.5 | 127.7 ± 17.3 | 127.5 ± 17.2 |
| Body mass index (kg/m 2 ) | 28.9 ± 5.9 | 28.8 ± 5.9 | 28.9 ± 5.9 | 29.0 ± 6.0 |
| Total cholesterol (mg/dl) ⁎ | 223.3 ± 39.5 | 223.9 ± 40.2 | 226.5 ± 40.0 | 224.7 ± 38.6 |
| Dietary omega-3 from fish, EPA + DHA (g/day) | 0.0 ± 0.0 | 0.1 ± 0.0 | 0.1 ± 0.1 | 0.3 ± 0.2 |
| Dietary fiber (g) | 13.6 ± 6.2 | 14.8 ± 6.1 | 16.1 ± 6.2 | 18.0 ± 6.9 |
| Combined fruits/vegetables (medium serving/day) | 3.2 ± 1.9 | 3.6 ± 1.8 | 4.0 ± 1.9 | 4.5 ± 2.1 |
| Atrial fibrillation | 0.75% | 0.96% | 0.88% | 0.82% |
⁎ To convert total cholesterol in mg/dl to mmol/L, multiply by 0.259.
A total of 378 cases of incident AF (including 42 cases of atrial flutter) occurred, representing 0.85% of the total sample. In the age-adjusted models, no association was found between weekly nonfried fish intake and incident AF. After additional adjustment for baseline risk factors (model 2) and dietary factors (model 3), no association was observed ( Table 2 ). Similar findings were observed for omega-3 fatty acid intake ( Table 3 ), dark fish/tuna intake ( Table 4 ), and fried fish intake (data not shown).
| Model | Nonfried Fish Consumption/wk (Medium Serving) | p Value for Trend | |||
|---|---|---|---|---|---|
| Quartile 1 (<1/2 serving/wk) | Quartile 2 (1/2−1 serving/wk) | Quartile 3 (1–2 servings/wk) | Quartile 4 (≥2 servings/wk) | ||
| Model 1 (age-adjusted) | 1.00 (Referent) | 1.30 (0.98–1.73) | 1.23 (0.93–1.63) | 1.17 (0.88–1.57) | 0.626 |
| Model 2 (age plus baseline risk factors) | 1.00 (Referent) | 1.04 (0.76–1.42) | 1.08 (0.79–1.46) | 1.07 (0.78–1.46) | 0.735 |
| Model 3 (model 2 plus dietary factors) | 1.00 (Referent) | 1.01 (0.74–1.39) | 1.04 (0.76–1.42) | 1.02 (0.73–1.42) | 0.916 |
| Model | Dietary Omega-3 from Fish Intake (g/day) | p Value for Trend | |||
|---|---|---|---|---|---|
| Quartile 1 (<0.049) | Quartile 2 (0.049−<0.092) | Quartile 3 (0.092−<0.157) | Quartile 4 (≥0.157) | ||
| Model 1 (age-adjusted) | 1.00 (Referent) | 1.09 (0.82–1.45) | 1.13 (0.85–1.51) | 1.11 (0.83–1.48) | 0.507 |
| Model 2 (age plus baseline risk factors) | 1.00 (Referent) | 1.02 (0.74–1.39) | 1.10 (0.81–1.51) | 1.08 (0.78–1.48) | 0.420 |
| Model 3 (model 2 plus dietary factors) | 1.00 (Referent) | 1.00 (0.73–1.37) | 1.07 (0.78–1.47) | 1.02 (0.73–1.44) | 0.579 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
