Kenji Wakai

Overview of Regular Dialysis Treatment in Japan (as of 31 December 2008)

A nationwide statistical survey of 4124 dialysis facilities was conducted at the end of 2008 and 4081 facilities (99.0%) responded. The number of patients undergoing dialysis at the end of 2008 was determined to be 283 421, an increase of 8179 patients (3.0%) compared with that at the end of 2007. The number of dialysis patients per million at the end of 2008 was 2220. The crude death rate of dialysis patients from the end of 2007 to the end of 2008 was 9.8%. The mean age of the new patients begun on dialysis was 67.2 years and the mean age of the entire dialysis patient population was 65.3 years. For the primary diseases of the new patients begun on dialysis, the percentages of patients with diabetic nephropathy and chronic glomerulonephritis were 43.3% and 22.8%, respectively. Among the facilities that measured bacterial count in the dialysate solution in 2008, 52.0% of facilities ensured that a minimum dialysate solution volume of 10 mL was sampled. Among the patients treated by facility dialysis, 95.4% of patients were treated three times a week, and the average time required for one treatment was 3.92 ± 0.53 (SD) h. The average amounts of blood flow and dialysate solution flow were 197 ± 31 and 487 ± 33 mL/min, respectively. The number of patients using a polysulfone membrane dialyzer was the largest (50.7%) and the average membrane area was 1.63 ± 0.35 m2. According to the classification of dialyzers by function, the number of patients using a type IV dialyzer was the largest (80.3%). The average concentrations of each electrolyte before treatment in patients treated with blood purification by extracorporeal circulation were 138.8 ± 3.3 mEq/L for serum sodium, 4.96 ± 0.81 mEq/L for serum potassium, 102.1 ± 3.1 mEq/L for serum chloride, and 20.7 ± 3.0 mEq/L for HCO3‐; the average serum pH was 7.35 ± 0.05. Regarding the type of vascular access in patients treated by facility dialysis, in 89.7% of patients an arteriovenous fistula was used and in 7.1% an arteriovenous graft was used. The percentage of hepatitis C virus (HCV)‐positive patients who were HCV‐negative in 2007 was 1.04%; the percentage is particularly high in patients with a period of dialysis of 20 years or longer. The risk of becoming HCV‐positive was high in patients with low serum creatinine, serum albumin, and serum total cholesterol levels, and/or a low body mass index before beginning dialysis.

Estimation of glomerular filtration rate by the MDRD study equation modified for Japanese patients with chronic kidney disease

BackgroundAccurate estimation of the glomerular filtration rate (GFR) is crucial for the detection of chronic kidney disease (CKD). In clinical practice, GFR is estimated from serum creatinine using the Modification of Diet in Renal Disease (MDRD) study equation or the Cockcroft-Gault (CG) equation instead of the time-consuming method of measured clearance for exogenous markers such as inulin. In the present study, the equations originally developed for a Caucasian population were tested in Japanese CKD patients, and modified with the Japanese coefficient determined by the data.MethodsThe abbreviated MDRD study and CG equations were tested in 248 Japanese CKD patients and compared with measured inulin clearance (Cin) and estimated GFR (eGFR). The Japanese coefficient was determined by minimizing the sum of squared errors between eGFR and Cin. Serum creatinine values of the enzyme method in the present study were calibrated to values of the noncompensated Jaffé method by adding 0.207 mg/dl, because the original MDRD study equation was determined by the data for serum creatinine values measured by the noncompensated Jaffé method. The abbreviated MDRD study equation modified with the Japanese coefficient was validated in another set of 269 CKD patients.ResultsThere was a significant discrepancy between measured Cin and eGFR by the 1.0 × MDRD or CG equations. The MDRD study equation modified with the Japanese coefficient (0.881 × MDRD) determined for Japanese CKD patients yielded lower mean difference and higher accuracy for GFR estimation. In particular, in Cin 30–59 ml/min per 1.73 m2, the mean difference was significantly smaller with the 0.881 × MDRD equation than that with the 1.0 × MDRD study equation (1.9 vs 7.9 ml/min per 1.73 m2; P <?0.01), and the accuracy was significantly higher, with 60% vs 39% of the points deviating within 15%, and 97% vs 87% of points within 50%, respectively (both P <?0.01). Validation with the different data set showed the correlation between eGFR and Cin was better with the 0.881 × MDRD equation than with the 1.0 × MDRD study equation. In Cin less than 60 ml/min per 1.73 m2, the accuracy was significantly higher, with 85% vs 69% of the points deviating within 50% (P <?0.01), respectively. The mean difference was also significantly smaller (P <?0.01). However, GFR values calculated by the 0.881 × MDRD equation were still underestimated in the range of Cin over 60 ml/min per 1.73 m2.ConclusionsAlthough the Japanese coefficient improves the accuracy of GFR estimation of the original MDRD study equation, a new equation is needed for more accurate estimation of GFR in Japanese patients with CKD stages 3 and 4.

The relationship between green tea and total caffeine intake and risk for self-reported type 2 diabetes among Japanese adults.

Context Caffeine intake from coffee has been associated with a lower incidence of diabetes. Researchers have not studied the relationship of green tea, a popular beverage in Japan, where the incidence of diabetes is increasing rapidly. Contributions The authors estimated the intake of caffeine-containing beverages in a community-based survey in Japan and measured the 5-year incidence of diabetes. Greater intake of green tea or coffee was associated with a reduced incidence of diabetes. The effect of green tea was largely observed in women, was dose-related, and reflected caffeine intake. Implications Higher intake of caffeine, whether from coffee or green tea, is associated with a lower incidence of diabetes. The Editors The prevalence of type 2 diabetes has increased worldwide, particularly in Asian countries where it was previously low (1). In Japan, population-based studies have shown a 2-fold increase in the prevalence of diabetes during the past 2 decades, from 5% to 10% to 10% to 15% (2). Several cohort studies done in Europe and in the United States reported an association between coffee consumption, a major source of caffeine, and reduced risk for diabetes (3-7). Although these studies did not show any association between consumption of black tea and the risk for diabetes, they did not examine the effect of green or oolong teas, major sources of caffeine in Asian countries. Consumption of green tea is common in Japan; 80% of the population drinks green tea, and the average consumption per capita is 2 cups per day (8). We wanted to determine whether there is a relationship between consumption of green tea and the risk for type 2 diabetes and, if so, whether caffeine fully accounts for this relationship. To examine these questions, we analyzed data from a large cohort study of 19487 middle-aged men and women in 25 communities across Japan. We also examined the effect of age, sex, body mass index (BMI), family history, smoking status, alcohol use, magnesium intake, and physical activity on the association between this mode of caffeine consumption and risk for diabetes. Methods The Japan Collaborative Cohort Study for Evaluation of Cancer Risk (JACC Study) started between 1988 and 1990. A total of 110792 individuals (46465 men and 64327 women) who were 40 to 79 years of age and living in 45 communities across Japan participated in municipal health screening examinations and completed self-administered questionnaires regarding lifestyle and medical history of cardiovascular disease and cancer (9). Informed consent was obtained before the completion of the questionnaire. Participants from 25 of the 45 communities completed 5-year follow-up surveys. Among 35690 participants (15177 men and 20513 women) who were 40 to 65 years of age at baseline without a history of type 2 diabetes, stroke, coronary heart disease, or cancer, 17413 individuals (49%; 6727 men and 10686 women) completed the 5-year follow-up questionnaire and provided valid responses on tea or coffee consumption and history of type 2 diabetes. The data from these 17413 individuals were used for the analyses. The mean age of the nonrespondents was 1 year younger for men (52.3 vs. 53.3 years of age) and did not differ for women (53.1 vs. 53.0 years of age) compared with the respondents. The mean BMI and the prevalence of a BMI of 25.0 kg/m2 or greater did not differ between the respondents and nonrespondents. Mean BMI was 22.7 kg/m2 versus 22.8 kg/m2 for men and 22.8 kg/m2 versus 22.9 kg/m2 for women, respectively; the prevalence of overweight was 18% versus 19% for men and 21% versus 22% for women, respectively. The ethical committees at Nagoya University and the University of Tsukuba approved the study. Assessment of Consumption of Tea and Coffee and Caffeine Intake At baseline, consumption of tea and coffee was assessed by using a self-administered dietary questionnaire. Participants were asked to state their average consumption of green tea, black tea, oolong tea, and coffee during the previous year. They could select any of 4 frequency responses: less than once a week, about 1 to 2 times a week, about 3 to 4 times a week, and almost every day. Participants who selected the response of almost every day were also asked to state their average consumption of these beverages in number of cups per day. We classified the categories of consumption as less than 1 cup per week, 1 to 6 cups per week, 1 to 2 cups per day, 3 to 5 cups per day, and 6 or more cups per day. The highest 2 or 3 consumption categories were combined for coffee, black tea, and oolong tea because of the small number of participants in these categories. The consumption of decaffeinated coffee or tea was not recorded because these products were not commercially available in Japan in the early 1990s. The total intake of caffeine was calculated by adding the caffeine content from each specific beverage (1 cup for coffee or tea) and multiplying it by the participants weight proportional to the frequency of caffeine use. We estimated the size of the cup for each beverage from a validation study (10) and the caffeine content per 100 mL of each beverage from the Japan Food Tables (11). The estimated caffeine content was 153 mg per cup (170 mL) of coffee, 30 mg per cup (200 mL) of green tea, 51 mg per cup (170 mL) of black tea, and 38 mg per cup (190 mL) of oolong tea. The mean caffeine intake was 229 mg/d for men and 215 mg/d for women. Relative proportions of caffeine intake by beverage were 46% from green tea, 44% to 47% from coffee, 3% from black tea, and 4% to 5% from oolong tea. For reproducibility, the Spearman correlation coefficients between the 2 questionnaires, administered 1 year apart for 85 participants (8 men and 77 women), were 0.79 for green tea, 0.87 for coffee, 0.77 for black tea, and 0.56 for oolong tea (10). The validity of the data was confirmed for the 85 participants by comparing the data from the questionnaire with those from four 3-day dietary records collected approximately 3 to 4 months apart (10). The mean frequency of consumption of green tea was 25.4 cups per week according to the questionnaire and 30.1 cups per week according to four 1-week dietary records (Spearman correlation coefficient, 0.47). The respective mean frequencies were 8.0 cups and 7.1 cups per week with a correlation coefficient of 0.79 for coffee, 1.4 cups and 1.6 cups per week with a correlation coefficient of 0.70 for black tea, and 1.8 cups and 1.2 cups per week with a correlation coefficient of 0.55 for oolong tea. When we restricted the data to the 77 women, the results were essentially the same. Assessment of Diabetes Cases Participants who reported having diabetes newly diagnosed by physicians on the 5-year follow-up questionnaire were considered to have incident diabetes. To examine the validity of self-reporting of diabetes, we compared self-report data with laboratory findings and treatment status in a sample of 1230 men and 1837 women. We considered elevated glucose concentrations (fasting serum glucose concentration 7.8 mmol/L [140 mg/dL] or a randomly measured concentration of 11.1 mmol/L [200 mg/dL]) or treatment with oral hypoglycemic agents or insulin to indicate new cases of diabetes. Recent criteria from the American Diabetes Association (12) were not used because the cases in our study were diagnosed before 1995. The sensitivity of self-reporting was 70% for men and 75% for women; the specificity was 95% for men and 98% for women. Statistical Analysis To examine potential confounding variables reported from previous studies (3-7), we presented baseline characteristics according to the frequency of consumption for each beverage. Tests for trends were conducted by using the median values of confounding variables in each category of beverage; the linear regression model was used for continuous variables, and the logistic regression model was used for categorical variables. The odds ratios for incident type 2 diabetes were calculated in each category of beverage consumption and in each quartile of caffeine intake; less than 1 cup per week or the lowest quartile was used as the reference category. We estimated age, sex, and BMI-adjusted odds ratios and multivariable odds ratios using the logistic regression model, adjusting for age (in years), sex, sex-specific quintiles of BMI (weight in kilograms divided by the square of height in meters), parental history of diabetes (yes or no), smoking status (never, former, or current [1 to 19, 20 to 29, or 30 cigarettes/d]), alcohol intake (never, former, or current [1 to 22, 23 to 45, 46 to 68, or 69 g/d]), sex-specific quintiles of magnesium intake, hours of walking (<0.5, 0.5, 0.6 to 0.9, and 1.0 h/d), and hours of participation in sports (<1, 1 to 2, 3 to 4, and 5 h/wk). Sex-specific quintiles of BMI and magnesium intake were used because of different distributions between the sexes. We adjusted for magnesium intake because previous cohort studies indicated an inverse association between magnesium intake and risk for diabetes (13, 14). We conducted a test for trend by treating median values of each category of beverage or caffeine intake as continuous variables. We examined the association between caffeine intake and the risk for diabetes stratified by age group (40 to 54 years and 55 to 65 years), sex, family history of diabetes (yes or no), current smoking status (yes or no), current alcohol intake (yes or no), magnesium intake (below and above the sex-specific median), BMI (<25.0 kg/m2 and 25.0 kg/m2), hours of walking (<0.5 and 0.5 h/d), and hours of participation in sports (<5 and 5 h/wk). The interactions with these stratified variables were tested by using cross-product terms of caffeine intake and the stratified variables. All analyses were conducted by using the SAS statistical package, version 8.2 (SAS Institute Inc., Cary, North Carolina). P values for statistical tests were 2-tailed, and 95% CIs were estimated. Role of the

Epidemiological features of Moyamoya disease in Japan: findings from a nationwide survey

To estimate an annual number of patients treated for Moyamoya disease in Japan and to describe the clinico-epidemiological features, a nationwide epidemiological survey was conducted in 1995. The study consisted of two questionnaires, which were distributed to departments randomly selected, of neurosurgery, neurology and pediatrics in hospitals throughout Japan. The first questionnaire inquired the number of the patients treated in 1994 and the second one detailed clinico-epidemiological information of each patient reported. Following major epidemiological findings emerged from the study: (a) The total annual number of patients treated for Moyamoya disease was estimated as 3900 (95% confidence interval (CI) 3500-4400) in Japan 1994, with the prevalence and incidence rates of 3.16 and 0.35 per 100,000 population, respectively; (b) the sex ratio (females to males) of the patients was 1.8; (c) the peak of age distribution of the patients was observed in 10-14 years old and a smaller peak in their forties; (d) the age at onset was under 10 years old in 47.8% of the patients, but some had developed the disease at the age of 25-49 years; (e) family history of Moyamoya disease was found in 10.0% of the patients; and (f) about 75% of the patients had normal activity of daily life or working ability even before treatment. The present findings were quite comparable with those obtained in the previous nationwide epidemiological survey in 1990.

Prevalence and Clinicoepidemiological Features of Moyamoya Disease in Japan Findings From a Nationwide Epidemiological Survey

Background and Purpose— The objectives of the present study were to estimate an annual number of patients with moyamoya disease in Japan and to describe the clinicoepidemiological features of the disease. Methods— The study consisted of 2 questionnaire surveys, which were distributed to randomly selected departments of neurosurgery, internal medicine, neurology, cerebrovascular medicine, and pediatrics in hospitals throughout Japan. The first survey inquired about the number of the patients treated in 2003, and the second requested additional detailed clinicoepidemiological information about each patient identified in the first survey. Results— In 2003, the total number of patients treated in Japan was estimated at 7700 (95% confidence interval, 6300 to 9300). Sex ratio (women to men) of the patients was 1.8. For men, the peak of moyamoya disease was observed in patients aged 10 to 14 years and for women aged 20 to 24 years. Annual rate of newly diagnosed cases in 2003 was 0.54 per 100 000 population. Family history of moyamoya disease was found in 12.1% of the patients. The majority (77.9%) were treated as outpatients. Conclusions— Although the clinicoepidemiological features of the patients in the present study were almost similar to those obtained in previous ones, the estimated prevalence of moyamoya disease in Japan has almost doubled during the recent decade (3900 in 1994 and 7700 in 2003). The increase could partly be explained by the increase in newly diagnosed cases (0.35 in 1994 and 0.54 in 2003 per 100 000 population).

Dietary risk factors for inflammatory bowel disease A Multicenter Case‐Control Study in Japan

&NA; To evaluate the role of dietary factors in the etiology of inflammatory bowel disease (IBD), we conducted a multicenter hospital‐based case‐control study in a Japanese population. Cases were IBD patients aged 15 to 34 years [ulcerative colitis (UC) 111 patients; Crohns disease (CD) 128 patients] within 3 years after diagnosis in 13 hospitals. One control subject was recruited for each case who was matched for sex, age, and hospital. A semiquantitative food frequency questionnaire was used to estimate preillness intakes of food groups and nutrients. All the available control subjects (n = 219) were pooled, and unconditional logistic models were applied to calculate odds ratios (ORs). In the food groups, a higher consumption of sweets was positively associated with UC risk [OR for the highest versus lowest quartile, 2.86; 95% confidence interval (CI), 1.24 to 6.57], whereas the consumption of sugars and sweeteners (OR, 2.12; 95% CI, 1.08 to 4.17), sweets (OR, 2.83; 95% CI, 1.38 to 5.83), fats and oils (OR, 2.64; 95% CI, 1.29 to 5.39), and fish and shellfish (OR, 2.41; 95% CI, 1.18‐4.89) were positively associated with CD risk. In respect to nutrients, the intake of vitamin C (OR, 0.45; 95% CI, 0.21 to 0.99) was negatively related to UC risk, while the intake of total fat (OR, 2.86; 95% CI, 1.39 to 5.90), monounsaturated fatty acids (OR, 2.49; 95% CI, 1.23 to 5.03) and polyunsaturated fatty acids (OR, 2.31; 95% CI, 1.12 to 4.79), vitamin E (OR, 3.23; 95% CI, 1.45 to 7.17), and n‐3 (OR, 3.24; 95% CI, 1.52 to 6.88) and n‐6 fatty acids (OR, 2.57; 95% CI, 1.24 to 5.32) was positively associated with CD risk. Although this study suffers from the shortcoming of recall bias, which is inherent in most retrospective studies (prospective studies are warranted to confirm the associations between diet and IBD risk), the present findings suggest the importance of dietary factors for IBD prevention.

Common Defects of ABCG2, a High-Capacity Urate Exporter, Cause Gout: A Function-Based Genetic Analysis in a Japanese Population

Dysfunctional genotype combinations of polymorphic adenosine 5′-triphosphate–binding cassette transporter gene ABCG2/BCRP, which encodes a high-capacity urate secretion transporter in human gut and kidney, are major causes of gout. Gout, the “Disease of Kings” as it is often known, is a painful medical condition characterized by sharp acute pain in bone joints, due to the high deposition of uric acid crystals from the blood serum into the surrounding cartilage. It affects approximately 1% of the U.S. population and remains a significant public health concern. The prevalence of gout is much higher in certain Asian ethnic groups, and is also reportedly rising in African Americans. Current medical treatments are aimed at ameliorating pain severity, but as the underlying genetic etiology of the disease unfolds, new targets for future therapies are likely to be found. Although genome-wide association studies (GWAS) have enabled the calculation of risk predispositions for a wide variety of complex diseases, the relation of gene function to the causality of disease-related mutations has remained largely unclear. A recent U.S. population–based study supported an association between urate levels and gout in individuals carrying variants in a multifunctional transporter gene, ABCG2. This study identified Q141K as a high-risk variant in nearly 10% of gout cases in Caucasians. Now, a team led by Hirotaka Matsuo report that in a Japanese population, another risk variant in ABCG2, namely the Q126X nonfunctional mutation, confers an even higher risk associated with an increase in uric acid deposition in the blood and may cause gout in Asians. Because this gene is responsible for giving rise to a protein that transports harmful waste products and metabolites out of the kidney and gut, they extensively validate the biological activity of ABCG2 using functional assays in vitro that effectively recapitulate human data obtained from Japanese individuals afflicted with the disease. These findings lend weight to previously reported GWAS; moreover, these newly identified specific high-risk variants that block urate secretion may serve as potential intervention points for quelling the disease. Gout based on hyperuricemia is a common disease with a genetic predisposition, which causes acute arthritis. The ABCG2/BCRP gene, located in a gout-susceptibility locus on chromosome 4q, has been identified by recent genome-wide association studies of serum uric acid concentrations and gout. Urate transport assays demonstrated that ABCG2 is a high-capacity urate secretion transporter. Sequencing of the ABCG2 gene in 90 hyperuricemia patients revealed several nonfunctional ABCG2 mutations, including Q126X. Quantitative trait locus analysis of 739 individuals showed that a common dysfunctional variant of ABCG2, Q141K, increases serum uric acid. Q126X is assigned to the different disease haplotype from Q141K and increases gout risk, conferring an odds ratio of 5.97. Furthermore, 10% of gout patients (16 out of 159 cases) had genotype combinations resulting in more than 75% reduction of ABCG2 function (odds ratio, 25.8). Our findings indicate that nonfunctional variants of ABCG2 essentially block gut and renal urate excretion and cause gout.

Prevalence of chronic kidney disease (CKD) in the Japanese general population predicted by the MDRD equation modified by a Japanese coefficient

BackgroundThe number of patients with end-stage renal disease (ESRD) in Japan has continuously increased in the past three decades. In 2005, 36 063 patients whose average age was 66 years entered a new dialysis program. This large number of ESRD patients could be just the tip of the iceberg of an increasing number of patients with chronic kidney disease (CKD). However, to date, a nationwide epidemiological study has not been conducted yet to survey the CKD population.MethodsData for 527 594 (male, 211 034; female, 316 560) participants were obtained from the general adult population aged over 20 years who received annual health check programs in 2000–2004, from seven different prefectures in Japan: Hokkaido, Fukushima, Ibaraki, Tokyo, Osaka, Fukuoka, and Okinawa prefectures. The glomerular filtration rate (GFR) for each participant was estimated from the serum creatinine values, using the abbreviated Modification of Diet in Renal Disease (MDRD) study equation modified by the Japanese coefficient.ResultsThe prevalences of CKD stage 3 in the study population, stratified by age groups of 20–29, 30–39, 40–49, 50–59, 60–69, 70–79, and 80–89 years, were 1.4%, 3.6%, 10.8%, 15.9%, 31.8%, 44.0%, and 59.1%, respectively, predicting 19.1 million patients with stage 3 CKD in the Japanese general adult population of 103.2 million in 2004. CKD stage 4 + 5 was predicted in 200 000 patients in the Japanese general adult population. Comorbidity of hypertension, diabetes, and proteinuria increased as the estimated GFR (eGFR) decreased. The prevalence of concurrent CKD was significantly higher in hypertensive and diabetic populations than in the study population overall when CKD was defined as being present with an eGFR of less than 40 ml/min per 1.73 m2 instead of less than 60 ml/min per 1.73 m2.ConclusionsAbout 20% of the Japanese adult population (i.e., approximately 19 million people) are predicted to have stage 3 to 5 CKD, as defined by a GFR of less than 60 ml/min per 1.73 m2.

DIETARY INTAKE AND SOURCES OF ISOFLAVONES AMONG JAPANESE

We examined the dietary intake and sources of isoflavones (daidzein and genistein) among Japanese subjects based on dietary records (DRs). The subjects comprised two groups: 1,232 who completed one-day DRs (Group 1) and 88 men and women who kept four four-day (16-day) DRs. For quantitative data on the level of daidzein and genistein in soy foods, we extensively reviewed the literature, particularly for Japanese soy foods, and adopted the median value for each food. The median intake of daidzein was 12.1 and 9.5 mg/day among Groups 1 and 2, respectively, while the corresponding values for genistein were 19.6 and 14.9 mg/day. The top four foods (tofu, miso, natto, and fried tofu) covered about 90% of the population intake of daidzein and genistein. It did not seem feasible to estimate ones intake of isoflavones by using dietary recording/recall in epidemiological studies, since the day-to-day variation in intake was too large, the within-person coefficient of variation being 89.1% for daidzein and genistein. Therefore, we should use other methods, such as food-frequency questionnaires, focusing on the four major sources of isoflavones, to assess individual isoflavone intake.

Fish, ω-3 Polyunsaturated Fatty Acids, and Mortality From Cardiovascular Diseases in a Nationwide Community-Based Cohort of Japanese Men and Women: The JACC (Japan Collaborative Cohort Study for Evaluation of Cancer Risk) Study

OBJECTIVES The objective of our study was to test the hypothesis that fish or omega-3 polyunsaturated fatty acids (PUFA) intakes would be inversely associated with risks of mortality from ischemic heart disease, cardiac arrest, heart failure, stroke, and total cardiovascular disease. BACKGROUND Data on associations of dietary intake of fish and of omega-3 PUFA with risk of cardiovascular disease among Asian societies have been limited. METHODS We conducted a prospective study consisting of 57,972 Japanese men and women. Dietary intakes of fish and omega-3 PUFA were determined by food frequency questionnaire, and participants were followed up for 12.7 years. Hazard ratios and 95% confidence intervals were calculated according to quintiles of fish or omega-3 PUFA intake. RESULTS We observed generally inverse associations of fish and omega-3 PUFA intakes with risks of mortality from heart failure (multivariable hazard ratio [95% confidence interval] for highest versus lowest quintiles = 0.76 [0.53 to 1.09] for fish and 0.58 [0.36 to 0.93] for omega-3 PUFA). Associations with ischemic heart disease or myocardial infarction were relatively weak and not statistically significant after adjustment for potential risk factors. Neither fish nor omega-3 PUFA dietary intake was associated with mortality from total stroke, its subtypes, or cardiac arrest. For mortality from total cardiovascular disease, intakes of fish and omega-3 PUFA were associated with 18% to 19% lower risk. CONCLUSIONS We found an inverse association between fish and omega-3 PUFA dietary intakes and cardiovascular mortality, especially for heart failure, suggesting a protective effect of fish intake on cardiovascular diseases.