Total fish consumption and total stroke risk. Effect modification.

  • For total stroke risk, stratified analysis was performed to examine the possibility that effects might differ between different levels of intake (Table 1). Four different levels of consumption were considered, relative to nonconsumption: < 1, 1, 2, and ≥ 3 servings/week. Data from 17 cohorts, including 19.971 cases could be included. Data from Bravata DM [35] could not be included because levels of intake were not clearly defined, and no RR's were provided. The protective effect became stronger with increasing levels of consumption (RR = 0.98, 0.94, 0.93, and 0.89 for < 1, 1, 2, and ≥ 3 servings/wk, respectively). Effects sizes were heavily dependent on data from one single cohort which included the majority of cases (Kinjo Y [10]). Exclusion of data from this cohort strenghtened all effect sizes (RR = 0.97, 0.89, 0.86, and 0.86 for < 1, 1, 2, and ≥ 3 servings/wk, respectively).
  • For total stroke risk, analysis stratified by gender showed that (non)significant protective effects (P = < 0.10) were restricted to female cohorts (Table 2). In addition, within-cohort comparisons (n = 5) showed that RR's were consistently lower among women. Average RR's are 0.99 for men, and 0.80 for women.
    But authors from 3 other cohorts stated that effects did not differ significantly between men and women (Kinjo Y [10], Montonen J [13], Mozaffarian D [24]).
  • For total stroke risk, (non)significant protective effects were restricted to findings among US and Asian cohorts (Tables 3-5). Average RR's suggested possible protective effects among Asian cohorts (6 cohorts: RR = 0.87), and US cohorts (8 cohorts: RR = 0.77), but not among European cohorts (7 cohorts: RR = 0.96).
  • For ischemic stroke risk, (non)significant protective effects were restricted to US cohorts (Tables 6-8). Average RR's suggested a possible protective effect among US cohorts (3 cohorts: RR = 0.57), but not among European cohorts (3 cohorts: RR = 0.92), and Asian cohorts (4 cohorts: RR = 0.98).
  • For hemorrhagic stroke risk, (non)significant protective effects were restricted to findings among Asian cohorts (Tables 9-11). Average RR's suggested a possible protective effect among Asian cohorts (4 cohorts: RR = 0.87), and European cohorts (3 cohorts: RR = 0.84), but not among US cohorts (3 cohorts: RR = 1.16).
  • In one cohort a significant protective effect against ischemic stroke risk was found among men with median intake (0.91-1.14 g/d) of alpha-linolenic acid only (He K [14]).
  • No significant effect modification was found by intakes of vitamin E (He K [14]) or aspirin (Iso H [14], He K [14], Mozaffarian D [24]).
  • No significant effect modification was found by serum cholesterol (Montonen J [13]) systolic blood pressure (Mozaffarian D [2005]), or hypertension (Montonen J [13], Mozaffarian D [24]).
  • No significant effect modification was found by age (Gillum RF [17, Mozaffarian D [24]), race (Gillum RF [17]), education (Mozaffarian D [25]), rural/urban status (Kinjo Y [10]), BMI (Montonen J [13]), or smoking status (Kinjo Y [10], Mozaffarian D [24]).

Table 1.
Prospective studies of total fish and total stroke risk.
Stratified by intake in servings fish/week:
AuthorCohort nameCases< 1 serving/week1 serving/week2 servings/week≥ 3 servings/week
42) Larsson SC (2011)The Swedish Mammography Cohort1,680HR = 1HR = 0.91HR = 0.95HR = 0.86
38) Kaushik S (2008)The Blue Mountains Eye Study69HR = 1HR = 0.51HR = 0.62HR = 0.62
36) Yamagishi K (2008)The JACC Study972HR = 1HR = 1HR = 0.98HR = 0.92
34) Myint PK (2006)The EPIC-Norfolk Study217 men, and

204 women		Men: RR = 1

Women: RR = 1		Men: RR = 1.12

Women: RR = 0.74		Men: RR = 1.12

Women: RR = 0.74		Men: RR = 1.34

Women: RR = 0.86
32) Nakamura Y (2005)The NIPPON DATA80288RR = 1*RR = 0.75RR = 0.75RR = 0.89
31) Ness AR (2005)The Boyd Orr Cohort83RR = 0.90RR = 0.79RR = 2.01RR = 2.01
29) Folsom AR (2004)The Iowa Women's Health Study313RR = 1.15RR = 0.95RR = 0.90RR = 1.06
27) Sauvaget C (2003)The Hiroshima/Nagasaki Life Span Study1,260HR = 0.82HR = 0.82HR = 0.74HR = 0.73
24) Mozaffarian D (2005)The Cardiovascular Health Study626HR = 0.94HR = 0.75HR = 0.75HR = 0.75
19) Yuan JM (2001)The Shanghai Cohort Study480RR = 1RR = 1RR = 0.84RR = 0.93
17) Gillum RF (1996)The NHANES I262 white men

251 white women		White men: RR = 1.27

White women: RR = 0.78		White men: RR = 1.23

White women: RR = 0.77		White men: RR = 0.85

White women: RR = 0.55		White men: RR = 0.85

White women: RR = 0.55
15) Morris MC (1995)The Physician's Health Study173RR = 1RR = 0.9RR = 0.8RR = 0.7
14) He K (2002)The Health Professsional's Follow-up Study608RR = 0.87RR = 0.74RR = 0.67RR = 0.75
14) Iso H (2001)The Nurses' Health Study574RR = 0.97RR = 0.78RR = 0.73RR = 0.61
13) Montonen J (2009)The Finnish Mobile Clinic Health Survey659RR = 1RR = 1.01RR = 1.05RR = 1.04
10) Kinjo Y (1999)No cohort name defined11,030RR = 1RR = 0.98RR = 0.98RR = RR = 0.92
2) Orencia AJ (1996)The Western Electric Study222HR = 0.98HR = 0.98HR = 0.94HR = 1.26
Total number of cases: 19,971Average RR = 0.98Average RR = 0.94Average RR = 0.93Average RR = 0.89
Excluding data from Kinjo Y [10]Total number of cases: 8.941Average RR = 0.97Average RR = 0.89Average RR = 0.86Average RR = 0.86
* = Lowest level of consumption was not the reference goup. RR for the lowest level of consumption was set to 1. Other RR's were adjusted for this difference in setting.

Table 2.
Prospective studies of total fish and stroke risk.
Stratified by gender:
AuthorCohort nameCasesRelative Risk among menRelative Risk among women
42) Larsson SC (2011)The Swedish Mammography Cohort1,680-RR = 0.84 (0.71-0.98; P = 0.049)
36) Iso H (2007)The JACC Study1,230 men, and

1,014 women		HR = 1.00 (0.87-1.16)HR = 0.87 (0.73-1.02; P = < 0.10)
35) Bravata DM (2007)No cohort name defined369HR = 0.89 (0.59-1.36)-
34) Myint PK (2006)The EPIC-Norfolk Study217 men, and

204 women		RR = 1.34 (0.93-2.93; P = 0.26)RR = 0.86 (0.60-1.24; P = 0.29)
29) Folsom AR (2004)The Iowa Women's Health Study313-RR = 1.06 (0.67-1.67; P = 0.65)
23) Nagata C (2002)The Takayama Study137 men, and

132 women		HR = 1.19 (0.78-1.81; P = 0.37)HR = 0.87 (0.58-1.30; P = 0.49)
22) Wennberg M (2007)The MONICA Study189 men, and

128 women		OR = 1.24 (1.01-1.51; P = 0.04)OR = 0.90 (0.73-1.12; P = 0.35)
19) Yuan JM (2001)The Shanghai Cohort Study480RR = 1.05 (0.77-1.43; P = 0.47)-
17) Gillum RF (1996)The NHANES I262 white men,

251 white women		RR = 0.85 (0.49-1.46)RR = 0.55 (0.32-0.93; P = 0.05)
15) Morris MC (1995)The Physician's Health Study173RR = 0.6 (0.3-1.6; P = 0.13)-
14) He K (2002)The Health Professsional's Follow-up Study608RR = 0.83 (0.53-1.29; P = 0.81)-
14) Iso H (2001)The Nurses' Health Study574-RR = 0.48 (0.21-1.06; P = 0.06)
6) Lapidus L (1986)No cohort name defined13-No significant association
2) Orencia AJ (1996)The Western Electric Study222HR = 1.26 (0.74-2.16)-
1) Keli SO (1994)The Zutphen Study42HR = 0.71 (0.38-1.33)-
Total number of cases: 3,929 men, and
4,309 women		Average RR = 0.99Average RR = 0.80


Table 3.
Prospective studies of total fish and total stroke risk among US cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
35) Bravata DM (2007)No cohort name defined369RiskHR = 0.89 (0.59-1.36)
29) Folsom AR (2004)The Iowa Women's Health Study313MortalityRR = 1.06 (0.67-1.67; P = 0.65)
24) Mozaffarian D (2005)The Cardiovascular Health Study626RiskHR = 0.74 (0.54-1.02; P = 0.04)
17) Gillum RF (1996)The NHANES I262 white men,

251 white women,

107 black men and women		RiskWhite men: RR = 0.85 (0.49-1.46).

White women: RR = 0.55 (0.32-0.93; P = 0.05).

Black men and women: RR = 0.51 (0.30-0.88; P = < 0.05).
15) Morris MC (1995)The Physician's Health Study173RiskRR = 0.6 (0.3-1.6; P = 0.13)
14) He K (2002)The Health Professsional's Follow-up Study608RiskRR = 0.83 (0.53-1.29; P = 0.81)
14) Iso H (2001)The Nurses' Health Study574RiskRR = 0.48 (0.21-1.06; P = 0.06)
2) Orencia AJ (1996)The Western Electric Study222RiskHR = 1.26 (0.74-2.16)
Total number of cases: 3,505Average RR = 0.77


Table 4.
Prospective studies of total fish and total stroke risk among European cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
42) Larsson SC (2011)The Swedish Mammography Cohort1,680RiskRR = 0.84 (0.71-0.98; P = 0.049)
34) Myint PK (2006)The EPIC-Norfolk Study217 men, and

204 women		RiskMen: RR = 1.34 (0.93-2.93; P = 0.26).

Women: RR = 0.86 (0.60-1.24; P = 0.29).
31) Ness AR (2005)The Boyd Orr Cohort83MortalityRR = 2.01 (1.09-3.69; P = 0.01)
22) Wennberg M (2007)The MONICA Study189 men, and

128 women		RiskMen: OR = 1.24 (1.01-1.51; P = 0.04).

Women: OR = 0.90 (0.73-1.12; P = 0.35).
13) Montonen J (2009)The Finnish Mobile Clinic Health Survey659RiskRR = 1.01 (0.81-1.27; P = 0.80)
6) Lapidus L (1986)No cohort name defined13RiskNo significant association
1) Keli SO (1994)The Zutphen Study42RiskHR = 0.71 (0.38-1.33)
Total number of cases: 3,202Average RR = 0.96


Table 5.
Prospective studies of total fish and total stroke risk among Asian cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
36) Yamagishi K (2008)The JACC Study972MortalityHR = 0.91 (0.74-1.13; P = 0.40)
32) Nakamura Y (2005)The NIPPON DATA80288MortalityRR = 1.28 (0.71-2.32; P = 0.50) for the highest vs second quintile of consumption
27) Sauvaget C (2003)The Hiroshima/Nagasaki Life Span Study1,462MortalityHR = 0.85 (0.75-0.98; P = 0.02)
23) Nagata C (2002)The Takayama Study137 men, and

132 women		MortalityMen: HR = 1.19 (0.78-1.81; P = 0.37).

Women: HR = 0.87 (0.58-1.30; P = 0.49).
19) Yuan JM (2001)The Shanghai Cohort Study480MortalityRR = 1.05 (0.77-1.43; P = 0.47)
10) Kinjo Y (1999)No cohort name defined11,030MortalityRR = 0.86 (0.79-0.94)
Total number of cases: 14,501Average RR = 0.87


Table 6.
Prospective studies of total fish and ischemic stroke risk among US cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
24) Mozaffarian D (2005)The Cardiovascular Health Study537Ischemic stroke riskHR = 0.70 (0.50-0.99; P = 0.02)
14) He K (2002)The Health Professsional's Follow-up Study377Ischemic stroke riskRR = 0.54 (0.31-0.94; P = 0.28)
14) Iso H (2001)The Nurses' Health Study303Ischemic stroke riskRR = 0.38 (0.12-1.19; P = 0.09)
Total number of cases: 1,217Average RR = 0.57


Table 7.
Prospective studies of total fish and ischemic stroke risk among European cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
42) Larsson SC (2011)The Swedish Mammography Cohort1,310Cerebral infarction riskRR = 0.87 (0.73-1.04; P = 0.19)
22) Wennberg M (2007)The MONICA Study147 men, and

111 women		Ischemic stroke riskMen: OR = 1.25 (1.00-1.56; P = 0.04).

Women: OR = 0.93 (0.74-1.17; P = 0.51).
13) Montonen J (2009)The Finnish Mobile Clinic Health Survey364Thrombosis or embolia riskRR = 0.99 (0.73-1.35; P = 0.96)
Total number of cases: 1,932Average RR = 0.92


Table 8.
Prospective studies of total fish and ischemic stroke risk among Asian cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
36) Yamagishi K (2008)The JACC Study319Ischemic stroke mortalityHR = 0.93 (0.65-1.34; P = 0.78)
32) Nakamura Y (2005)The NIPPON DATA80165Cerebral infarction deathRR = 1.11 (0.50-2.47; P = 0.70) for the highest vs second quintile of consumption
27) Sauvaget C (2003)The Hiroshima/Nagasaki Life Span Study665Cerebral infarction mortalityHR = 0.94 (0.77-1.14; P = 0.50)
10) Kinjo Y (1999)No cohort name defined4,084Embolism and thrombosis mortalityRR = 0.99 (0.86-1.14)
Total number of cases: 5,233Average RR = 0.98


Table 9.
Prospective studies of total fish and hemmorhagic stroke risk among US cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
24) Mozaffarian D (2005)The Cardiovascular Health Study73Hemorrhagic stroke riskHR = 0.93 (0.37-2.33; P = 0.66)
14) He K (2002)The Health Professsional's Follow-up Study106Hemorrhagic stroke riskRR = 1.55 (0.45-5.35; P = 0.70)
14) Iso H (2001)The Nurses' Health Study181Hemorrhagic stroke riskRR = 1.02 (0.34-3.10; P = 0.33)
Total number of cases: 360Average RR = 1.16


Table 10.
Prospective studies of total fish and hemmorhagic stroke risk among European cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
42) Larsson S (2011)The Swedish Mammography Cohort233Hemorrhagic stroke riskRR = 0.67 (0.42-1.08; P = 0.08)
22) Wennberg M (2007)The MONICA Study39 men, and

15 women		Hemorrhagic stroke riskMen: OR = 1.14 (0.69-1.88; P = 0.59).

Women: OR = 0.61 (0.23-1.57; P = 0.31).
13) Montonen J (2009)The Finnish Mobile Clinic Health Survey80Intracerebral haemorrhage riskRR = 1.23 (0.63-2.42; P = 0.41)
Total number of cases: 367Average RR = 0.84


Table 11.
Prospective studies of total fish and hemmorhagic stroke risk among Asian cohorts:
AuthorCohort nameCasesEnd pointRelative Risk (RR)
36) Yamagishi K (2008)The JACC Study180 intraparenchymal hemorrhage, and

153 subarachnoid hemorrhage		MortalityIntraparenchymal hemorrhage: HR = 0.95 (0.62-1.47; P = 0.58).

Subarachnoid hemorrhage: HR = 0.96 (0.55-1.68; P = 0.84).
32) Nakamura Y (2005)The NIPPON DATA8063Cerebral hemorrhage deathRR = 0.93 (0.20-4.28; P = 0.97) for the highest vs second quintile of consumption
27) Sauvaget C (2003)The Hiroshima/Nagasaki Life Span Study354Intracerebral haemorrhage mortalityHR = 0.70 (0.54-0.92; P = 0.008)
10) Kinjo Y (1999)No cohort name defined4,773Cerebral haemorrhage mortalityRR = 0.87 (0.76-0.98)
Total number of cases: 5,523Average RR = 0.87