Randomized controlled trial characteristics

Two RCTs evaluated the effect of oily fish (i.e., grouper, sea bream, kingfish, emperor, snapper)¹ or tuna,³ among schoolchildren on intermediate CVD outcomes. In the Fish Feeding Study conducted in Oman, children age 10 years were randomized to receive a school meal with either 100 grams of oily fish (n=96) or a cheese/salad sandwich (control; n=102) five times per week for 12 weeks.¹ Total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides were measured at post-intervention. An RCT conducted in Mexico examined the effect of 6, 7, or 8 grams of tuna (n~20-32); however, the number of meals provided, length of intervention, and control conditions were not described. Blood pressure, total cholesterol, and triglycerides were assessed following an unknown period, and results were stratified by sex.³ Compliance was not reported in either study.

Prospective cohort study characteristics

Two PCSs, one from the OPUS School Meal Study (Denmark) and one from the Boyd Orr Cohort (United Kingdom [U.K.]) assessed fish intake, analyzed continuously² or in quartiles,⁴ respectively. The OPUS School Meal Study, assessed fish intake at age 10 years using a 7-day dietary record at baseline, 3 months follow-up, and 6 months follow-up.² Intermediate outcomes were assessed after 3 and 6 months, including diastolic blood pressure, HDL cholesterol, and triglycerides. In the Boyd Orr Cohort, fish intake at age 7.5 years was indirectly assessed between 1937 and 1939 using a 7-day household inventory, dividing the total food expenditure by the total number of household members.⁴ This long-term longitudinal study followed these children for 60 years at which time stroke mortality and coronary heart disease (CHD) mortality were assessed.⁴

Intermediate outcomes (cardiovascular risk factor assessment)

Blood pressure: One RCT and one PCS did not find a significant relationship between intake of 6, 7, or 8 grams of tuna (length of intervention unknown and control group unspecified) or fish intake (grams per day), respectively, and blood pressure at approximately age 10-12 years.^2,3

Blood lipid levels: One RCT found a statistically significant relationship between fish intake and blood lipids: obese girls age 11-12 years who consumed 6, 7, or 8 grams of tuna had lower total cholesterol levels compared to controls. This was not significant in obese boys. Additionally, obese boys and girls age 11-12 years who consumed 6, 7, or 8 grams of tuna had significantly lower triglycerides compared to the control group.³ This study did not report the frequency or duration of the intervention; therefore, it was difficult to interpret the effect of the intervention. The OPUS PCS detected a statistically significant association between fish intake at age 10 years and fasting triglyceride levels three months later; these data were not reported.² However, interpretation of data on triglyceride change is difficult when not reported concurrently with a validated surrogate endpoint for CVD such as LDL cholesterol. In contrast, an RCT did not find an effect of an oily fish meal intervention (100 grams per meal, 5 times per week for 12 weeks; n=96) on triglyceride levels compared to cheese/salad sandwich (control, n=102).¹

Two studies (one RCT and one PCS) did not find a significant relationship between fish intake and HDL cholesterol levels.^1,2 Only one study (an RCT) assessed LDL cholesterol levels and did not find a statistically significant effect of oily fish intake.¹

CVD Endpoint outcomes

CVD-related mortality: One PCS (N=4,028) detected a statistically significant linear trend between greater fish intake at age 7.5 years and greater risk of stroke mortality after 60 years of follow-up. When examining comparisons between quartiles of intake, there was a statistically significant association between the highest quartile of fish intake at age 7.5 years (mean=44.5 grams, standard deviation [SD]=15.5), compared to the lowest quartile (mean=1.8 grams, SD=2.4) of intake and greater risk of stroke mortality approximately 60 years later.⁴ There was not a statistically significant difference in risk of stroke mortality between the second and third quartiles of fish intake (means=11.3 and 21.6 grams, respectively) and the lowest intake quartile. No association was detected between oily fish intake at age 7.5 years and stroke mortality (n=4,028). However, there was no statistically significant association between either fish intake or oily fish intake at age 7.5 years and CHD mortality after approximately 60 years. Results were interpreted with caution because this study had a serious limitation related to the method of fish assessment. Specifically, the dietary assessment completed between 1937 and 1939 was based on a weighed inventory of all foods available in the household divided by the total number of household members. Thus, no information was provided regarding the validity or accuracy of the actual intake of fish among children.

Assessment of the evidenceⁱⁱ

As outlined and described below, the body of evidence examining seafood consumption during childhood and adolescence and risk of CVD was assessed for the following elements used when grading the strength of evidence.

• Risk of Bias:

  • RCTs (Table 2)

    • Risk of bias due to randomization: it was unclear whether one trial randomized participants using a non-biased method.³
    • Risk of bias due to deviations from intended interventions:

      • Dietary compliance was not reported in either trial.^1,3
      • Results were difficult to interpret due to incomplete information on fish intake, the control group, and timing of outcome assessment in one trial.³
  • PCSs (Table 3)

    • Risk of bias due to confounding: both studies did not control for all key confounders, particularly family history of CVD.^2,4
    • Risk of bias due to selection of participants: follow-up period may have been different for all participants in one study.⁴
    • Risk of bias due to classification of exposures: fish intake was measured indirectly using a non-validated 7-day household inventory and a non-validated 7-day dietary record.^2,4
    • Risk of bias due to missing data: participants were excluded for missing data on other variables besides the exposure and outcome data in both studies.²
• Consistency:

  • Consistency could not be assessed due to an insufficient number of studies and serious methodological limitations.
• Directness:

  • Three studies were designed to examine seafood intake during childhood and adolescence and CVD outcomes.
  • One PCS was derived from a cluster randomized crossover trial designed to examine the effect of school lunches based on the new Nordic diet on metabolic syndrome and cognition.²
• Precision:

  • In one study, subgroupings had relatively small sample sizes.³
• Generalizability:

  • Two studies were conducted in Europe (U.K. and Denmark),^2,4 one in Mexico,³ and one in Oman.¹
  • One study exclusively enrolled obese adolescents.³
  • In the one study that provided information on parental education level, 66% reported having had “higher education”.²

Other considerations

A large, comprehensive search was conducted in multiple databases for this systematic review. Risk of publication bias is always of potential concern, however, both relatively small and large studies were included in this review, reporting both null and statistically significant results. Therefore, risk of publication bias is likely low across this body of evidence.