Skip to main content
. 2018 Nov 27;2018(11):CD012345. doi: 10.1002/14651858.CD012345.pub3

Summary of findings 2. Higher polyunsaturated fatty acid (PUFA) compared to lower PUFA for cardiovascular disease ‐ additional key outcomes.

Higher PUFA compared to lower PUFA ‐ dichotomous secondary outcomes for prevention of cardiovascular disease
Patient or population: people with or without existing cardiovascular disease, men and women
 Setting: includes free‐living participants and those living in institutions. Includes participants from all continents but most events and assessments occurred in trials carried out in Europe or North America.
 Intervention: higher PUFA intake
 Comparison: lower PUFA intake
Eligible trials compared higher with lower total PUFA intakes. The intervention had to be dietary supplementation, or a provided diet, or advice on diet. The advice, foodstuffs or supplements had to aim to increase or decrease total PUFA intake, or a dietary component high in total PUFA intake such as vegetable oil, or, if no clear aim was stated (but implied, such as aiming to provide a 'heart health', 'reduced fat' or 'Mediterranean' diet) then the intervention had to achieve an increase or decrease of at least 10% of the baseline total PUFA level.
Outcomes Anticipated absolute effects* (95% CI) Relative effect
 (95% CI) № of participants
 (trials) Certainty of the evidence
 (GRADE) Comments
Risk with lower PUFA Risk with higher PUFA
Cardiovascular mortality
 Follow‐up: range 12 months to 96 months No CVD at baseline (primary prevention) RR 1.02
 (0.82 to 1.26) 15,107
 (16 RCTs) ⊕⊕⊝⊝
 Lowa Increasing PUFA intake may have little or no effect on cardiovascular mortality (risk alters from 4.8% to 4.9% in the study population), low‐quality evidence.
36 per 1000 31 per 1000
 (19 to 50)
CVD at baseline (secondary prevention)
57 per 1000 64 per 1000
 (52 to 77)
Cardiovascular events
Follow‐up: range 12 months to 96 months		 No CVD at baseline (primary prevention) RR 0.89
 (0.79 to 1.01) 17,799
 (21 RCTs) ⊕⊕⊕⊝
 Moderateb Increasing PUFA intake probably reduces risk of CVD events (from 14.6% to 13.0% in the study population, NNT = 63), moderate‐quality evidence.
54 per 1000 46 per 1000
 (39 to 54)
CVD at baseline (secondary prevention)
233 per 1000 208 per 1000
 (175 to 245)
Coronary heart disease mortality
 Follow‐up: range 12 months to 96 months No CVD at baseline (primary prevention) RR 0.91
 (0.78 to 1.06) 8810
 (9 RCTs) ⊕⊕⊝⊝
 Lowc Increasing PUFA intake may reduce risk of CHD death (from 6.6% to 6.1% in the study population, NNT = 200), low‐quality evidence.
52 per 1000 44 per 1000
 (16 to 122)
CVD at baseline (secondary prevention)
68 per 1000 61 per 1000
 (53 to 72)
Atrial fibrillation and arrhythmias
 Follow‐up: range 12 months to 60 months No CVD at baseline (primary prevention) RR 0.87
 (0.72 to 1.06) 11692
 (11 RCTs) ⊕⊝⊝⊝
 Very lowd The effect of increasing PUFA intake on atrial fibrillation is unclear as the evidence is of very low quality.
26 per 1000 34 per 1000
 (25 to 46)
CVD at baseline (secondary prevention)
119 per 1000 95 per 1000
 (80 to 114)
Adiposity ‐ body weight, kg
 Follow‐up: range 12 months to 60 months Mean body weight was 81.0 kg MD 0.76 higher
 (0.34 higher to 1.19 higher) ‐ 7100
 (13 RCTs) ⊕⊕⊕⊝
 Moderatee Higher PUFA intake probably has little or no effect on body weight.
Adiposity ‐ BMI, kg/m2
 follow‐up: range 12 months to 60 months Mean BMI was 26.9 kg/m2 MD 0.17 higher
 (0.08 lower to 0.42 higher) ‐ 4798
 (8 RCTs) ⊕⊕⊝⊝
 Lowf Higher PUFA intake may have little or no effect on BMI.
Serum total cholesterol (TC, mmol/L)
 Follow‐up: range 12 months to 96 months Mean serum TC was 5.46 mmol/L MD 0.12 lower
 (0.23 lower to 0.02 lower) ‐ 8072
 (27 RCTs) ⊕⊕⊕⊕
 Highg Higher PUFA intake has little or no effect on TC.
Serum fasting triglyceride (TG, mmol/L)
 Follow‐up: range 12 months to 72 months Mean serum TG was 1.57 mmol/L MD 0.12 lower
 (0.2 lower to 0.04 lower) ‐ 3905
 (20 RCTs) ⊕⊕⊕⊝
 Moderateh Higher PUFA intake probably reduces TG levels.
Serum high‐density lipoprotein (HDL, mmol/L)
 Follow‐up: range 12 months to 60 months Mean serum HDL 1.31 mmol/L MD 0.01 lower
 (0.02 lower to 0.01 higher) ‐ 4674
 (18 RCTs) ⊕⊕⊕⊝
 Moderatei Higher PUFA intake probably has little or no effect on HDL.
Serum low‐density lipoprotein (LDL, mmol/L)
 Follow‐up: range 12 months to 60 months Mean serum LDL 2.86 mmol/L MD 0.01 lower
 (0.09 lower to 0.06 higher) ‐ 3362
 (15 RCTs) ⊕⊕⊕⊝
 Moderatej Higher PUFA intake probably has little or no effect on LDL.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
 
 BMI: Body Mass Index; CI: confidence interval; CVD: cardiovascular disease; MD: mean difference; OR: odds ratio; PUFA: polyunsaturated fatty acids; RR: risk ratio
GRADE Working Group grades of evidence
 High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
 Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
 Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
 Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

aCardiovascular mortality

  1. Risk of bias: limiting trials to those at low summary risk of bias, low risk from allocation concealment, from attention bias, from compliance, by trial funding and trial size suggests small benefits and harms from increasing PUFA intake. Tends to confirm lack of important effect. Not downgraded.
  2. Inconsistency: I2 statistic less than 50%, not downgraded.
  3. Indirectness: most events occurred in men, and in trials carried out in high‐income nations. Not downgraded.
  4. Imprecision: 95% confidence intervals range from important benefit to important harm. Downgraded twice.
  5. Publication bias: some suggestion that one or two small trials may be missing. If added in they would tend to increase the RR. Not a large effect, not downgraded.

bCardiovascular events

  1. Risk of bias: sensitivity analyses suggested reduced risk of CVD events with more PUFA, lack of effect, and some harm. Downgraded once.
  2. Inconsistency: I2 statistic less than 50%, not downgraded.
  3. Indirectness: most events occurred in men, and in trials carried out in high‐income nations. Not downgraded.
  4. Imprecision: over 17,000 participants randomised, of whom more than 2400 experienced CVD events. 95% confidence intervals excluded important harms. Not downgraded.
  5. Publication bias: no suggestion of missing trials in the funnel plot. Not downgraded.

cCoronary heart disease mortality (CHD)

  1. Risk of bias: all sensitivity analyses concurred that increased PUFA reduced risk of CHD deaths. Not downgraded.
  2. Inconsistency: I2 statistic less than 50%, not downgraded.
  3. Indirectness: most events occurred in men, and in trials carried out in high‐income nations. Not downgraded.
  4. Imprecision: over 8800 participants randomised and over 500 CHD deaths. However, the 95% confidence intervals didn't exclude important harm. Downgraded once.
  5. Publication bias: some suggestion of publication bias. If present replacing missing trials would tend to raise the risk ratio towards 1.0 (no effect). Downgraded once.

dAtrial fibrillation and arrhythmias

  1. Risk of bias: no included trials were at low risk of compliance problems, all other sensitivity analyses suggested reduced risk of AF with increased PUFA. However there was no dose response, a suggestion of benefit in short trials, and harm in longer trials supported by meta‐regression. Downgraded once.
  2. Inconsistency: I2 statistic greater than 50%. Downgraded once.
  3. Indirectness: most events occurred in men, and in trials carried out in high‐income nations. Not downgraded.
  4. Imprecision: 95% confidence intervals exclude serious harm, but included the null. Downgraded once.
  5. Publication bias: no suggestion of missing trials in the funnel plot. Not downgraded.

eAdiposity ‐ body weight

  1. Risk of bias: sensitivity analyses assessing effects of different biases all suggested greater weight gain in those taking higher total PUFA. Not downgraded.
  2. Inconsistency: I2 statistic greater than 50% but partially explained by type of intervention and duration of intervention. Not downgraded.
  3. Indirectness: weight was assessed in both men and women, but all trials were conducted in high‐income countries. Not downgraded.
  4. Imprecision: 95% confidence intervals only included increased weight with increased PUFA intake. Not downgraded.
  5. Publication bias: the funnel plot suggests that some trials with less weight gain in the higher PUFA arm may be missing. Two trials with weight data could not be included in meta‐analysis, but they also suggested greater weight gain in the higher PUFA arm. Other missing trials, if due to publication bias, are likely to have not been published because they suggested increased weight in the higher PUFA arm, so are likely to support the main analysis. Downgraded once.

fAdiposity ‐ Body Mass Index (BMI)

  1. Risk of bias: sensitivity analyses assessing effects of different biases all suggested greater weight gain in those taking higher total PUFA. Not downgraded.
  2. Inconsistency: I2 statistic greater than 50%, and not explained by subgrouping. Downgraded once.
  3. Indirectness: weight was assessed in both men and women, but all trials were conducted in high‐income countries. Not downgraded.
  4. Imprecision: 95% confidence intervals did not include important benefits, but did include the null. Downgraded once.
  5. Publication bias: no suggestion of missing data. Not downgraded.

gSerum total cholesterol (TC)

  1. Risk of bias: sensitivity analyses all suggested greater lipid reduction with higher PUFA intake. Not downgraded.
  2. Inconsistency: I2 statistic greater than 50%, and while no single factor explains this there were greater TC reductions with low statin use, higher PUFA dose, lower baseline PUFA, and replacement of saturated fats and monounsaturated fats. Not downgraded.
  3. Indirectness: data provided by men and women, and comes from high‐income and low‐ to middle‐income countries. Not downgraded.
  4. Imprecision: data came from thousands of participants and 95% confidence intervals did not include harm. Not downgraded.
  5. Publication bias: funnel plot not interpretable, known missing data are consistent with data used in meta‐analysis. Not downgraded.

hSerum triglycerides (TG)

  1. Risk of bias: sensitivity analyses all suggested greater lipid reduction with higher PUFA intake. Not downgraded.
  2. Inconsistency: I2 statistic = 50%, without any clear explanation from subgrouping. Downgraded once.
  3. Indirectness: data provided by men and women, and comes from high‐income and industrialising countries. Not downgraded.
  4. Imprecision: data came from thousands of participants and 95% confidence intervals did not include harm. Not downgraded.
  5. Publication bias: no suggestion of missing data. Not downgraded.

iSerum HDL

  1. Risk of bias: consistent lack of effect of PUFA in all sensitivity analyses. Not downgraded.
  2. Inconsistency: I2 statistic less than 50%. Not downgraded.
  3. Indirectness: data provided by men and women, and comes from high‐income and industrialising countries. Not downgraded.
  4. Imprecision: data came from thousands of participants and confidence interval excludes important effects. Not downgraded.
  5. Publication bias: some trials with lower HDL appear to be missing. Downgraded once.

jSerum LDL

  1. Risk of bias: consistent lack of effect of PUFA in all sensitivity analyses. Not downgraded.
  2. Inconsistency: I2 statistic less than 50%. Not downgraded.
  3. Indirectness: data provided by men and women, and comes from high‐income and industrialising countries. Not downgraded.
  4. Imprecision: data came from thousands of participants and confidence interval excludes important effects. Not downgraded.
  5. Publication bias: some trials with lower LDL appear to be missing. Downgraded once.