Table 3.
Outcomes | Anticipated absolute effects* (95% CI) | Relative effect (95% CI) | No of participants (studies) | Certainty of evidence (GRADE)† | Comments | |
---|---|---|---|---|---|---|
Risk with low PUFA | Risk with high PUFA | |||||
Diagnosis of type 2 diabetes mellitus | 37 per 1000 | 40 per 1000 (30 to 53) | RR 1.08 (0.81 to 1.43) | 4481 (3 RCTs) | ⨁◯◯◯ VERY LOWa,b | Effect of increasing total PUFA on risk of diabetes diagnosis is unclear as evidence was of very low quality. Downgraded once for risk of bias and twice for imprecision |
Diagnosis of impaired glucose tolerance | 0 per 1000 | 0 per 1000 (0 to 0) | Not estimable | (0 RCTs) | - | No RCTs assessed effect of total PUFA on diagnosis of impaired glucose tolerance |
Glycated haemoglobin (HbA1c, %) | Mean HbA1c 8.6% | Mean HbA1c, 0.08% higher (0.41% lower to 0.56% higher) | - | 172 (3 RCTs) | ⨁⨁◯◯ LOWc,d | Increasing total PUFA may make little or no difference to glycated haemoglobin. Downgraded once each for imprecision and risk of bias |
Plasma glucose, fasting (mmol/L) | Mean plasma glucose 8.1 mmol/L | Mean plasma glucose 0.04 mmol/L lower (0.18 lower to 0.11 higher) | - | 182 (3 RCTs) | ⨁⨁◯◯ LOWe,f | Increasing total PUFA may make little or no difference to plasma glucose. Downgraded once each for imprecision and risk of bias |
Insulin, fasting (pmol/L) | Mean insulin 61.7 pmol/L | Mean insulin 0.6 pmol/L lower (10.33 lower to 9.14 higher) | - | 157 (3 RCTs) | ⨁⨁◯◯ LOWc,d | Increasing total PUFA may make little or no difference to fasting insulin. Downgraded once each for imprecision and risk of bias |
HOMA-IR | Mean HOMA-IR 1.8 | Mean HOMA-IR 0.34 lower (0.88 lower to 0.2 higher) | - | 93 (1 RCT) | ⨁◯◯◯ VERY LOWg,h,i | Effect of increasing total PUFA on HOMA-IR is unclear, as evidence is of very low quality. Downgraded once each for imprecision, indirectness and risk of bias |
HOMA-IR=homoeostatic model assessment for insulin resistance; RCT=randomised controlled trial; RR=risk ratio.
Patient or population: people with and without diabetes; setting: these are long term trials, so participants live in the community; intervention: high total PUFA; comparison: low total PUFA.
Risk in intervention group (and its 95% CI) is based on assumed risk in comparison group and relative effect of intervention (and its 95% CI). Note that GRADE describes risk and 95% CI without using negative numbers; for example, GRADE states “0.02% lower (0.07 lower to 0.04 higher),” which would normally be described as “–0.02% (−0.07% to 0.04%).”
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.
a. Risk of bias: no studies were at low summary risk of bias or at low risk from compliance problems. Downgraded once.
b. Imprecision: 4481 participants in 3 RCTs, 175 diagnoses; 95% CI did not exclude important benefits or harms. Downgraded twice.
c. Risk of bias: lack of effect consistent in fixed effects analysis, and sensitivity analyses on concentration, but no included trials were at low summary risk of bias. Downgraded once.
d. Imprecision: 95% CI does not exclude important harms or benefits. Downgraded once.
e. Risk of bias: effect did not alter with fixed effects analysis; the single study at low risk from compliance suggested a small reduction in glucose with increased PUFA, but no trials were at low summary risk of bias or had low risk from allocation concealment. Downgraded once.
f. Imprecision: data based on 182 participants in 3 trials; 95% CI does not exclude important benefits or harms. Downgraded once.
g. Imprecision: data reflect a single study in 93 participants (a small proportion of the participants in the whole study); 95% CI does not exclude important harm. Downgraded once.
h. Risk of bias: effect not altered in fixed effects analysis, but the single study was not at low risk from summary risk of bias or compliance. Downgraded once.
i. Indirectness: subgroup of a single trial reported. Downgraded once.