Table 3.
Outcome | No. of Studies | Study Design | Certainty Assessment | RR [95% CI] | Certainty | ||||
---|---|---|---|---|---|---|---|---|---|
Risk of Bias | Inconsistency | Indirectness | Imprecision | Other Considerations | |||||
T2D risk reduction | Seven | randomized trials | not serious | not serious a | not serious | not serious | none | 0.53 [0.41, 0.67] | ⨁⨁⨁⨁ HIGH |
CI = confidence interval; GRADE = grading of recommendations assessment, development, and evaluation; RR = risk ratio; T2D = type 2 diabetes. a Although there was significant heterogeneity (I2 = 65%, p = 0.01), the removal of one study [34] explained some of the heterogeneity, which changed it from significant to non-significant (I2 = 36%, p = 0.16). However, the estimate of effect did not change appreciably. Furthermore, this inconsistency was not considered serious as the magnitude of effect remained large and in the same direction across all the studies (RR < 0.72).