Effectiveness of royal jelly supplementation in glycemic regulation: A systematic review

Table 4 GRADE assessment for long-term effectiveness of RJ treatment on glycemic control

GRADE criteria	Rating	Support for judgement	Overall quality of evidence
Outcome: Long term glycemic control (n = 14 studies)
RoB (assessed on Cochrane RoB Collaboration Tool)	No	Only one study had low RoB for all categories. Most studies had at least one item at high or unclear RoB	High
	Serious (-1)1
	Very serious (-2)
Inconsistency	No	Generally, positive clinical effects demonstrated but some studies indicate null effects. There are also large variations in magnitude of effect. Heterogeneity is notable between the studies (in population, intervention and outcome assessment)	Moderate
	Serious (-1) 1
	Very serious (-2)
Indirectness	No1	Evidence synthesized from studies addresses review question with respect to population, interventions and outcome	Low1
	Serious (-1)
	Very serious (-2)
Imprecision	No	All studies have groups with small sample sizes (≤ 20), with no indication that they meet required sample sizes to detect difference in outcome; observable but statistically insignificant measures in many studies suggest sample sizes were too small to detect difference. 95% confidence intervals of effect size estimates mostly suggest an appreciable benefit for treatment, but there are several that suggest possibility of no meaningful effect	Very low
	Serious (-1)
	Very serious (-2) 1
Publication bias	Undetected1	There is chance of publication bias considering the review is entirely “small-scale” trials; this area of research is not well-established and there is potential for publication bias, but none was overtly detected
	Strongly suspected (-1)
Other	Large effect (+11 or +2)	Standardized mean difference of studies (effect size) indicates large magnitude of effect of treatment. Dose-response relationships observed
	Dose response (+11 or +2)
	No plausible confounding (+1 or +2)

¹Indicates decision. RoB: Risk of Bias. Adapted from Ryan and Hill[21] and Wei et al[22].