Table 2.

GRADE assessment of individual food source of fructose-containing sugars

Certainty assessment								Study event rates (%)	Effect	Certainty
No. of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Other considerations		Relative risk (95% CI)
Sugar-sweetened beverages intake on incident gout (follow-up median 17 years)
2	Observational studies	No serious risk of bias	No serious inconsistency	Serious indirectness*	No serious imprecision	Undetected†	Large magnitude of effect.‡ Dose–response.§	1533/125 299 (1.22)	2.08 (1.40 to 3.07)	⊕⊕⊕◯ Moderate*†‡§ Due to downgrade for indirectness and upgrade for large magnitude effect and dose–response.
Fruit juice intake on incident gout (follow-up median 17 years)
2	Observational studies	No serious risk of bias	No serious inconsistency	Serious indirectness*	No serious imprecision	Undetected†	None	1533/125 299 (1.22)	1.73 (1.17 to 2.57)	⊕◯◯◯ Very low*† Due to downgrade for indirectness.
Fruit intake on incident gout (follow-up median 9.87 years)
2	Observational studies	No serious risk of bias	Very serious inconsistency¶	Serious indirectness*	Serious imprecision**	Undetected†	None	983/75 383 (1.3)	0.89 (0.27 to 2.87)	⊕◯◯◯ Very low*†¶** Due to downgrade for inconsistency, indirectness and imprecision.

*Downgrade for indirectness as the study population is specific to a group of the population like professionals, nurses or runners.

†No downgrade for publication bias as publication bias could not be assessed due to lack of power for assessing funnel plot asymmetry and small study effect (<10 cohort included in our meta-analysis).

‡Upgrade for a large magnitude of effect (RR >2.0).

§Upgrade for a dose response gradient as the GLST dose–response analysis revealed a significant linear relationship between sugar-sweetened beverage intake and incident gout (p=0.0001).

¶Downgrade for very serious inconsistency, as the two studies included had opposite associations, and there was evidence of substantial interstudy heterogeneity (I²=94%, p<0.0001). Due to the small number of studies included in the analysis, subgroup analysis was not performed.

**Downgrade for serious imprecision, as the lower bound of the 95% CI (RR: 0.27) includes clinically important benefit (RR: <0.9), while the upper bound of the 95% CI (RR: 2.87) crosses the minimally important difference of 10% (RR: >1.1).

GLST, generalised least squares trend estimation models; RR, risk ratio.