Table 2

Summary of the quality assessment (risk of bias)

Bias type	Quality criteria	Miyazaki et al30	Chang et al3135	Bennell et al29	Woollard et al32	Henriksen et al33
Selection	Were the descriptions of the groups and the distribution of prognostic factors sufficient?	A	I	I	A	A
	Were the groups assembled at a similar point in their disease progression?	A	I	I	A	A
	Were the groups comparable on all important confounding factors?	I	I	I	U	A
Detection	Was the joint load estimate reliably ascertained?	U	U	U	U	I
	Was adequate adjustment made for the effects of these confounding variables?	A	I	A	I	I
	Was outcome assessment blind to exposure status?	A	A	I	I	A
	Was follow-up long enough for the outcomes to occur?	A	A	A	A	A
	What proportion of the cohort was followed-up?*	U	A	U	A	A
Attrition	Were dropout rates and reasons for drop-out similar across groups?	I	I	I	A	A
Overall risk of bias†		Very serious limitations; high risk of bias	Very serious limitations; high risk of bias	Very serious limitations; high risk of bias	Very serious limitations; high risk of bias	Very serious limitations; high risk of bias
Methodological quality		Low	Low	Low	Low	Low

*Adequate=follow-up proportion >80%; unclear 50–80%; inadequate=<50%.
†Risk of bias within studies is assessed using GRADE's approach to study limitations: No serious limitation defined as all criteria being adequately described (high methodological quality); serious limitations defined as one criterion being inadequately described or >1 criterion being unclearly described (moderate methodological quality); very serious limitation defined as >1 criterion being inadequately described (low methodological quality).
A, adequately described; I, inadequately described; U, unclear.