Table 3.

Statistical comparison of the profiles of limitations between studies with positive and negative results

	Risk of bias				Imprecision
	Total number of key questions judged to be at high risk of bias	Number of key questions judged to be at high risk of bias in favour of a null result	Number of key questions judged to be at high risk of bias with uncertain direction on study outcome	Studies with key question judged to be at high risk of bias in favour of an effect of exposure	Studies judged to have concern regarding statistical power	Studies judged to have concern regarding missing adjustment for multiple comparisons when relevant
	Median (Interquartile range)	Median (Interquartile range)	Median (Interquartile range)	n (%)	n (%)	n (%)
Studies with positive results	3.0 (4.0)	2.0 (4.0)	1.0 (2.0)	2 (29%)	6 (86%)	2 (29%)
Studies with negative results	2.0 (2.0)	2.0 (1.5)	0 (1.5)	5 (26%)	15 (71%)	10 (48%)
Test statistics and p-values	Mann-Whitney-U: U = 55.5 p = 0.351,4	Mann-Whitney-U: U = 65.0 p = 0.682,4	Mann-Whitney-U: U = 54.5 p = 0.322,4	Fisher’s exact: p = 1.002	Fisher’s exact: p = 0.643	Fisher’s exact: p = 0.663

Significance levels: ¹α = 0.05, ²α = 0.017, ³α = 0.025

⁴Exact significance