Table 4.5.

Risk of Bias ‐ Distribution of the Studies Not Analysing STAR Data

Risk of bias item	Judgement								Total number of studies
	High	Low	Unclear	1	2	3	4	5
Sequence generation	80	0	2	‐	‐	‐	‐	‐	82
Allocation concealment	80	0	2	‐	‐	‐	‐	‐	82
Blinding *	‐	‐	0	0	0	0	82	0	82
Incomplete data[Link], 1:	‐	‐	2	2	1	1	0	0	6
Selective reporting[Link], 2:	‐	‐	0	5	0	1	0	3	9
Other bias[Link], 3:	‐	‐	0	4	1	1	0	4	10
Confounding[Link], 4:	‐	‐	0	2	0	1	2	51	56

Notes: *: The judgement is based on a 5‐point scale where 1 indicates low risk of bias and 5 indicates high risk of bias. Studies scoring 5 on any item of the risk of bias tool were not included in the data synthesis and therefore, it was not relevant to judge on the remaining items for these studies.

^1:Not judged for the eighteen studies that did not provide enough data to calculate an effect estimate, for the 51 studies scoring 5 on the confounding item, the four studies scoring 5 on the other bias item and the three studies scoring 5 on the selective reporting item.

^2:Not judged for the eighteen studies that did not provide enough data to calculate an effect estimate, for the 51 studies scoring 5 on the confounding item and the four studies scoring 5 on the other bias item..

^3:Not judged for the eighteen studies that did not provide enough data to calculate an effect estimate, for the 51 studies scoring 5 on the confounding item and the three studies scoring 5 on the selective reporting item.

^4: Not judged for the eighteen studies that did not provide enough data to calculate an effect estimate, for the one study using a randomised design and neither for the seven studies that scored 5 on the selective reporting and other bias items.