Supplementary Table 2.

Risk of Bias using ROBINS-I tool

Study	Confounding	Selection	Intervention Measurement	Missing Data	Outcome Measurement	Reported Results	Overall
Hu et al	Low	Low	Moderate	Low	Low	Low	Low
Encinas-Ullán et al	Moderate	Low	Moderate	Low	Low	Low	Low
Choi et al	Moderate	Moderate	Moderate	Low	Low	Low	Moderate
Boraiah et al	Moderate	Moderate	Moderate	Moderate	Low	Low	Moderate
Sirkin et al	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Bone et al	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Molina et al	Moderate	Low	Moderate	Moderate	Moderate	Moderate	Moderate
White et al	Moderate	Moderate	Moderate	Moderate	Low	Moderate	Moderate
Gardner et al	Moderate	Moderate	Moderate	Low	Moderate	Low	Moderate
Gehr et al	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Danoff et al	Low	Moderate	Moderate	Low	Low	Low	Low
Silluzio et al	Moderate	Moderate	Moderate	Moderate	Moderate	Low	Moderate
Zeng et al	Moderate	Moderate	Moderate	Low	Moderate	Low	Moderate
Conroy et al	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Harris et al	Moderate	Moderate	Moderate	Moderate	Low	Moderate	Moderate
Kapukaya et al	Moderate	Moderate	Low	Low	Moderate	Low	Low
Yildiz et al	Moderate	Moderate	Moderate	Low	Moderate	Low	Moderate

Note: Moderate = the study is sound for a non-randomised study with regard to this domain but cannot be considered comparable to a well-performed randomised trial; Low = The study is comparable to a well-performed randomised trial with regard to this domain