Table 1. Risk of bias summary showing judgments about each risk of bias domain in ROBINS-I and overall bias assessment across all domains

	Domain 1: Bias owing to confounding	Domain 2: Bias in selection of participants into the study	Domain 3: Bias in classification of interventions	Domain 4 ITT: Bias owing to deviations from intended interventions: effects of assignment to intervention
First author, publication year	LYL	JB	KB	LYL	JB	KB	LYL	JB	KB	LYL	JB	KB
Margolis 2005²⁷		Low	Low		Low	Low		Low	Low		NI	NI
Margolis 2012²⁸		Low	Low		Moderate	Moderate		Low	Low		NI	NI
Basak 2013²⁹	Critical	Critical		Moderate	Moderate		Low	Low		NI	NI
Adams 1985³¹	Critical	Critical		NI	NI		Low	Low		NI	NI

Domain 4: Bias owing to deviations from intended interventions: effect of starting and adhering to intervention

Domain 5: Bias owing to missing data

Domain 6: Bias in measurement of outcomes

Domain 7: Bias in selection of the reported result

Overall bias assessment across all domains

LYL

Low

Moderate

Serious

Low

Moderate

Low

Serious

Moderate

Serious

Authors: LYL, JB, KB. ITT = intention-to-treat. NI = no information. ROBINS-I = Risk of Bias in Non-randomised Studies - of Interventions.