Table 4.

Quality of included studies assessed using self-designed tool

Quality criteria		Reference																Author’s recommendations
		[19]	[20]	[21]	[22]	[23]	[1]	[24]	[25]	[26]	[27]	[28]	[29]	[30]	[31]	[12]	[32]
Selection and completeness of follow-up	Patient baseline characteristics with no/minor significant differences between FLS and no-FLS group	No	Yes	Yes	Yes	No	Yes	Yes	NR	Yes	Yes	Yes	Yes	No	No	Yes	Yes	Participants in two groups should be carefully selected with no/minor significant differences in characteristics to avoid selection bias
	All patients were included and analyzed in both FLS and no-FLS cohorts	Yes	Yes	No	Yes	Yes	No	No	Yes	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	All patients should be included and analyzed regardless of whether they were seen in the FLS clinic
	Inclusion/exclusion criteria are clearly described for FLS and no-FLS group	Yes	Part	Part	Part	Part	Part	Yes	Part	Yes	Part	Yes	Yes	Yes	Yes	Yes	Yes	Inclusion/exclusion criteria should be clearly described for completeness of reporting reason
	At least 50% eligible patients attend FLS	Yes	NR	NR	NR	NR	Yes	NR	NR	NR	NR	Yes	NR	Yes	No	Yes	NR	The proportion of FLS attending is expected to be at least 50% to provide confidence of the results
	Loss to follow-up ≤20% in FLS and no-FLS group	Yes	Yes	Part	NR	NR	NR	NR	NR	Part	NR	Yes	NR	NR	NR	Yes	NR	The loss of follow-up for both groups is expected to be less than 20% to guarantee statistical power for the results
Exposure	Clear description of care for FLS and no-FLS group	Yes	Part	Part	Yes	No	Part	No	Part	Part	Part	Yes	Part	Part	Part	Part	Part	Fracture care including BMD testing, treatment, education, long-term adherence, etc. should be clearly described for both groups
Outcome	Outcomes assessed in FLS and no-FLS groups using similar method	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	The same statistical methods should be used in both groups to assess the outcomes
Statistical accuracy and analyses	Analyses of outcomes accounted for relevant confounders	Yes	No	No	Yes	Yes	No	Yes	No	Yes	No	Yes	No	Yes	Yes	Yes	Yes	Relevant confounders should be fully adjusted using statistical models, such as multivariable cox regression model
	Sample size is based on power calculation	No	No	No	Yes	No	No	No	No	Yes	No	No	No	No	No	Yes	No	To avoid insufficient statistical power for the results, sample size should be based on power calculation
	Analyses of outcomes account for competing risk of death	No	No	No	No	Yes	No	Yes	No	No	No	No	No	No	Yes	No	Yes	Competing risk analysis should be included in studies designed to evaluate risk of subsequent fracture
Total score		7	5	3.5	6.5	4.5	4	5	3	6	4	8	4.5	5.5	5.5	8.5	6.5

Yes, fully fulfilled the criteria; No, not fulfilled the criteria; Part, partially fulfilled the criteria

NR not reported, BMD bone mineral density, FLS fracture liaison service