Table 5. Sensitivity analysis of mortality by risk of bias and industry sponsorship.

Risk of bias	No. of trials	Total sample size	Risk ratio (95% CI)	Heterogeneity
				Q	df	I2, %	P value for I2
Low risk in sequence generation	3	331	0.98 (0.54–1.77)	4.89	2	59.1	0.087
Low risk in allocation concealment	3	331	0.98 (0.54–1.77)	4.89	2	59.1	0.087
Low risk in blinding of outcome assessors	2	296	0.95 (0.41–2.19)	4.88	1	79.5	0.027
Low risk in incomplete outcome reporting	4	390	1.05 (0.71–1.53)	5.10	3	41.1	0.165
Presence of industry sponsorship	2	296	0.95 (0.41–2.19)	4.88	1	79.5	0.027