Table 3.

Subgroup Analysis of VA/SCD Risk by Individual Sulfonylurea

Sulfonylurea versus metformin	HR (95% CI)	Cohort size	Number of events	Follow‐up person‐years/incidence per 1000 person‐years	E‐value (HR)
Glicazide	1.74 (1.57–1.93)	10 116	744	49 983/14.89 (13.83–15.99)	2.87
Glipizide	2.15 (1.54–3.01)	399	36	1998/18.02 (12.62–24.94)	3.72
Tolbutamide	4.70 (3.58–6.17)	302	56	1503/37.26 (28.14–48.38)	8.87
Glibenclamide	1.16 (0.89–1.51)	1216	60	6157/9.75 (7.44–12.54)	1.00*
Glimepiride	2.79 (1.82–4.29)	180	22	829/26.54 (16.63–40.18)	5.02

A large E‐value implies that any unmeasured confounder must be strong to explain away the effect of sulfonylurea over metformin use in the risk of developing VA/SCD. HR indicates hazard ratio; and VA/SCD, ventricular arrhythmia or sudden cardiac death.

^*The CI of the hazard ratio crossed 1; thus, its E‐value is 1.