Table 2

Subgroup analysis of the risk of incident dementia among ADM treatment comparisons (MET monotherapy served as the reference)

	Age			BMI			HbA1c
	≤75 years	>75 years	P value*	18.5–25 kg/m²	≥25 kg/m²	P value*	≤7%	>7%	P value*
	n=466 855	n=92 251		n=37 476	n=521 082		n=175 806	n=383 300
Primary analysis: 1-year exposure†
SU	1.12 (1.08 to 1.16)	1.11 (1.07 to 1.16)	0.14	1.07 (0.99 to 1.15)	1.14 (1.09 to 1.18)	0.06	1.14 (1.10 to 1.17)	1.09 (1.03 to 1.15)	0.04
TZD	0.67 (0.64 to 0.71)	0.91 (0.86 to 0.96)	0.01	0.98 (0.88 to 1.08)	0.72 (0.68 to 0.77)	0.01	0.79 (0.76 to 0.82)	0.71 (0.65 to 0.78)	0.02
MET and SU	1.15 (1.11 to 1.19)	1.16 (1.11 to 1.22)	0.26	0.98 (0.90 to 1.06)	1.17 (1.12 to 1.23)	0.01	1.15 (1.11 to 1.19)	1.14 (1.08 to 1.20)	0.53
MET and TZD	0.86 (0.82 to 0.9)	0.91 (0.86 to 0.96)	0.01	0.90 (0.81 to 1.00)	0.89 (0.84 to 0.94)	0.69	0.93 (0.89 to 0.97)	0.82 (0.76 to 0.90)	0.01
SU and TZD	1.12 (1.07 to 1.17)	1.05 (0.99 to 1.11)	0.01	1.06 (0.96 to 1.17)	0.90 (0.85 to 0.96)	0.19	1.00 (0.96 to 1.04)	1.13 (1.05 to 1.22)	0.01
Exploratory analysis: 2-year exposure†
SU	1.13 (1.08 to 1.18)	1.12 (1.06 to 1.17)	0.53	1.02 (0.93 to 1.12)	1.15 (1.10 to 1.21)	0.01	1.11 (1.11 to 1.19)	1.13 (1.06 to 1.21)	0.19
TZD	0.69 (0.65 to 0.73)	0.72 (0.67 to 0.78)	0.02	0.74 (0.65 to 0.84)	0.63 (0.59 to 0.68)	0.32	0.67 (0.64 to 0.71)	0.60 (0.53 to 0.67)	0.04
MET and SU	0.98 (0.94 to 1.03)	1.07 (1.01 to 1.14)	0.01	0.79 (0.71 to 0.88)	0.9 (0.86 to 0.96)	0.01	0.93 (0.89 to 0.97)	0.88 (0.83 to 0.94)	0.04
MET and TZD	0.69 (0.65 to 0.74)	0.81 (0.75 to 0.88)	0.01	0.84 (0.73 to 0.96)	0.74 (0.69 to 0.80)	0.01	0.67 (0.64 to 0.71)	0.75 (0.67 to 0.83)	0.02
SU and TZD	1.03 (0.97 to 1.08)	0.93 (0.87 to 1.01)	0.01	1.02 (0.91 to 1.15)	0.88 (0.82 to 0.94)	0.01	0.84 (0.80 to 0.88)	1.04 (0.94 to 1.13)	0.01

Boldfaced values representing estimated HRs were statistically different from 0 with p value of <0.05.
The interactions between treatment and baseline age, BMI, and HbA1c were all statistically significant with a p value of <0.001.
*P value for testing the interaction effects between treatment and baseline age, BMI, and HbA1c categories.
†Models were adjusted for age, calendar year of individual baseline, statin use, HbA1c, systolic blood pressure, diastolic blood pressure, and congestive heart failure history collected at baseline because these variables remained unbalanced after matching, in at least one comparison.
ADM, antidiabetic medication; BMI, body mass index; HbA1c, glycated hemoglobin; MET, metformin; SU, sulfonylurea; TZD, thiazolidinedione.