Table 2.

HRs (95% CIs) for the association between baseline age-specific AMH tertiles and risk of type 2 diabetes in women of the Doetinchem Cohort Study

Population	Lowest age-specific AMH tertile	Middle age-specific AMH tertile	Highest age-specific AMH tertile (reference)	p value for trend
Total study population (n = 3293, 163 cases)	1.62 (1.06, 2.48)*	1.24 (0.81, 1.92)	1.00	0.02
Exclusion of AMH measurements within 2 years prior to type 2 diabetes diagnosis (n = 3285, 155 cases)	1.55 (1.00, 2.40)*	1.19 (0.76, 1.85)	1.00
Exclusion of women with a missing AMH measurement at baseline (n = 3104, 148 cases)	1.62 (1.04, 2.52)*	1.29 (0.83, 2.00)	1.00
Exclusion of women who ever used HRT (n = 1803, 95 cases)a	1.26 (0.72, 2.20)	0.74 (0.41, 1.32)	1.00
Exclusion of women who potentially had PCOS (n = 3025, 138 cases)	1.57 (0.97, 2.54)	1.13 (0.71, 1.83)	1.00

Cox proportional hazards models were adjusted for the following baseline variables: age (years), parity (nulliparous, parous), current oral contraceptive use (yes, no), menopausal status (premenopausal, postmenopausal), BMI (kg/m²), educational attainment (low, middle, high), current smoking (yes, no), alcohol consumption (glasses/day), physical activity (inactive, active), hypertension (yes, no), total cholesterol (mmol/l)

^aNumbers differed between imputation sets, as the variable ever HRT use itself was imputed; presented numbers are average sample sizes and average numbers of cases

*p <0.05