Table 3.

Sensitivity analyses: HR and 95% CI (Q4 vs. Q1) for incident type 2 diabetes according to baseline predicted serum vitamin D

	N	Incident T2D	HR (95% CI)
Main analysis Q4 vs. Q1a	18,594	209	0.48 (0.26–0.88)
Excluding participants with cancer at baselineb	18,103	198	0.48 (0.26–0.89)
Excluding participants with hypertension at baselinec	16,631	113	0.28 (0.11–0.72)
Excluding participants with hypertriglyceridemia at baselined	17,393	136	0.44 (0.20–0.99)
Excluding participants with hypercholesterolemia at baselinee	15,446	114	0.50 (0.22–1.17)
Excluding extreme daily energy intake (< p1 or > p99)	20,108	217	0.50 (0.27–0.90)
Excluding outliersf	17,823	192	0.45 (0.24–0.86)
Including only participants with family history of T2D	2854	80	0.34 (0.10–1.10)
Including only overweight participantsg	5538	173	0.19 (0.07–0.56)
Including only participants with sedentary lifestyleh	9297	135	0.28 (0.09–0.93)

^aAdjusted for sex, age, marital status, smoking status (current, former or never smoker), smoking pack-years, weight change, years of university, TV hours/day, family history of T2D, physical activity (MET-h/week), Trichopoulou’s 9-point score/Mediterranean dietary pattern, energy intake (kcal/day), sugar-sweetened beverage consumption (servings/day), snacking, following a special diet, prevalent hypertension, prevalent cancer, prevalent hypercholesterolemia, prevalent hypertriglyceridemia, obesity (kg/m²)

^bWithout prevalent cancer adjustment

^cWithout prevalent hypertension adjustment

^dWithout prevalent hypertriglyceridemia adjustment

^eWithout prevalent hypercholesterolemia adjustment

^fWithout outliers located within ± 1.5 interquartile range of the average of predicted vitamin D

^gParticipants with BMI ≥ 25 kg/m² were included

^hParticipants with sedentary lifestyle (MET-h/week < p50)