Table 3.

Subgroup analysis. Hazard ratios (95% confidence intervals) for incidence of depression according to baseline regular and decaffeinated coffee consumption.

Cups/Day	Regular Coffee Consumption				p for Trend
	<1	1	>1 and <4	≥4
Cups/day (median)	0	1	2.5	5
N	6315	3433	4193	472
Cases	84	49	61	5
Person-years	61,621	34,065	43,130	5212
Crude HR	1 (ref.)	1.01 (0.71–1.44)	0.97 (0.69–1.35)	0.65 (0.26–1.60)	0.569
Model 1	1 (ref.)	1.00 (0.70–1.42)	0.96 (0.69–1.34)	0.64 (0.26–1.59)	0.533
Model 2	1 (ref.)	0.96 (0.67–1.37)	0.84 (0.59–1.18)	0.43 (0.17–1.07)	0.095
Additionally adjusted for decaffeinated coffee consumption	1 (ref.)	0.97 (0.68–1.39)	0.87 (0.61–1.23)	0.44 (0.18–1.11)	0.141
Cups/day	Decaffeinated Coffee Consumption				p for Trend
	<1		1	>1
Cups/day (median)	0		1	2.5
N	12,700		1268	445
Cases	167		21	11
Person-years	127,007		12,674	4348
Crude HR	1 (ref.)		1.25 (0.79–1.96)	1.90 (1.03–3.51)	0.033
Model 1	1 (ref.)		1.20 (0.76–1.89)	1.77 (0.96–3.26)	0.065
Model 2	1 (ref.)		1.20 (0.76–1.89)	1.54 (0.82–2.87)	0.142
Additionally adjusted for regular coffee consumption	1 (ref.)		1.15 (0.72–1.82)	1.46 (0.78–2.76)	0.218

Results from Cox regression models. Age was the underlying time variable in all analyses. Model 1: adjusted for sex and stratified for age (decades) and recruitment period. Model 2: adjusted for age, sex, alcohol intake (linear and quadratic term), years of university education, marital status, smoking, body mass index, total energy intake, adherence to the Mediterranean diet, between-meal snacking and following special diets, leisure-time physical activity (METS-h/week), hours of TV watching, hypertension at baseline, baseline high blood cholesterol, self-perception of competitiveness, anxiety, and psychological dependence, and use of anxiolytics.