Table 2.

Association between log plasma levels of caffeine and its metabolites and the estimated 60 ml espresso cups daily consumed in SKIPOGH (considering all caffeine source)

		Caffeine + paraxanthine + theophylline plasma levels a
		Model 1		Model 2
	Predictors	Estimates (95% Confidence Interval)	p-value	Estimates (95% Confidence Interval)	p-value
SKIPOGH 1 (N = 428)	Intercept	6.92 (6.51; 7.32)	< 10 − 3	7.18 (6.64; 7.71)	< 10 − 3
	60 ml espresso b	0.03 (0.01; 0.05)	0.007	0.00 (−0.03; 0.04)	0.89
	Change of slope			−0.62 (−1.08; −0.16)	0.008
	Change of slope ˣ 60 ml espresso b			0.16 (0.08; 0.25)	< 10 − 3
SKIPOGH 2 (N = 610)	Intercept	6.48 (6.14; 6.82)	< 10 − 3	6.53 (6.06; 7.00)	< 10 − 3
	60 ml espresso b	0.03 (0.01; 0.04)	< 10 − 3	0.02 (−0.01; 0.05)	0.14
	Change of slope			−0.19 (−0.58; −0.21)	0.36
	Change of slope ˣ 60 ml espresso b			0.06 (0.00; 0.12)	0.04

Model 1 was not adjusted for the change of slope, while Model 2 was adjusted for the change of slope at 5 and 6 cups of 60 ml espresso per day in SKIPOGH1 and SKIPOGH2, respectively. All models were adjusted for age, smoking and time spent between last caffeine intake and blood intake

SKIPOGH 1 and 2 had 3% increase in log plasma levels after each 60 ml espresso cup, ignoring the change of the slope (Model 1). Plasma SKIPOGH1 and 2 levels were 1043 (exp(6.92 + 0.03)) and 672 (exp(6.48 + 0.03)) ng/ml for one 60 ml espresso per day

Considering the change of the slope (Model 2), log plasma levels increased by 16% for 5 60 ml espresso cups in SKIPOGH1 and 6% for 6 in SKIPOGH2. Thus, SKIPOGH1 plasma levels were 1541 (exp(7.18 + 0.16)) after 5 60 ml espresso cups, whereas SKIPOGH2 plasma levels were 728 (exp(6.53 + 0.06)) after 6 cups. In both waves, more espresso intake did not raise log plasma caffeine and its metabolites

^aLog transformed

^bThe estimated number of 60 ml espresso cups consumed per day

ˣInteraction term