Table 2.

Linear mixed effects models on outcomes

Oxygen saturation	Test	Model 1			Model 2
		Est	95% CI	P value (p-BH)	Est	95% CI	P value (p-BH)
Blood	T1D versus CON	−0.5	−0.9 to −0.0	0.034 (0.041)	−0.3	−0.9 to 0.4	0.424 (0.424)
	Standing versus supine	0.6	0.4 to 0.8	<0.001 (<0.001)	0.6	0.4 to 0.8	<0.001 (<0.001)
	Standing versus supine, T1D versus CON	−0.4	−0.6 to −0.2	<0.001 (<0.001)	−0.4	−0.6 to −0.1	0.001 (0.002)
Arm	T1D versus CON	–	–	–	–	–	–
	Standing versus supine	–	–	–	–	–	–
	Standing versus supine, T1D versus CON	–	–	–	–	–	–
Forehead	T1D versus CON	−0.7	−1.6 to 0.2	0.149 (0.224)	−1.5	−3.0 to 0.1	0.058 (0.116)
	Standing versus supine	−1.0	−1.6 to 0.7	<0.001 (<0.001)	−1.0	−1.2 to 0.7	<0.001 (<0.001)
	Standing versus supine, T1D versus CON	0.1	−0.2 to 0.3	0.683 (0.683)	0.1	−0.2 to 0.6	0.584 (0.683)

Results are presented as estimated mean differences with 95% CIs. Estimates are on original scale. Differences in saturation were modelled by a linear mixed-effect model with a participant-specific random intercept to account for the correlation of repeated measures within participants. Diabetes and whether the participants were standing or not, sex, age and smoking were included as fixed effects (model 1). A second level of confounder adjustment was also tested, by further adjusting for Body Mass Index, heart rate, systolic blood pressure, hemoglobin A1c, hemoglobin and baroreflex sensitivity (model 2). Estimates from arm oxygen saturation levels are not reported as normality of model residuals could not be obtained.

CON, healthy control; Est, estimate; p-BH, p value corrected for multiple testing using the Benjamini-Hochberg procedure; T1D, type 1 diabetes.