Vascular and Myocardial Structure and Function in Adolescents with Type 1 Diabetes: The CARDEA Study

Table 3

Multivariable linear regression estimates (95% CI) on the relation between having type 1 diabetes versus not on arterial pressure, blood lipids, and early cardiovascular markers.

	Case complete analysis		Multiple imputation analysis
	β coefficient for type 1 diabetes (95% CI)	β coefficient for type 1 diabetes by sex (95% CI)	β coefficient for type 1 diabetes (95% CI)	β coefficient for type 1 diabetes by sex (95% CI)
Arterial pressure
	0.45 (0.04; 0.86)	−0.44 (−0.99; 0.11)	0.40 (0.03; 0.78)	−0.49 (−1.00; 0.03)
zDBP	0.19 (0.01; 0.37)	–	0.21 (0.04; 0.38)	–
Lipids
HDL-c (mmol/L)	0.04 (−0.04; 0.12)	–	0.03 (−0.05; 0.10)	–
LDL-c (mmol/L)	0.22 (0.01; 0.42)	–	0.21 (0.02; 0.40)	–
Triglycerides (SYM%)	13.6 (0.6; 26.6)	–	16.5 (4.2; 28.9)	–
Arterial stiffness^†,‡
PWV (m/s)	0.04 (−0.18; 0.26)	–	0.05 (−0.17; 0.27)	–
Endothelial function^†,‡
VTI (cm)	−0.85 (−3.64; 1.93)	–	−1.84 (−4.43; 0.76)	–
Acceleration (cm/s^2)	−80.7 (−142.0; −19.4)	–	−77.4 (−133.1; −21.6)	–
CMR^†
Left ventricular ejection fraction^‡ (%)	1.28 (−0.73; 3.30)	–	0.94 (−0.91; 2.79)	–
Left ventricle mass indexed to (g/m)	−1.87 (−7.70; 3.95)	−6.59 (−14.30; 1.13)	−1.02 (−6.45; 4.40)	−7.89 (−15.16; −0.62)
Papillary muscle mass (SYM%)	−33.8 (−81.5; 13.9)	–	−25.6 (−68.0; 16.7)	–
Average wall (mm)	0.40 (−0.57; 1.37)	−1.46 (−2.74; −0.18)	0.50 (−0.36; 1.36)	−1.47 (−2.65; −0.29)

CI, confidence interval; CMR, cardiac magnetic resonance; HDL-c, high-density lipoprotein cholesterol; LDL-c, low-density lipoprotein cholesterol; PWV, pulse wave velocity; VTI, velocity time integral; zDBP, diastolic blood pressure z-score; zSBP, systolic blood pressure z-score. Models were adjusted for age, sex, ethnicity, android-to-gynoid ratio, and familial income. Triglycerides and papillary muscle mass variables were log-transformed into sympercents because of their heavily skewed distribution. Results in bold indicate p-value < 0.05. Beta coefficient interpretation can be interpreted as follows, using zDBP as an example, multiple imputation analysis: having type 1 diabetes is associated with a 0.21 (95% CI: 0.04; 0.38) higher zDBP compared to not having type 1 diabetes, for a fixed level of the covariates. Significant interactions terms between having type 1 diabetes and sex (males = 1, females = 0) were found for these variables. Interpretation for interaction terms goes as follow, using complete case models as an example. For zSBP: Having type 1 diabetes was associated with a 0.45 SD (95% CI: 0.04; 0.86) higher zSBP in girls, while no difference by type 1 diabetes status was observed in boys (beta (95% CI): 0.01 SD (−0.36; 0.38)). LV mass indexed to height: no difference was observed in girls (beta (95% CI): −1.87 g/m (−7.70; 3.95)). In boys, having type 1 diabetes was associated with lower LV mass indexed to height (beta (95% CI): −8.46 g/m (−13.57; −3.35)). Average wall thickness: having type 1 diabetes was associated with a higher wall thickness in girls although the lower CI bound crossed the null (beta (95% CI): 0.40 mm (−0.57; 1.37)). In boys, having a type 1 diabetes was associated with a lower wall thickness in boys (beta (95% CI): −1.06 (−1.90; −0.21)). We refer to Table S3 for estimated betas and 95% CI for sex-specific multivariable models with multiply imputed data. ^†Additionally adjusted for systolic blood pressure z-score. ^‡Additionally adjusted for heart rate during the test.