Table 2. Associations of mutually adjusted CMR metrics with vascular risk factors and prevalent cardiovascular disease in multivariable logistic regression models with full confounder adjustment.
| CMR metric | Vascular risk factors | Prevalent cardiovascular disease | ||||||
|---|---|---|---|---|---|---|---|---|
| Hypertension | Diabetes | High cholesterol | Smoking (current) | AF | Stroke | IHD | MI | |
| LAVi (mL/m2) | 1.24* (1.21–1.28) 7.59 x10-46 |
0.87* (0.83–0.92) 1.31 x10-6 |
0.96* [0.94, 0.99]0.0129 |
0.88* [0.83, 0.94] 1.65 x10-4 |
1.30* [1.18, 1.44] 4.15x10-7 |
0.96 [0.88, 1.05] 0.3661 |
1.14* [1.08, 1.20] 2.73 x10-6 |
1.09 [1.01, 1.18] 0.0377 |
| LAEF (%) | 0.99 (0.96–1.02) 0.6064 |
0.94* (0.89–0.98) 0.0110 |
0.99 [0.96, 1.02] 0.3950 |
0.93* [0.87, 0.99] 0.0266 |
0.40* [0.36, 0.43] 6.15 x10-91 |
0.88* [0.82, 0.96] 0.0027 |
0.82* [0.78, 0.86] 1.36 x10-14 |
0.82* [0.76, 0.88] 3.82 x10-8 |
| LVM: LVEDV | 1.43* (1.38-1.48) 2.40 x10-99 |
1.20* (1.14-1.27) 1.10x10-11 |
1.10* [1.07, 1.14] 6.28 x10-9 |
1.29* [1.21, 1.38] 1.05x10-13 |
0.80* [0.71,0.90] 2.16x10-4 |
1.04 [0.95, 1.14] 0.4062 |
0.85* [0.81,0.91] 1.19x10-7 |
0.75* [0.68, 0.82] 1.52x10-10 |
| LVGFI (%) | 0.93* (0.90–0.96) 1.07 x10-5 |
0.87* (0.82–0.92) 1.89 x10-6 |
1.00 [0.96, 1.03] 0.7519 |
0.88* [0.82, 0.94] 2.43 x10-4 |
1.23* [1.11, 1.37] 8.50x10-5 |
0.95 [0.87, 1.05] 0.3310 |
0.88* [0.84, 0.94] 2.63 x10-5 |
0.71* [0.65, 0.77] 7.96 x 10-16 |
| GLS (%) | 1.03* (1.00–1.07) 0.0251 |
1.15* (1.09–1.21) 9.34 x10-7 |
0.97 [0.94, 1.00] 0.0382 |
1.12* [1.05, 1.20] 9.47 x10-4 |
1.11 [1.01, 1.22] 0.0387 |
1.04 [0.96, 1.14] 0.3420 |
1.01 [0.95, 1.06] 0.8414 |
1.07 [0.99, 1.16] 0.0836 |
Results are odds ratios, 95% confidence intervals, and P-values. Models are logistic regression models with disease of interest entered as the response (outcome) variable. For the vascular risk factor models, covariates include mutually entered PCA rotated CMR metrics (LAV, LAEF, LVM/LVEDV, GLS, and LVGLFI), age, sex, ethnicity, deprivation, education, body mass index, physical activity, and all the VRFs (except the one set as the model outcome). For the prevalent cardiovascular disease models covariates include mutually entered PCA rotated CMR metrics (LAV, LAEF, LVM/LVEDV, GLS, and LVGLFI), age, sex, ethnicity, deprivation, education, body mass index, physical activity,hypertension, high cholesterol, diabetes, and smoking.
AF, atrial fibrillation; CMR, cardiovascular magnetic resonance; LVGFI, left ventricular global function index; GLS, global longitudinal strain; i, indexation to body surface area; IHD, ischaemic heart disease; LAEF, left atrial ejection fraction; LAV, maximum left atrial volume; LVEDV, left ventricular end-diastolic volume; LVM, left ventricular mass; MI, myocardial infarction; PCA, principal component analysis.
Statistically significant P-values with a false discovery rate of 0.05, giving an approximate threshold of 0.025.