Table 3

Mixed models assessing the relation between variables and progression risk of the Fazekas score and infarctions

Fixed effectsFazekas scoreInfarctions
OR (95% CI)P valueOR (95% CI)P value
Model 1 (sex and phenotype separate)
Age1.28 (1.24 to 1.32)<0.00011.26 (1.17 to 1.35)<0.0001
Sex
 Women11
 Men7.4 (2.4 to 23.2)0.0006169 (17 to 1697)<0.0001
Phenotype
Non-classical11
 Classical52.9 (11.0 to 254.8)<0.000164 (2.5 to 1625)0.0119
Model 2 (sex and phenotype combined groups)
Age1.28 (1.24 to 1.33)<0.00011.26 (1.17 to 1.35)<0.0001
Sex, phenotype
 Women, classical11
 Men, classical11.0 (3.2 to 38.4)0.0002224 (19 to 2618)<0.0001
 Men, non-classical0.06 (0.01 to 0.49)0.0091.42 (0.04 to 52.6)0.8483
 Women, non-classical0.05 (0.01 to 0.39)0.00370.07 (0.00 to 7.61)0.2619
Model 2 + <6 months of ERT11
 ≥6 months of ERT1.34 (0.96 to 1.88)0.08211.43 (0.73 to 2.81)0.2951
Model 2+ years treated with ERT****
Model 2 + changes in eGFR0.99 (0.98 to 1.01)0.49970.96 (0.93 to 0.99)0.018
Model 2 + changes in LVMi on MRI1.03 (1.00 to 1.05)0.03681.03 (0.99 to 1.07)0.1726
Model 2+ no hypertension11
 Hypertension1.12 (0.26 to 4.82)0.88210.61 (0.05 to 6.88)†0.6859
Model 2+ changes in BAD0.90 (0.57 to 1.43)0.64921.62 (0.65 to 4.05)0.2983
Model 2 + LDL-cholesterol1.21 (0.58 to 2.64)0.62641.26 (0.31 to 5.13)0.7514
Model 2 + no AF11
 AF0.77 (0.32 to 1.85)0.56170.45 (0.08 to 2.49)0.3584
Model 2 + no ischaemic heart disease11
 Ischaemic heart disease1.92 (0.57 to 6.49)0.29170.75 (0.09 to 6.47)0.796
Model 2 + no valve dysfunction1
Valve dysfunction0.43 (0.10 to 1.86)0.2575
Model 2 + no systolic dysfunction or LVOTO1
 Systolic dysfunction or LVOTO0.28 (0.04 to 2.09)0.2132
Model 2 + MRI scanner Ingenia11
 Intera2.36 (1.56 to 3.57)<0.00011.74 (0.75 to 4.03)0.1946
Model 2 + Fazekas scale1.94 (1.32 to 2.85)0.0008
  • In all models adding additional variables to model 2, the effect sizes of age and sex and phenotype divided subgroups remained similar and are therefore not presented. To reduce the false positive rate (type I errors), we regarded p values <0.01 as significant.

  • *Models did not converge, probably due to collinearity.

  • †The model was unable to run with non-classical patients included, probably due to the low number of non-classical patients with infarctions. Thus, for this analysis only classical patients were included.

  • AF, atrial fibrillation; BAD, basilar artery diameter; eGFR, estimated glomerular filtration rate; ERT, enzyme replacement therapy; LDL, low-density lipoprotein; LVMi, left ventricular mass index; LVOTO, left ventricular outflow tract obstruction.