Table 3

Event prediction based on dichotomised and categorical variables

Univariable analysisMultivariable analysis: model 1 (8 leads)Multivariable analysis: model 2 (3 leads)
WaldP valueHR (95% CI)WaldP valueHR (95% CI)WaldP valueHR (95% CI)
Prediction of 5-year all-cause mortality
Age >75 years32.6<0.0012.181 (1.669 to 2.850)
Female sex0.510.4760.891 (0.648 to 1.224)
Non-ischaemic aetiology2.320.1270.834 (0.660 to 1.053)
CRT-D device implanted21.0<0.0011.710 (1.360 to 2.152)
Heart rate >75 bpm19.7<0.0011.685 (1.338 to 2.121)8.000.0052.956 (1.392 to 6.274)8.060.0052.973 (1.401 to 6.305)
LVEF <25%14.5<0.0011.563 (1.241 to 1.967)
QRS duration >120 ms24.9<0.0012.010 (1.528 to 2.643)
QTc >450 ms8.050.0051.393 (1.108 to 1.752)
QRS-T angle >110°14.2<0.0011.554 (1.235 to 1.954)
TpTe >100 ms0.170.8950.985 (0.783 to 1.238)
QRS n-pl 8 leads >median20.4<0.0011.730 (1.364 to 2.193)7.190.0073.051 (1.350 to 6.894)
T wave n-pl 8 leads >median0.810.3671.111 (0.884 to 1.397)
QRS n-pl 3 leads >median33.2<0.0012.034 (1.598 to 2.589)7.780.0053.189 (1.413 to 7.204)
T wave n-pl 3 leads >median0.950.3291.121 (0.891 to 1.410)
Prediction of 5-year ICD shocks
Age >75 years0.200.6550.909 (0.598 to 1.381)
Female sex7.550.0060.538 (0.346 to 0.837)7.480.0060.539 (0.346 to 0.840)7.080.0080.548 (0.352 to 0.854)
Non-ischaemic aetiology3.970.0460.751 (0.566 to 0.995)
CRT-D device implanted3.290.0700.768 (0.577 to 1.021)
Heart rate >75 bpm2.350.1250.790 (0.584 to 1.068)
LVEF <25%<0.010.9891.002 (0.755 to 1.330)
QRS duration >120 ms1.650.1991.209 (0.905 to 1.614)
QTc >450 ms0.240.6230.933 (0.708 to 1.230)
QRS-T angle >110°0.700.4040.888 (0.672 to 1.173)
TpTe >100 ms4.330.0370.743 (0.562 to 0.983)
QRS n-pl 8 leads >median11.20.0011.609 (1.217 to 2.128)8.930.0031.536 (1.159 to 2.035)
T wave n-pl 8 leads >median12.2<0.0011.647 (1.245 to 2.180)9.520.0021.559 (1.176 to 2.067)
QRS n-pl 3 leads >median2.760.0971.261 (0.959 to 1.658)
T wave n-pl 3 leads >median5.430.0201.389 (1.054 to 1.832)5.380.0201.389 (1.052 to 1.833)
  • Univariable and multivariable (backwards stepwise elimination) Cox regression analyses for the prediction of outcome events during 5-year follow-up (see the text for the distinction of the two multivariable models). Risk factors entered as dichotomised variables, n-pl indices of QRS and T wave loops stratified according to population median. Statistically significant results are shown in bold. In addition to the p values, the levels of Wald statistics are also shown. The grey areas indicate the variables excluded in the different multivariable models.

  • bpm, beats per minute; CRT-D, cardiac resynchronisation therapy-defibrillator; ICD, implantable cardioverter-defibrillator; LVEF, left ventricular ejection fraction; n-pl, non-planarity; TpTe, T-peak to T-end.