Table 1.
Baseline characteristics.
Fig 1.
QT PGS by cases (diLQTS) and controls.
Kernel density estimates, Epanechnikov kernel function.
Table 2.
Risk of diLQTS by medication.
Fig 2.
Marginal (adjusted) probability of diLQTS by level of QT PGS for predictors.
A. Heart failure (HF), B. Atrial fibrillation (AF), C. Dofetilide, D. Amiodarone.
Fig 3.
Nonlinear evaluation of QT PGS.
A. Adjusted quintiles, B. Adjusted restricted cubic splines, C. Unadjusted quintiles, D. Unadjusted restricted cubic splines.
Table 3.
Fig 4.
ROC curves for various models predicting diLQTS.
Fig 5.
Decision-tree example for integration of QT PGS.
Entry into the model is based on a patient for whom a known QT-prolonging medication is to be prescribed. Nodes corresponding to any heart failure diagnosis (HF Dx), use of an Class III antiarrhythmic agent (AAD), and QT PGS ≥ 2 SD above the mean are used to define the decision process. Listed within the leaves (bottom) are the proportion of subjects with diLQTS, and the coverage as a number and percentage of the total population.