Evaluation of various methods of digoxin dosing

The ability to precisely predict serum digoxin concentrations using 12 published methods in a group of 85 patients was undertaken. Two methods of estimating creatinine clearance and two estimates of ideal body weight were employed as input variables using the 12 dosing methods. This resulted in 40 relationships from which correlation coefficients and linear regression constants were derived for predicted versus measured serum digoxin concentrations. The correlation coefficients between predicted and measured serum digoxin ranged from -0.393 to 0.389. Possible explanations for the low correlation coefficients are interpatient variability in the kinetics of digoxin, the small number of subjects used to generate some of the digoxin dosing methods, undetected patient noncompliance in the present study, the use of empirically derived dosing methods, and/or the use of rather homogeneous patient populations to develop a given method while this study is comprised of a heterogeneous group of patients. The methods studied tend to overpredict serum digoxin concentrations and therefore generally allow safe, first approximations for digoxin dosing.