Abstract
Three methods of estimating inter- and intrasubject variances and coefficients of variation of digoxin renal clearance are discussed and illustrated. All three methods gave essentially the same values for intrasubject variation; two of the methods (II and III), based on two-factor and one-factor analyses of variance, respectively, gave the same estimates of intersubject variation. The preferred method I involves directly calculating variances from the rows and columns of the tabular data, where the rows are repetitive treatments in the same subject and the mean variance represents intrasubject variance; a column represents different subjects' results on a given treatment and the mean of the column variances represents intersubject variance. Problems in attempting to use components of variance are discussed. Intrasubject coefficients of variation of digoxin renal clearance averaged 23.8 percent by method I and ranged from 15.2 to 28.6 percent. Intersubject coefficients of variation averaged 30.5 percent with a range of 17.5–41.8 percent by method I and averaged 42.1 percent with a range of 19.2–49.7 percent by methods II and III.