Computer-Assisted Digoxin Therapy

Abstract

In 42 patients requiring digitalis, and randomly divided into two groups, the performance of a computer program using patient size and renal function to compute digoxin dosage was compared to that of unaided physician judgment. Serum digoxin concentrations were measured repeatedly. Efficacy was measured by changes in the manifestations of heart failure, and toxicity by electrocardiographic criteria. For each patient, physicians specified a desired serum digoxin concentration and predicted this concentration at each visit. For one group, the computer program suggested the dosage needed to achieve the desired digoxin concentration.

Efficacy was the same in both groups, and there was no toxicity. Although the computer slightly outperformed the physicians, prediction and achievement errors were unacceptably large. Hence, much between-patient variability in serum digoxin concentrations remains unexplained after adjustments for dose, body size and renal function. This argues for measurement of digoxin concentrations and their use for feedback dosage adjustment. (N Engl J Med 289:441–446, 1973)