Evaluating treatments for preterm labor: possible solutions for some methodological problems

The usual evaluation procedures for labor-inhibiting therapies are criticized and a new measure is proposed. This measure (rate of pregnancy prolongation) is shown to fulfil the criteria of high sensitivity to treatment effects and of unbiasedness for the gestational age of patients. The use of analysis of covariance is proposed as a solution for the problem of confounding background variables. The advantages of the new evaluation procedure are illustrated by the analysis of a clinical trial of labor inhibition by hypnotic relaxation.