An Analysis of a Set of Medical Data with Missing Observations

Abstract

The efficacies of two tranquilizers and a placebo in reducing tension and related complaints were studied at three clinics over a period of six weeks. There was a considerable amount of missing data in this drug trial. The purposes of this project were: (1) To estimate the missing values by the regression technique and (2) to analyse the data by the multivariate analyses of variance method A significant time trend in changes of severity during the six-week period was found in all three clinics. The predominant trend was linear. But quadratic and cubic trends were also found in some clinics. As far as effectiveness of tranquilizers was concerned there was no conclusive answer. In two clinics, the tranquilizers were not proved to be better than the placebo. In the third clinic, however the 2 active drugs were significantly different from the placebo. This inconsistant result of drug effects among the 3 clinics might be due to (1) non-random allocation of subjects to the clinics and (2) the fact that a large proportion of subjects failed to complete the study