SOME ISSUES IN ESTIMATING THE EFFECT OF PROGNOSTIC FACTORS FROM INCOMPLETE COVARIATE DATA

In evaluating prognostic factors by means of regression models, missin g values in the covariate data are a frequent complication. There exis t statistical tools to analyse such incomplete data in an efficient ma nner, and in this paper we make use of the traditional maximum likelih ood principle. As well as an analysis including the incompletely measu red covariates, such tools also allow further strategies of data analy sis. For example, we can use surrogate variables to improve the predic tion of missing values or we can try to investigate a questionable 'mi ssing at random' assumption. We discuss these techniques using the exa mple of a clinical study where one important covariate is missing for about half the subjects. Additionally we consider two further issues: evaluation of differences between estimates from a complete case analy sis and analyses using all subjects and assessment of the predictive v alue of missing values.