Missing data in rheumatologic studies and the practical reality of modern approaches to the analysis of incomplete subject data: comment on the article by Baron et al and the editorial by Boers.