Application of the matched nested case-control design to the secondary analysis of trial data