Methodologic issues in the study of second malignant neoplasms and pregnancy outcomes.

For future studies of late effects, it will be necessary for investigators to consider methodologic issues carefully as a means of insuring the quality of research in this increasingly important area. Recognizing the advantages and limitations in univariate and multivariate analyses will help in selecting the best analytic approach. There are many methods for denoting risk. Selection of the method will often be dependent on study design, but should also be the one believed most effective for communicating the study results. There is little question there is a need for well-designed studies for second malignancies, especially those designed to evaluate late-occurring second cancer (i.e., in the second, third, and fourth decade following treatment). With the aging of the childhood cancer survivor population, questions regarding pregnancy issues are increasing. These questions range from those relating to fertility to others associated with mutagenesis. When considering studies of these late effects, it is important to recognize the need for selecting the most appropriate study design and statistical methods.