Treatment outcome studies. pitfalls in current methods and practice

The objective of a treatment outcome study is to investigate the heterogeneity in outcome between patients according to factors other than treatment, such as country, institution or physician. Results of treatment outcome studies have already been extensively presented in the medical literature. However, no clear methodology has emerged to perform treatment outcome studies and various methods have been used. This paper reviews the different types of questions addressed in treatment outcome studies, the different methodologies and the different endpoints used. Statistical techniques are mainly descriptive including tables, estimates of survival curves, but regression models have also been used. Most of the studies use registry data, while only a few use discharge data or data available from clinical trials.