The need for collaboration between clinicians and statisticians: some experience and examples.

INTRODUCTION: Very often we see poor communication and collaboration between clinicians and statisticians. Both sides may fail to realise the importance of a truly collaborative effort. In this paper we give examples to illustrate some problems clinicians and statisticians may encounter when they do not have the full support of each other. The aim is to convince both parties the importance of a truly collaborative effort.
METHODS: Real examples in various medical research areas are drawn from the authors' practical experience for illustration. The examples cover various research aspects such as the use of computer software, regression analysis and interpretation of findings.
RESULTS: Superficial collaboration between clinicians and statisticians may lead to serious problems and sub-optimal research practice that may not be obvious in the first sight. Some of the barriers to effective communication and collaboration are discussed.
CONCLUSIONS: On the one hand, robust statistical practice is vital in many medical research projects. On the other hand, medical thinking is important in the formulation and application of statistical strategies. Statistical inputs should be integrated into medical research projects throughout the whole research process. Sporadic contacts between clinicians and statisticians are not enough. Both parties must learn to communicate more effectively and to be willing to collaborate with each other.