Statistical collaboration to impact policy decisions.

Abstract

The objective of this paper is to support the following thesis: In health services research, a statistician can, and often should, contribute beyond what is generally defined as the 'classical' statistician project role. I use projects from RAND Health, the health services research division of the RAND Corporation, to demonstrate the issues that arise for the statistician in this expanded collaborative role. These projects are the 'HIV Cost and Services Utilization Study' (HCSUS) and the 'Ephedra: Clinical Efficacy and Side Effects Project'. HCSUS collected information on a nationally representative sample of people in care for HIV infection in order to determine what services were being delivered and their cost in order to guide policy decisions on the allocation of limited health care resources. The ephedra project was a systematic review conducted by the Southern California Evidence-Based Practice Centre to provide information on the existing science so that the National Institutes of Health could guide an expanded research effort to better understand the safety of ephedrine alkaloids. In order to collaborate on more than just the data analysis of a typical project, the statistician must proactively contribute to all project phases. These phases include problem formulation, study design, data collection, analysis, and the communication of results. The statistician needs to effectively and efficiently balance methodological, substantive, practical, and sometimes even ethical demands. As a result of contributing beyond standard statistical tasks, the statistician may directly impact health policy decisions.