Comparative effectiveness research: do we need to reevaluate research ills?
By Liz Scherer | Wednesday, March 10th, 2010
Editor’s note: The Disruptive Women in Health Care blog recently compiled an ebook exploring the issue of Comparative Effectiveness Research (CER) from a variety of viewpoints and perspective. We invite you to download the ebook or read the original posts.
By Liz Scherer. Comparative effectiveness research (CER): it’s the buzzword of the new decade. In fact, Congress recently passed legislation to provide more than $1B to support CER in hopes of improving utilization of existing therapies while simultaneously holding down healthcare costs. The ultimate goal of CER goes even further and paints a rosy vision of patient-centered care and personalized medicine.
However, perhaps these goals are loftier than originally imagined. Newly- published data appearing in this week’s JAMA show that the very research that is supposed to be forging the path for our nation’s health is filled with ills of its own.
An analysis of randomized and observational studies and meta-analyses published in six “high-impact” journals (i.e. NEJM, Lancet, JAMA, Annals of Internal Medicine, BMJ and Archives of Internal Medicine) demonstrated that there is a dearth of CER studies to guide policymaking or clinical decisions. Granted, this underscores the need to expand funding, preferably public funding to fill the gap. However, key findings also showed that only 32% of evaluable medication studies met the criteria for CER, efficacy outcomes were generally emphasized to a substantially greater extent than safety outcomes (only 19% of studies focused on safety), and that a critical element for promoting effective and efficient healthcare, i.e. cost-effective analyses, appeared in only 2% of the studies. Moreover, less than 50% of studies had active comparators and of these, less than a quarter used non-inferiority analyses, thereby obviating the ability to effectively evaluate and compare similar agents with different side effects profiles.













