Large amounts of missing data in clinical trials can undermine the scientific credibility of the studies, according to a special report by an expert panel commissioned by the National Research Council.
The panel, chaired by Roderick Little, the Richard D. Remington Collegiate Professor of Biostatistics at the University of Michigan School of Public Health, also found that common current methods for analysis do not adequately adjust for missing data, sometimes leading to unjustified conclusions about the efficacy of drugs, biologic products and some medical devices.
Little and his colleagues from 12 other universities and organizations concluded that results of clinical trials could be improved by trial design to limit missing data, use of flexible treatment regimens that allow for patient differences, better follow-up of participants and employment of more scientific methods to adjust for missing data.
"Too many investigators think that you can fix up missing data at the analysis stage," said Little, who is also a professor of statistics at the U-M College of Literature, Sciences, and the Arts, and a research professor at the U-M Institute for Social Research.
"There was a strong consensus that people are using data adjustment approaches that are too simplistic. There are better methods to handle missing data in clinical trials."
The report, which was requested by the Food and Drug Administration, is summarized in an article titled "The Prevention and Treatment of Missing Data in Clinical Trials," which appears in the October issue of the New England Journal of Medicine.
It also offers guidance to those who sponsor research, suggesting they work with investigators who have a good track record of following up with participants, and to consider incentives for complete data collection.
The report was commissioned as part of the FDA's Critical Path Initiative, which its website calls a "national strategy to drive innovation in the scientific processes through which medical products are developed, evaluated and manufactured." The initiative's goal is to get scientific discoveries translated into medical products that serve to prevent, cure and treat disease. Little said the panel's report is expected to lead to revised FDA guidelines on how to handle data in clinical trials.
Another U-M author is Susan Murphy, the Herbert E. Robbins Collegiate Professor of Statistics at the College of Literature, Sciences, and the Arts, who also has appointments at the Institute for Social Research and Medical School.
University of Michigan