In an effort to enhance transparency, members of the Genetic Risk Prediction Studies group — GRIPS — published 25 recommendations for reporting genetic risk prediction studies in 10 journals that run relevant papers.
Among other things, the GRIPS group recommends that researchers reporting studies of risk prediction using genetic factors clearly identify their articles in the title and abstract using keywords; provide specifics as to study design, participation, risk model construction, statistical methods, and related information; discuss the potential limitations of the study as well as its generalizability; and disclose the study's funding sources.
The publication of the GRIPS statement and recommendations in March represented the culmination of two years' work, shared among a group of risk prediction researchers, epidemiologists, geneticists, methodologists, statisticians, and journal editors.
Concerned over the variable quality of genetic risk prediction research, Cecile Janssens first met with the Centers for Disease Control and Prevention's Muin Khoury in February 2009 to discuss a plan to generate guidelines for researchers reporting such results. "We felt that the field might benefit from guidelines in an early stage," says Janssens, an associate professor of epidemiology at the Erasmus University Medical Center in Rotterdam, the Netherlands. After consulting with members of CDC's Human Genome Epidemiology Network, Janssens, Khoury, and their colleagues invited additional expert participants.
In December 2009, during a two-day meeting in Atlanta, the 25 GRIPS members nailed down draft guidelines, some of which they based on principles established by prior reporting guidelines — in particular, those from the Strengthening the Reporting of Genetic Association studies group in which seven GRIPS group members also participated.
It was after the meeting, however, that the team's work really began. From April to June 2010, GRIPS members debated the specifics of two draft articles, Janssens says. In particular, "the papers were rigorously screened for the presence of recommendations on how to do risk prediction studies," she says. "We did not want those."
As the authors also stress in their paper, the GRIPS guidelines are not meant to assess the quality of empirical studies. "The recommendations … address the reporting, not researchers' choices regarding [study] design, et cetera," Janssens says. "Because we all have our preferences for the design and analysis of these studies, we agreed that those choices were the authors' responsibility."
At Erasmus, the Janssens lab maintains a collection of all empirical genetic risk prediction papers; using that, a student in the group has assessed the reporting quality of the studies prior to the publication of the GRIPS statement. Janssens says her team intends to monitor uptake of the GRIPS recommendations going forward. Given her students' baseline evaluation, Janssens hopes that over time "we can demonstrate more clearly why the GRIPS statement is so valuable," she says.