BY Alaina Busch

Treatments for HIV, abnormal heart rhythms, osteoporosis and diabetes will be the initial focus of FDA's comparative effectiveness research program, with the agency's contractor planning to develop software to compare clinical trial designs, statistical techniques for subgroup analysis and tools to evaluate benefits and risk during a product's life cycle. Other efforts are continuing to standardize agency data and contribute information from insurers, FDA officials and other stakeholders said during a Drug Information Association computational science meeting Monday (March 14).

Johns Hopkins University professor Jodi Segal outlined four research projects moving forward in the Partnership in Applied Comparative Effectiveness Science program, a three-year research contract awarded through the economic stimulus and run by Hopkins. A team of biostatisticians, internists and an economist from Hopkins are working on the research in tandem with FDA investigators.

Hopkins received the contract in September and there are three subcontractors contributing claims data, and policy and technology expertise. The Lewin Group will supplement the effort with data from insurers and a multi-payer claims database that is under development, said Clifford Goodman, senior vice president of the group.

Using HIV data, one of the PACES studies will develop trial designs targeted at estimating the effects of treatments in subpopulations, Segal said. The research will generate software to compare adaptive and fixed clinical trial designs.

A second study will analyze data from cardiac resynchronization devices -- implanted products that help the heart beat properly -- to analyze how to combine data from disparate sources. The study is intended to produce Bayesian data models, which allow for the compilation of disparate data sources to form a broader conclusion.

Another analysis will look at the heterogeneity of treatment effects using osteoporosis data, Segal said. The team has developed an analytic framework to process multiple variables in subgroup analysis, including demographics, genetic variants and disease severity. The study will apply the framework to FDA studies involving subgroups, particularly those that lead to policy decisions. It will generate a graphical display of treatment effects in subgroups and possibly use external data from The Lewin Group, she said.

A fourth study, using data from diabetes trials, will develop tools to evaluate a product's risks and benefits, which are generally evaluated independently, Segal said. A multi-criteria decision analysis would allow for risk and benefit tradeoffs in complex situations.

Industry has been wary about how the research will affect product approvals and reimbursement, but stakeholders said comparative effectiveness decisions have already been made by payers when applying new products to the population.

"It really comes down to finding out what the decision makers want," said Robert Giffin, senior research director at the Center for Medical Technology Policy, which is also involved in the PACES project.

Giffin said industry will adjust to the surge of research in the area by changing the types of drugs in development and paying more attention to subgroup analysis. Companies are likely to shift to small clinical trials and align pre-and post-market studies, he said.

Agency officials also tackled industry concerns about access to the data, which is proprietary and FDA cannot share.

"It's not our data to give to anybody," said Lilliam Rosario, associate director of the FDA office of the chief scientist. FDA officials said they plan to publish the tools that are generated by the research, including clinical trial design methods.

Segal noted the unique nature and security considerations of the pilot program, which is a test case for accessing individual level data. "It really tests our ability to do good science with this data," she said.

FDA, meanwhile is working on conversion and validation of legacy data to allow the research to move forward. The Janus data repository will aggregate product data from various sources. FDA officials said HIV and osteoporosis data have been converted and said future data requirements will be more stringent to avoid conversions. They said they expect the Janus structure to change to address problems as they arise. The agency will also make a validation tool public to aid the process, officials said.

"We want to learn from this and we want to move forward," Rosario said, adding that future data will be more standardized.

She said by the end of the year, reviewers will have access to a "data mart," a precuser to Janus, which will be functional by 2013.