By Deb Borfitz
January 26, 2009 | Long before Lincoln Technologies was acquired in 2005 by Phase Forward, the firm began building tools to search for safety signals in the humungous health care database maintained by the U.S. Department of Defense (DoD). So it was unsurprising that the DoD last year chose Phase Forward’s Lincoln Safety Group to support a new drug safety initiative, enabling it to provide the FDA with rapid assessment capabilities.
The software development project will extend the capabilities of the Empirica Suite, the company’s pioneering pharmacovigilance and adverse event (AE) reporting solution, to accommodate the vagaries of longitudinal medical records and insurance claims data housed in the DoD database, says Phase Forward Lincoln Safety Group President Chan Russell. It will be the “first significant use” of the Empirica Suite outside of the clinical trials and post-marketing surveillance arena.
FDA researchers will access de-identified health care data on upwards of ten million Army active duty personnel, their family members, and retirees to determine which safety signals merit more thorough investigation. That makes the database large enough to identify even 1-in-40,000 AEs. Researchers will also be able to query the data to better separate real from imagined public health concerns.
Web-based data access is being provided through Phase Forward’s established production server facility in Marlborough, Mass., says Russell, who declined comment on the number of staff dedicated to the project and all of the key contract terms. General support services will be offered to no more than about 20 FDA and DoD personnel.
To date, the project has been driven largely by “specific concerns about drug-event associations,” says Russell. “On the DoD side, it’s a little like formulary management [by a] health care plan. If there is a safety problem affecting patients, they may want to change prescribing behaviors or make the [medicine] a second-line instead of a first-line drug.” The FDA’s perspective, in contrast, is “general protection of the public health.”
When a health concern arises, the first order of business is to determine if the DoD database is relevant; that is, if it represents enough people who have taken the suspect drug, says Russell. To date, Empirica has typically been used “to do comparisons between those exposed to a drug versus those who were not or to compare drugs in a class. We’ve had 10 or so of those kinds of [multi-session investigations] so far.”
Information in the DoD database is “quite different” than what sits in clinical trials databases, says Russell. The data’s main virtue is that it represents “what really happens with real patients,” including diagnoses made, lab tests given, and prescriptions ordered. “The data represents a true, real-world snapshot of a patient. Anything goes. You can have people on 50 [different] drugs.” The data’s main vice is that it’s more a collection of health stories than moment-in-time events, such as AE reports. “Some diagnoses strongly imply an AE, but a lot of times you have to infer it.”
Russell says Empirica’s task with the DoD project is “almost a mix of what it does with trials [in that it’s intrinsically comparative] and post-marketing data [in that it seeks the outliers]. With health care data, you can do both—more broadly ask questions and do focused comparisons.”
While epidemiological studies have been done for years to tease out safety issues, says Russell, up until now no one has tried to make the information readily accessible to the end users (notably physicians) who make decisions about patient care. The project will also provide the FDA with stronger evidence of safety problems before it undertakes an expensive, full-blown epidemiological study or mandates that a drug manufacturer conduct post-marketing trials. As a next step, the DoD may be asked to verify the accuracy of information in the database by, for example, cross-checking diagnostic codes with patient charts. “They’ll be milking the data for all it’s worth to get rid of confounding problems.”
One big task for Phase Forward will be to better define drug “exposure” and “outcome,” taking into account multiple separate exposures per patient and concomitant exposure to multiple medications as well as the presence or absence of multiple symptoms, says Russell. Should the FDA determine that a full-scale epidemiological study is warranted, Phase Forward will help refine the selection of patients considered for inclusion.
Russell adds that data-sharing may ultimately broaden to include other federal agencies, such as the CDC, whose purview covers vaccine safety.