Posted 06 January 2014
By Alexander Gaffney, RAC
Keeping track of adverse events is a tricky task, even for regulators. Even when a drug has undergone a rigorous premarket assessment process, some risks may not become evident until a product is used by millions. And for other drugs, a particularly rare but serious side effect may take months, if not years, to be identified.
But being a regulator in 2014 means having access to something regulators of old did not: Big data.
In late December 2013, the US Food and Drug Administration (FDA) quietly announced that it plans to begin leveraging that data to better assess drugs in post-market settings, allowing it to determine which drugs may be definitely dangerous, and which ones may need targeted warnings to tailor their appropriate use.
At present, side effects are reported to FDA voluntarily, usually by healthcare professionals or companies using a variation of Form FDA 3500. As of 2013, consumers can also report side effects using Form FDA 3500B.
The problem for regulators, however, is twofold. The first problem is that submission of the form is voluntary, and it is thought that the vast majority of adverse events are never reported to FDA. The second problem is the overall complexity of the form, which can result in inaccurate or incomplete reports.
The form is, in other words, not a tool FDA can completely rely on.
But increasingly, many healthcare systems are tracking their patients using electronic medical records (EMRs), sometimes referred to as electronic health records (EHRs). In recent years the US has poured tens of billions of dollars into incentivizing their use in hospital systems, and US government payers like Medicare are poised to soon tie reimbursement to use of EMRs.
Whereas Medicare may be most interested in outcomes, FDA has expressed its preliminary interest in adverse events. Its 20 December 2013 posting on the Federal Business Opportunities (FBO) website states that it is looking for a partner to develop for it a database of de-identified (scrubbed of individually-identifying patient information) EMR data.
"FDA seeks continued access to a longitudinal patient-level EMR data," the agency explained.
Such a system would allow FDA to "estimate the contribution of various risk factors," FDA said, including demographics, health history, diagnoses and procedures, laboratory test orders and results, use of drugs and biologics, and health encounters like hospitalizations or visits to a physician.
And when we said FDA wants access to big data, we do mean big data. The proposed database would include a minimum of 10 million patients, at least half of whom would be so-called "Active" patients. Half of all patients would have to have been in the system for three continuous years, allowing FDA to track health outcomes longitudinally.
Other useful links in the database will hopefully include mother-baby linkages and national death index data, FDA said.
FDA is accepting responses for the "Sources Sought" notice until 7 January 2014.