Sometimes the devil is in the details, but device makers had a broad set of concerns about the July draft guidance for the use of real-world evidence (RWE) in regulatory decision-making. Among the comments are those reflecting concern about the potential need for investigational device exemptions for some uses of RWE, and a perception that the scope of the draft seemed to exclude class II devices.

The FDA released the draft guidance in July, stating that it would decide whether a proposed collection of data for generating RWE would require an investigational device exemption (IDE). Among the methods for generating RWE cited in the draft are registry studies, but the agency also pointed to data derived from electronic health records as well, although the document was less than clear on the question of whether RWE could be used for class II devices. (See Medical Device Daily, July 27, 2016.)

Among those who commented to the docket was the D.C.-based 510(k) Coalition, which said in an Oct. 25, letter that the draft "relies too heavily on registries," but also that the draft "relies too heavily on the pre-submission process or prior approval" of such a study, such as an IDE. The group said it was "concerned about the seeming need for a pre-submission for every use of RWE," questioning whether the agency's reviewers could keep up with the additional workload.

On the point of the scope of the draft, the Coalition said the agency seemed to omit engineering analyses and bench testing as potential sources of RWE, but also posed the question of whether data from biobanks might likewise serve as RWE. Another consideration for the Coalition was that the draft seemed to address therapeutic devices to the apparent exclusion of diagnostics, but the group expressed skepticism as to whether the FDA's intent was to allow the use of RWE "as the actual valid scientific evidence sufficient to permit approval or clearance of a device." The coalition further indicated that the draft seemed to infer that consent would be required when the data in question are collected as part of routine medical care.

The Advanced Medical Technology Association's Oct. 25, 2016, comments to the docket echoed the 510(k) Coalition's remark about the draft's seeming reluctance to cite data other than registries as sources of RWE. AdvaMed also said the agency should clarify that the terms of the draft are applicable to 510(k) devices as well as PMA devices, and inquired into the absence of any mention of in vitro diagnostics in the draft.

AdvaMed made note of the National Evaluation System for health Technology (NEST) under development by the FDA via the Medical Device Innovation Consortium, but said the draft failed to describe how RWE would "bring new devices to the market faster." However, AdvaMed brought up the question of how FDA will assign relative weight to the various sources of RWE in making regulatory decisions, and argued that the collection of real-world data related to an approved or cleared device does not necessarily "become research data, subject to informed consent."

Adrian Hernandez, director of outcomes research at the Duke Clinical Research Institute, told Medical Device Daily, "the limitations on RWE would be that people want to make sure that quality exists" in the data. Hernandez said the use of a registry to expand the indications for use "does depend on the circumstance, but a well done registry can provide important information" about comparative benefit and risk, and can serve as an historical cohort. The data from a registry can also provide data on performance measures, presumably including for product development protocols for class III devices so long as those measures are pre-defined.

Hernandez said institutional review boards routinely approve the use of registries, and most healthcare systems use registry data to track the early use and outcomes of novel devices. On the question of whether an IRB would be uncomfortable with the use of registry data for an expanded indication in the absence of an approved IDE, he said, "it depends on the clinical context." He said an IRB would much more likely be amenable to such a thing when the intended expansion of indications doesn't stray far from the original indication for use.

On the point of informed consent, Hernandez said a requirement "would depend on a few issues. One is whether there are additional procedures that would not normally be done in routine care." Consent might also be needed when the registry study would seek information beyond that needed in the provision of routine care, which would also pertain to any follow-up that would otherwise not be done.

"There are some registries that are linked to other data sources, such as Medicare claims" that make use of indirect identifiers that protect privacy, Hernandez observed. In this scenario, the investigation would not require access to information that isn't already available, and he remarked, "in that fashion, you can have data from a registry with longitudinal outcomes," which might not require IRB approval.

Any move by the FDA to expand the potential use of RWE will have some influence on the credibility of such a notion for payers, but Hernandez noted that most payers are already on board with the use of registry data and other sources of RWE in their coverage determinations.

On the point of the use of EHRs in pursuit of RWE, Hernandez said interoperability is not quite the problem it once was because "some of them are mapped to common data models. Those systems can be queried in a fashion that allows data to be aggregated across multiple systems, even if they have different EHRs," he explained. One potential exception would be data generated by a procedure that is not specifically reimbursed, which might not be captured in an EHR. The same might be said for events that happen in the patient's home as well.

As for registry interoperability, Hernandez said, "that's another sticking point."