The document is a revision of draft guidance initially published in May 2015. It explains the circumstances in which it may be appropriate to leverage existing clinical data to support pediatric device indications and labeling, with a view toward increasing the availability of such devices.

The FDA released finalized guidance titled “Leveraging Existing Clinical Data for Extrapolation to Pediatric Uses of Medical Devices.” The guidance outlines the approach the FDA uses to determine whether extrapolation is appropriate, and describes how data can be used to increase precision for pediatric inferences. It also provides a road map for leveraging existing clinical data when submitting premarket approval applications (PMAs) and de novo requests, as well as for use in supporting humanitarian device exemptions (HDEs).

The finalized document clarifies that for PMA, HDE or de novopremarket submissions, it may be appropriate to extrapolate existing clinical data when the course of the disease or effects of the device are sufficiently similar in adults and pediatric patients. However, information should be extrapolated only when the end points used in the adult data sources are relevant to the pediatric population, when the data quality is high and when existing data are determined to be scientifically valid evidence.

Comments submitted by industry on the draft version, according to the FDA, sought clarification of the concept of “borrowing strength.” Thus, the finalized guidance explains that the quantitative information provided by existing adult data can be incorporated in one of two ways: The adult data can stand in as a substitute for pediatric data, or the adult data can be used to supplement pediatric data within a statistical model. This type of combination of data sources, the FDA writes, is known as borrowing strength, and it can significantly boost the sample size of a prospective pediatric study.

The agency specifies that while the exact model used to borrow strength may vary depending on the case, the extent of leveraging depends on the similarity between borrowed data and any pediatric data that will be collected – regardless of the model used.

The guidance also notes that the types of existing data sources that may be considered for extrapolation include data from a variety of clinical investigations (e.g., randomized controlled trials, single-arm studies and studies from any individual treatment arm), historical clinical data, reference samples and published literature.