Our study is unique in that it analyzes reported AE data over a 34-year period for injectable dermal fillers. To our knowledge, this novel method of calculating AE rates across dermal fillers and for individual products is the first of its kind that facilitates usage-normalized comparison of different filler types.
All OpenFDA data are self-reported and therefore have inherent limitations. Anyone can enter information on AEs in this system, including both patients and health care providers, so the quality of the input may be variable. However, this output is the only representation we have for nearly 35 years of AE history for this burgeoning category of popular aesthetic treatments. Another study limitation is that not everyone may know that reporting an AE in the OpenFDA is an option; therefore, we may be missing a portion of AEs due to underreporting. Underreporting may be especially at play in the years before the Internet was prevalent for residential use since access to the Internet would be required to report an AR on the website. However, examining the available data provides an important window into valuable information on complications that have occurred and have been reported for FDA-approved dermal fillers.
An additional challenge in constructing this study was assessing the total number of injectable dermal filler treatments being performed annually across filler types for normalization of the data. Although the absolute numbers of filler use as captured by the ASPS are smaller than the true total filler use across all injectors, the relative use of different filler products will be similar across all specialties because it reflects product popularity. Annual surveys on aesthetic procedures also are conducted by the American Society for Dermatologic Surgery and the American Association for Facial Plastic and Reconstructive Surgery, but neither one captures the relative usage of different filler types. Because individual filler companies do not publish their annual sales numbers by product, the ASPS data give us the best gauge of relative use of fillers by product type given the available information. We conclude that the comparison of AE rates would remain the same even if we had data for total annual filler use across specialties.
Our graphical depiction of the data clearly demonstrates the low AE profile of HA fillers, which is in line with the general consensus of their safety that has contributed to their vast popularity; however, this study represents the first time usage-normalized AE rates are compared to other filler compositions. Hyaluronic acid fillers have the unique feature of being able to be dissolved with the hyaluronidase enzyme, which can limit adverse event potential as compared to other ingredient classes of filler types and may be reflected in their low overall AE profile. The AE rate spike and resolution for collagen fillers represent what we refer to as a “normal learning curve” based on our analysis of the data set as a whole, suggesting an appropriate time course of increased familiarity with the product without inherent issues with the product itself. Multiple sequential anatomic site indications were approved for hydroxylapatite fillers from 2006 through 2015, which may have yielded overlapping learning curves for each approval, resulting in a rather uniform AE rate. The early drop-off in AE rates after the 2015 anatomic site approval may represent the beginning of a normal learning curve, and continued surveillance of AE rates would be of value to confirm this trend. We saw a similar 3-year learning curve for PLLA fillers as the curve for collagen fillers, suggesting a normal learning curve and no out-of-line safety issues. Polymethylmethacrylate fillers were approved in 2006 and were taken off the market for a period in the late 2000s, explaining the drop-off. Once they were back on the market, we do not see a typical learning curve for PMMA, which may warrant surveillance for safety by both clinicians and the FDA.
Our study represents a novel method of evaluating the safety of medical devices, specifically aesthetic fillers. We showed that every AE rate curve for different filler types tells a story. Reactions to AEs for new fillers should be placed in the context of whether they seem to be following the established learning curve.