PMD50 USING SOCIAL MEDIA TO CHARACTERIZE ADVERSE EVENTS: A CASE STUDY OF BREAST IMPLANT ILLNESS

Abstract

Reports from recipients of breast implants about a diverse set of symptoms has led to the use of the term “breast implant illness” (BII), though no precise clinical definition of a symptom complex has been established. Breast implant options include both saline- and silicone-containing devices with textured or smooth surfaces. Using social media analysis, we qualitatively explored patients’ experience of complaints associated with different types of breast implants. We identified posts on 3 breast implant patient support websites between 2010-2019. Using natural language processing, we searched the post texts for symptoms from the FDA’s list of suspected BII symptoms, such as redness, swelling, skin rash, fatigue and any other mentioned complications. We tabulated complaints reported in posts by implant type. Posts included ≥1 complaint. There were 2,518 posts mentioning saline and 5,470 mentioning silicone gel breast implants. The most common complaints for saline implants were rupture (19.5%), wrinkling/rippling (19.2%), redness/bruising/skin rash (16.1%), inflammation/irritation (10.4%), malposition/displacement/ asymmetry (6.4%), infection (6.1%), deflation (4.9%), capsular contracture (4.0%), and tiredness/fatigue (2.9%). The complaints for silicone implants followed a similar pattern except that redness/bruising/skin rash was more frequent (24.2%), wrinkling/rippling (14.0%) and deflation were less common (0.7%). Brain fog and trouble concentrating were rarely mentioned (<1%). Comparing smooth and textured breast implants, wrinkling/rippling was more frequently mentioned for textured (25.3%) vs smooth (14.9%), while redness/bruising/skin rash was more frequent with smooth (33.0%) vs textured (21.6%). Malposition/displacement was more common for textured (5.4%) vs smooth implants (1.6%). Tiredness/fatigue (4.8% vs 1.7%) and brain fog/trouble concentrating/confusion/not thinking clearly (1.3% vs 0.5%) were mentioned more for smooth vs textured implants. Social media analysis using NLP may be useful to provide insights and qualitative characterization of experience of complications among breast implant recipients. However, multiple factors may influence differences, not just implant type.