Abstract TH882: Pigmentation-Related Disparities in Smartwatch Photoplethysmography for Atrial Fibrillation Detection: A Scoping Review

Abstract

Background: Photoplethysmography (PPG) is widely used in smartwatches for detecting atrial fibrillation (AF), the most common cardiac arrhythmia. However, technical limitations related to melanin absorption, inadequate skin tone measures like the Fitzpatrick scale, and underrepresentation of persons with darker skin tones in validation studies are concerns that may compromise device accuracy and exacerbate health disparities. This scoping review aimed to systematically map the extent to which the literature acknowledges and addresses these pigmentation-related limitations. Methods: Following the Joanna Briggs Institute methodology, we conducted a scoping review of 6 databases: MEDLINE, EMBASE, Cochrane, Web of Science, Scopus, and ClinicalTrials.gov. Eligibility included English-language articles on PPG and cardiovascular disease published between January 1, 2015 to May 1, 2024 in adult populations and included grey literature (e.g., abstracts and reviews). Data were extracted to assess whether articles mentioned the identified limitations or reported results stratified by skin tone or race/ethnicity. Results: Of the 4,981 articles screened, 635 met inclusion criteria. Of those, 68 (10.6%) addressed technical limitations, 4 (0.6%) addressed skin tone measures, and 40 (6.2%) addressed underrepresentation. Of the 130 (20.4%) clinical studies that evaluated PPG-based AF detection, 11 (8.4%) addressed technical limitations, none addressed skin tone measures, and 11 (8.4%) addressed underrepresentation. Only two clinical studies reported stratified results. In a large-scale trial (n=455,699), the diagnostic yield for confirmed AF following an irregular rhythm detection was 2.2% (4/180) for Black non-Hispanic participants vs. 7.7% (293/3,812) for White non-Hispanic participants; however, formal statistical analysis and diagnostic accuracy were not reported. In a validation study, (n=111), the Black/African American subgroup was found to have a lower interval-level sensitivity (67.5%) compared to the White subgroup (97.2%). Conclusion: Pigmentation-related disparities and technical challenges of PPG due to skin tone are well-known yet seldom addressed as evidenced by our scoping review. Studies reporting stratified results further demonstrate the need for additional research. Future research should address the accuracy of PPG in darker skin tones, the accuracy of skin tone measures, and adequate recruitment of underrepresented populations.