P170 Identifying predictors of axial spondyloarthritis flare onset using daily patient-reported data collected via a smartphone app

Abstract

Abstract Background/Aims Smartphone apps for self-tracking symptoms and behaviour can offer novel insights into axial spondyloarthritis (axSpA) self-management and the experience of flares. Previous studies have predicted weekly flare occurrence in axSpA using physical activity data, and weekly flare occurrence in other rheumatic diseases using symptom data. However, no study has yet used daily flare, symptom and behaviour data to predict flare onset. The aims of this study were to examine the association between daily self-reported symptom and behaviour data with axSpA flare onset using data collected via a smartphone app. Methods A secondary analysis of 136 participants (Mage=52, 70% male) used clinical data collected at the Royal United Hospital for Rheumatic Diseases (Bath) and daily self-tracking data collected via the Project Nightingale smartphone app from 2018-21. Self-tracking data for 8 daily and 13 optional symptoms were scored on a 5-point scale. Participants recorded daily whether an axSpA flare started, continued, stopped or there was no flare. Paired t-tests were used to examine differences in symptoms during flare compared to non-flare days. A mixed effects binary logistic regression model was computed, with flare onset as the binary outcome, participant ID as the random effect, and the change (difference) in each symptom score from the preceding day as fixed effects. Odds ratios (OR) are reported to indicate the increased likelihood of flare onset given a one-point change in symptom score. Results Periods of flare were associated with greater anti-inflammatory use, lower confidence in self-management, more fatigue, worse mood, more pain, reduced exercise, worse sleep quality, and greater stress (p<.001), and lower adherence, more frequent chest pain, more hot flushes and greater impact of the menstrual cycle (p<.01) compared to periods without flare. The predictive model found that changes in certain symptoms and behaviours from the day preceding flare were associated with higher likelihood of flare onset: increased pain score (OR = 3.13, 95% CI [2.70, 3.70]), fatigue score (OR = 1.28, 95% CI [1.15, 1.43]) and anti-inflammatory use (OR = 1.25, 95% CI [1.08, 1.45]), and decreased amount of exercise (OR = 1.16, 95% CI [1.05, 1.30]). Conclusion This is the first study to use daily flare and symptom data to predict flare occurrence. Increases in pain, fatigue and anti-inflammatory use and decreased exercise from the day preceding flare significantly increased the likelihood of flare onset. Potential future integration of predictive flare features into smartphone apps for disease self-management could contribute to more effective prevention and targeting of imminent flares, to enhance individuals’ self-efficacy, disease outcomes and quality of life. Critically, such app development will require consideration of the needs and perspectives of people living with axSpA for optimal user experience. Disclosure A. Kypta-Vivanco: None. R. Barnett: None. R. Sengupta: Honoraria; Dr. Raj Sengupta has received honoraria or consultancy fees from Abbvie, Biogen, Novartis, Celgene, Lilly, Chugai, MSD and UCB. Grants/research support; Dr. Raj Sengupta has received grants/ research support from AbbVie, Celgene, Novartis and UCB. Other; Dr Raj Sengupta is also on the advisory boards for AbbVie, Biogen, Chugai, Lilly, Novartis and UCB.