Patient-reported quality of life and biometric signals: A real-time window into vaso-occlusive crisis risk
No Thumbnail Available
Authors
Summers K.
Agrippa O.
Yusuf M.
Rana K.
Hegemann I.
Madsen J.T.
Anie K.
Telfer P.
Lugthart, S.
Check for full-text access
Issue Date
2025
Type
Article
Language
Keywords
Alternative Title
Abstract
Introduction: Accounting for approximately 95% of SCD-related hospitalisations and with severe impacts on quality of life, vaso-occlusive crises (VOCs) remain a significant challenge for individuals with Sickle Cell Disease (SCD). This work therefore explored the potential for predicting VOCs through biometric signals collected by wearables and electronic patient-reported outcomes (ePROs), aiming to enable earlier intervention and improve outcomes. Method(s): Data was analysed from 62 UK patients (Jul 24- Feb 25) who had provided informed consent, including continuously captured wearable biometric data at minute-level resolution, and daily ePROs collected through a dedicated app. Using linear models adjusted for sex, gender, and genotype, variation in ePRO and activity data was analysed across the 7 days pre-and peri-VOC, with significance defined at p < 0.05. Health states were assessed using EQ-5D- 5L at baseline, pre-VOC (7 days), and during VOCs, deriving health utility index scores from the UK/England EuroQoL value set. VOCs and linked hospitalisations were self-reported as Yes or No, and days with recorded "Maybe" VOC responses were excluded to reduce misclassification. VOC events <5 days apart were treated as a single episode. Baseline was defined as excluding the 7 days pre-and post-Yes VOCs. Result(s): Median age was 36 (range 17-68), and 76% were female. Genotype distribution included HbSS (73%), HbSC (23%), and others (5%). Compared to both pre-VOC and baseline values, VOC days exhibited the lowest Health State and EQ-5D- 5L scores and declined significantly (Figure 1A), with scores 1 day pre-VOC already significantly lower than baseline. All five EQ-5D- 5L domain severity scores were in turn significantly elevated on VOC days, with significant elevation from baseline of pain/ discomfort and usual activity scores 1-2 days prior to VOC. Significant reductions were revealed through hour-by- hour analysis (Figure 1B) in night-time respiratory rate (21:00- 04:00) and SpO2 levels (22:00-09:00) pre-and during VOCs, compared to baseline, in parallel with lower heart rate variability (HRV) at each timepoint. Across 180 VOCs, 32 (18%) were associated with a patient-reported hospitalisation. In the 1-7 days pre-hospitalised VOC, participants recorded increased activity than non-hospitalised VOCs: longer moderate and soft activity duration, higher step counts, and elevated pulse rates (Figure 1C). Activity levels exhibited significant drops during hospitalised VOCs compared to non-hospitalised VOCs, including step counts, active time, moderate and intense activity, alongside sleep duration. Hospitalised VOCs were furthermore associated with higher mean pulse rates and SpO2 than non-hospitalised VOCs, in addition to lower respiratory rates. Conclusion(s): This work observed significant pre-VOC changes in ePROs, particularly regarding increased pain/ discomfort and decreased QoL scores. Real-time biometric monitoring revealed hourly-level deviations in respiratory rate, SpO2, and HRV pre-VOC in comparison to baseline, highlighting their potential utility as early indicators in VOC prediction. With hospitalised VOCs exhibiting distinct wearable patterns both pre-and during VOCs, indicating particular potential in stratifying hospitalisation risk, these findings ultimately support the feasibility of future remote VOC prediction strategies in disease management, and highlight key physiological signals pre-VOC that could inform critical proactive decision-making.
Description
Citation
Publisher
License
Journal
British journal of haematology