TL;DR
This study demonstrates the feasibility of using smartwatches to collect daily symptom data from individuals with multiple long-term conditions (MLTC-M). The findings highlight fatigue as a primary symptom and reveal the potential for this technology to track symptom variability and identify dynamic relationships between symptoms, offering new avenues for clinical care and research.
What Problem Does This Paper Address?
Managing multiple long-term conditions (MLTC-M), also known as multimorbidity, is a growing global health challenge. Individuals with MLTC-M experience a complex web of interconnected symptoms that fluctuate over time, significantly impacting their quality of life. Understanding this symptom variability and how different symptoms interact is crucial for effective diagnosis, treatment, and self-management. However, traditional methods of symptom tracking are often infrequent and may not capture the day-to-day fluctuations that patients actually experience. This paper addresses the need for more dynamic and granular data collection methods to better understand and manage the complex symptom landscape of individuals with MLTC-M.
Methods
The study employed a prospective observational feasibility design called "Watch Your Steps." Participants with MLTC-M used a smartwatch to report on seven core symptoms (pain, fatigue, wellbeing, mood, stress, function, and sleep quality) daily, along with up to eight additional symptoms personalized to their specific conditions. This data collection occurred over a 90-day period. The analysis involved summarizing symptom ratings at individual and group levels, plotting daily symptom trajectories, and investigating relationships between symptoms and demographic factors. Exploratory statistical tests were used due to the study's underpowered nature for robust hypothesis testing.
Datasets
The study utilized data collected from 52 participants in the "Watch Your Steps" feasibility study. Participants provided daily symptom ratings via a smartwatch over a 90-day period. No external datasets were used for training or evaluation.
Key Insights and Ideas
- Smartwatches as a viable tool for symptom tracking in MLTC-M: The study successfully demonstrated that smartwatches can be used to collect rich, longitudinal symptom data from individuals with multiple chronic conditions, paving the way for more dynamic health monitoring.
- Fatigue as a dominant symptom: Fatigue emerged as the most problematic symptom reported by participants, underscoring its significant impact on individuals with MLTC-M.
- Interconnectedness of symptoms: The research revealed strong correlations between fatigue, pain, and functional impairment, suggesting a complex interplay between these symptoms that clinicians and patients should be aware of.
- Individual symptom trajectories are highly variable: The study highlighted that symptom experiences are not static and can fluctuate significantly from day to day, emphasizing the importance of capturing this variability for a true understanding of a patient's condition.
- Potential for early detection and intervention: By tracking daily symptom fluctuations, there's a potential to detect early signs of deterioration, enabling timely interventions and personalized care.
Key Results
- Fifty-two participants submitted symptom ratings, with an overall symptom question completion rate of 45%.
- The majority of participants (64%) had MLTC-M affecting three or more disease areas, and the musculoskeletal system was the most commonly affected (67%).
- Fatigue was rated as the most problematic symptom overall, followed by pain.
- Increased comorbidity (3+ disease areas) was associated with poorer ratings for symptoms.
- Fatigue ratings were strongly correlated with pain (R=0.65) and level of perceived dysfunction (R=0.68).
- Modest negative correlations were observed between fatigue and wellbeing (R=-0.49), sleep quality (R=-0.47), and mood (R=-0.43).
- Exploratory analysis suggested that patients with increased comorbidity or female sex tended to have higher fatigue ratings, though these findings were not statistically significant.
Limitations and Future Work
The authors acknowledge several limitations:
- Small sample size and generalizability: The study had a relatively small number of participants who were motivated and digitally literate, limiting the generalizability of the findings to the broader MLTC-M population.
- Underpowered for statistical testing: The dataset was insufficient for robust statistical testing, meaning that observed differences and relationships should be interpreted with caution.
- Missing data: The study did not investigate whether missing data was random or due to participants feeling particularly well or unwell.
- Within-day variability: The analysis focused on daily summaries, and within-day symptom variability was not explored, despite its potential contribution to the burden of living with MLTCs.
- Self-reported disease areas: The stratification by disease area was underpowered for robust conclusions, and disease areas were self-reported, which may be inaccurate.
Future work suggested by the authors includes:
- Advanced analytical methods: Developing sophisticated methods to describe and summarize changing symptom patterns over time, moving beyond simple linear or cause-and-effect models.
- Integration with other data sources: Linking self-reported patient symptoms with clinician-reported data from electronic health records to verify diagnoses and support clinical decision-making.
- Modeling time-series data: Utilizing time-series data to explore causal relationships and predict future disease activity.
- Investigating physical activity: Examining how patterns of physical activity, assessed via smartwatch sensors, relate to fatigue.
- Prediction and just-in-time interventions: Leveraging symptom tracking data for early detection and prediction of events, leading to timely adaptive interventions delivered through new care pathways or digital interventions.
- Scalability: Addressing the need for provided smartwatches in future studies, given the current low penetration of these devices in the MLTC population.
Summary updated