The effect of COVID-19 on the home behaviours of people affected by dementia

Alina-Irina Serban, Eyal Soreq, Payam M. Barnaghi, Sarah Daniels, Rafael A. Calvo, David J. SharpNpj Digit. Medicine 2022 · OpenReview

TL;DR

This study leverages unobtrusive home sensor data from 31 individuals with dementia in the UK to quantify the impact of COVID-19 lockdowns on their daily behaviors. The findings reveal an increase in in-home activity and a significant decrease in time spent outdoors during lockdown periods, highlighting the utility of passive monitoring for tracking social isolation in vulnerable populations.

What Problem Does This Paper Address?

The COVID-19 pandemic imposed unprecedented public health measures, such as lockdowns and social distancing, which disproportionately affected vulnerable populations, including the elderly and individuals with dementia. These measures, while aimed at preventing infection, raised significant concerns about increased social isolation and its detrimental effects on cognitive and mental health. Directly assessing how these measures impacted the daily lives and behaviors of people living with dementia (PLwD) was crucial. Traditional methods of assessing behavior, like self-reporting or direct observation, are challenging with PLwD due to cognitive impairments and compliance issues. This paper addresses the need for an objective, continuous, and unobtrusive method to monitor the home behaviors of PLwD and understand their response to pandemic-related restrictions, particularly concerning social isolation.

Methods

The researchers employed a passive home monitoring system utilizing a network of Internet of Things (IoT) sensors. These included:

  • Passive Infrared (PIR) sensors: Placed in key rooms (bedroom, lounge, kitchen, bathroom, hallway) to detect movement.
  • Door sensors: Installed on front and back doors, and the fridge door, to record opening and closing events.
  • Smart plugs: Connected to kitchen appliances (kettle, toaster, microwave) to monitor usage.

Data from these sensors were collected continuously for 498 days, spanning from December 1, 2019, to April 12, 2021. This period encompassed six distinct phases of the pandemic, including national lockdowns.

A key methodological contribution is a custom algorithm developed to quantify time spent outside the home. This algorithm combines door sensor data (identifying entry/exit events) with the absence of in-home PIR activity. Events shorter than 3 minutes or longer than 9 hours were excluded to filter out sensor noise and overnight absences.

Statistical analysis primarily involved Linear Mixed-Effects (LME) modeling. This approach was used to examine changes in total household activity and time spent outside across different pandemic periods, while accounting for individual household heterogeneity and occupancy (single vs. multiple occupants).

Datasets

The study utilized data collected from 31 households of individuals diagnosed with dementia or mild cognitive impairment in the UK. The data collection period spanned from December 1, 2019, to April 12, 2021, covering pre-pandemic baseline, the initial outbreak, and multiple lockdown and relaxation phases. This dataset is unique in its continuous, long-term monitoring of a vulnerable population during an unprecedented global event. No new benchmark datasets were introduced; rather, existing sensor data was leveraged.

Key Insights and Ideas

  • Proactive Behavioral Change: PLwD began altering their behavior (increasing in-home activity) before official government lockdown measures were announced, suggesting an intuitive response to the perceived threat of the pandemic.
  • Lockdown's Impact on Mobility: The study provides quantitative evidence of how drastically lockdowns restricted movement, with a significant reduction in time spent outdoors, particularly during the first national lockdown.
  • Passive Monitoring as a Digital Biomarker: The research strongly advocates for the potential of unobtrusive home sensor data as a "digital biomarker" for social isolation and changes in daily routines in vulnerable individuals. This can inform social care and healthcare interventions.
  • Occupancy Matters for Isolation: While overall activity changes were not significantly different between single and multiple occupancy households, the paper notes that single occupants are inherently more prone to social isolation, a critical factor for targeted support.
  • Daytime vs. Nighttime Activity: Changes in activity were more pronounced during daytime hours, suggesting that the pandemic's impact was primarily on daily routines and engagement rather than nocturnal patterns.

Key Results

  • Increased Home Activity: The onset of the COVID-19 pandemic was associated with a 14% increase in total home activity (LME ANOVA, p < 0.001). This increase was observed across both single and multiple occupancy households.
  • Reduced Outdoor Time: During the first lockdown (P3), the average time spent outside fell from a baseline of 55 minutes to 26 minutes. This was the most pronounced reduction observed across all pandemic periods.
  • Daytime Activity Shift: Activity levels changed significantly during daytime hours (05:00-23:00) but not during the night (23:00-05:00), with a significant interaction between the pandemic period and time of day.
  • Entryway Event Changes: A reduction in entryway events was observed during the first lockdown and relaxation period, followed by a relative increase in back door usage, potentially indicating more garden activity.

Limitations and Future Work

Limitations acknowledged by the authors: * Sensor Presence Impact: While unlikely, the presence of sensors could theoretically influence participant behavior. However, the study design, with participants enrolled before the pandemic and no discussion of COVID-19 protocols with them, mitigates this concern. * Limited Behavioral Detail: The current sensor setup provides information on movement and appliance use but cannot infer detailed cognitive function or specific behaviors beyond what the sensors capture. * Incomplete Household Coverage: The sensor placement, while focused on active areas, may not cover all rooms, potentially leading to an incomplete understanding of behavior in less-used spaces.

Future Work suggested: * Expanding Sensor Capabilities: Deploying additional sensors to capture a wider range of behavioral and cognitive information. * Integrating with Health/Social Care: Developing practical ways to incorporate these digital measures into health and social care systems to proactively identify and support vulnerable individuals at risk of social isolation. * Broader Application: Extending this approach to assess compliance with public health measures and identify unintended consequences of interventions in other vulnerable groups.

Summary updated

Abstract

The COVID-19 pandemic has dramatically altered the behaviour of most of the world’s population, particularly affecting the elderly, including people living with dementia (PLwD). Here we use remote home monitoring technology deployed into 31 homes of PLwD living in the UK to investigate the effects of COVID-19 on behaviour within the home, including social isolation. The home activity was monitored continuously using unobtrusive sensors for 498 days from 1 December 2019 to 12 April 2021. This period included six distinct pandemic phases with differing public health measures, including three periods of home ‘lockdown’. Linear mixed-effects modelling is used to examine changes in the home activity of PLwD who lived alone or with others. An algorithm is developed to quantify time spent outside the home. Increased home activity is observed from very early in the pandemic, with a significant decrease in the time spent outside produced by the first lockdown. The study demonstrates the effects of COVID-19 lockdown on home behaviours in PLwD and shows how unobtrusive home monitoring can be used to track behaviours relevant to social isolation.
BibTeX
@inproceedings{serban2022effect,
  author = {Alina-Irina Serban and Eyal Soreq and Payam M. Barnaghi and Sarah Daniels and Rafael A. Calvo and David J. Sharp},
  title = {{The effect of COVID-19 on the home behaviours of people affected by dementia}},
  booktitle = {Npj Digit. Medicine},
  year = {2022},
  url = {https://openreview.net/forum?id=DK8TiRgEvZ},
}
Figure