Predicting Major Decline in Older Adults: The Role of Time Spent Out of Home

Wan Tai M. Au-Yeung, Lyndsey M. Miller, Joel S. Steele, Nora Mattek, Zachary T. Beattie, Lisa C. Silbert, Jeffrey A. Kaye

Research output: Contribution to journalComment/debatepeer-review

1 Scopus citations


Background: For older adults in community settings, predicting undesired outcomes (needing higher levels of care or death) is important for advancing preventative care and achieving successful aging. We examined whether a combination of baseline characteristics (demographic, clinical) and a continuous digital measure of general mobility and socialization (time-out-of-home) would be predictive of transitions of care and mortality. Method: Longitudinal data from community-dwelling older adults in the Collaborative Aging Research using Technology (CART) Study was examined for occurrences of undesired life transitions: moving residences due to needing more assistance, receiving non-temporary in-home assistance, and death. In-home sensor data (passive infrared motion and contact sensors) was used to create an algorithm for time when no one was home (TOH). For homes with a transition recorded, a 90-day window preceding the transition was used to calculate average TOH per day; for homes that did not have transitions, a 90-day window was randomly selected. Logistic regression was used to predict whether mean duration of TOH (hours), age, gender, Clinical Dementia Rating Scale Sum of Boxes (CDR-SOB), and Cumulative Illness Rating Scale (CIRS) could predict undesired transitions in single-resident and dual-resident homes of older adults. Result: 177 participants residing in 123 homes were included in the analysis: 56.1% of the homes were single-resident homes. Mean age in years was 72 (6.4), 52% of participants were women, mean CDR-SOB = 0.30 (0.46), and mean CIRS = 22 (3.9). Transition to needing greater assistance or death was significantly associated with less TOH (OR 0.66 [95% CI 0.44-0.98]) after controlling for known risk factors that included age, CIRS, CDR-SOB and gender. Older age (OR=1.07 [CI 0.98-1.16]), higher CDR-SOB (OR=2.66 [CI 0.80-8.87]), higher CIRS (OR=1.05 [CI 0.90-1.21]), and being male (OR=0.55 [CI 0.19-1.64]) were associated with a major transition or death with wide CI’s among these mediating factors. Conclusion: Less duration of TOH (likely reflecting mobility and motivation) is predictive of major decline in older adults after controlling for known demographic and clinical risk factors. Digital behavioral biomarkers assessing older adults’ day-to-day behaviors may provide valuable information in addition to clinical assessments.

Original languageEnglish (US)
Article numbere066879
JournalAlzheimer's and Dementia
Issue numberS2
StatePublished - Dec 2022

ASJC Scopus subject areas

  • Epidemiology
  • Health Policy
  • Developmental Neuroscience
  • Clinical Neurology
  • Geriatrics and Gerontology
  • Cellular and Molecular Neuroscience
  • Psychiatry and Mental health


Dive into the research topics of 'Predicting Major Decline in Older Adults: The Role of Time Spent Out of Home'. Together they form a unique fingerprint.

Cite this