Data-Driven Approaches to Frailty Operationalization and Prediction of Adverse Outcomes

• Author / Creator

  Bohn, Linzy

• Objective: We advance the literature on frailty measurement and conceptualization by applying data-driven analytic technologies to multiple aging morbidity indicators in order to detect clusters of deficits or individual features that elevate risk for (a) frailty emergence and progression and (b) adverse cognitive outcomes or trajectories. Studies 1-3 are presented in Chapters 2-4, respectively. In Study 1, we examined whether baseline frailty profiles could be empirically determined; extracted profiles predicted frailty progression and neurocognitive slowing; and results generalized across sex. In Study 2, we aimed to extract longitudinal frailty profiles (or statuses); characterize patterns of frailty progression; and identify predictors of baseline frailty classifications and transitions. In Study 3, we sought to identify frailty-related features that increase risk for cognitive impairment and dementia; calculate a data-driven frailty index; and (c) examine whether frailty levels vary across clinical cohorts and complementary frailty index operationalizations.
  Overall Method:
  Study 1 participants (n = 649) were cognitively normal (CN) adults from the Victoria Longitudinal Study who contributed data for baseline multi-morbidity assessment and longitudinal trajectory analyses. Exploratory factor analysis (EFA) was applied to 50 multi-morbidity items, revealing 7 separable domains. The proportion of deficits accumulated in each domain was submitted to latent profile analysis. The extracted profiles were tested as predictors of level and change trajectories in a 50-item frailty index and a latent neurocognitive speed variable.
  Study 2 participants (n = 3,074) were clinical cohorts from the National Alzheimer’s Coordinating Center with amnestic mild cognitive impairment (aMCI) or Alzheimer’s disease (AD). Participants contributed baseline and 2-year follow-up data for 43 multi-morbidity items and baseline risk characteristics for prediction analyses. EFA was applied to 43 multi-morbidity items, revealing 5 separable domains. The proportion of deficits accumulated was submitted to latent transition analysis.
  Study 3 participants (n = 255) were from the Comprehensive Assessment of Neurodegeneration in Aging. Participants contributed cross-sectional aging morbidity indicators (n = 84). We used random forest analysis to identify the most important features that discriminate cohorts with subjective cognitive impairment (SCI), MCI, or AD from CN controls. A 30-item data-driven frailty index and a complementary 81-item index was calculated.
  Results:
  Study 1: We detected three early frailty profiles: not-clinically frail (84%), mobility-type frailty (9%), and respiratory-type frailty (7%). Mobility-type frailty predicted accelerated deficit accumulation and neurocognitive slowing, followed by respiratory-type frailty, and not-clinically frail. Results were robust across sex.
  Study 2: We detected two baseline statuses: Not-Clinically Frail (91%) and Moderately Frail (moderate ambulatory impairment endorsed; 9%). At follow-up, Not-Clinically Frail (56%), Moderately Frail (19%), Mildly Frail (mild ambulatory impairment; 21%), and Severely Frail statuses (severe ambulatory impairment; 4%) were detected. Moderately Frail participants were more likely to remain in statuses characterized by a higher frailty burden, and discriminated by age, male sex, AD diagnosis, and global cognition.
  Study 3: Central risk elevating characteristics included quality of life (QoL), lymphocytes, and neutrophils for SCI; QoL, male sex, lymphocytes, and eyesight for MCI; and QoL, olfaction, visual contrast sensitivity, male sex, and instrumental activities of daily living for AD. We also detected features that were selectively sensitive to SCI, MCI, and AD. Clinical cohorts reported (a) comparable levels of frailty and (b) higher levels of frailty on the 30-item index as compared to CN controls and the 81-item index.
  Discussion: In a programmatic series of three studies, this dissertation research applied a suite of data-driven technologies to the challenge of resolving several empirical and clinical inconsistencies in the multi-morbidity and aging literature. Study 1 provides novel insight into critical early domains of frailty (e.g., mobility impairment) that serve as portals into broader and chronic frailty. Study 2 demonstrated that these domains are relevant for aMCI and AD in that (a) frailty statuses varied along a continuum of ambulatory impairment and (b) moderate impairment exacerbated risk for adverse frailty transitions. Study 3 identified selected important features of frailty that elevate clinical risk for SCI, MCI, and/or AD. Overall, these results demonstrate that some features of frailty are important across a clinical spectrum of aging and AD and thus may increase prediction accuracy in clinical-research setting. Further, early interventions targeting mobility and related functional impairments may prevent frailty emergence and progression, as well as downstream negative outcomes.

• Subjects / Keywords

  • frailty
  • cognitive impairment
  • dementia

• Graduation date

  Spring 2022

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-9ehk-p090

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.