Trajectories of muscle strength, muscle function, body composition and bone mineral density in later life: Analysis of determinants, interrelationships and consequences using data from the Health ABC Study

Abstract

Musculoskeletal disorders, such as sarcopenia and osteoporosis, are common among older people. Preventive strategies require understanding of the determinants, interrelationships and consequences of changes in muscle strength, function and body composition in older age.
Considering novel statistical techniques and using data from 3075 men and women (aged 70-79 years) from the Health, Aging and Body Composition Study, this thesis describes 9-year changes in grip strength, gait speed, appendicular lean mass (ALM), whole body fat mass and total hip BMD, and identifies determinants and consequences of lower levels and greater declines in these characteristics.
Declines were linear for ALM but accelerated with age for other characteristics. Declines in characteristics were positively correlated, suggesting they co-occur. Determinants of lower levels of most characteristics included older age, shorter stature and lower physical activity; older age and poorer diet quality predicted greater declines in some characteristics. Lower levels and greater declines in all characteristics predicted hospital admission (excluding ALM and fat mass) and mortality. Lower levels of all characteristics and greater hip BMD declines predicted fragility fracture. Lower grip strength, greater declines in ALM and hip BMD, and lower levels and greater declines in gait speed predicted falls.
These results have clinical implications. First, healthier lifestyles, represented by higher diet quality and physical activity, predicted higher levels and reduced declines in musculoskeletal characteristics. Therefore, encouraging healthier lifestyles may improve musculoskeletal health in older age. Second, interventions to maximize peak levels of musculoskeletal parameters in early adulthood, and to delay age-related declines, may reduce the burden of musculoskeletal morbidity in later life.
Statistical implications of this thesis are that latent class trajectory and growth mixture models were of limited use for identifying groups of participants with varying rates of decline in characteristics. Applying a linear mixed effects model to z-scores obtained from the LMS method and then extracting the random slopes was a feasible method for deriving change measures in these characteristics. Bivariate dual change score models were unsuitable for examining
interrelationships between changes in characteristics as many failed to converge. Longitudinal cohort studies of musculoskeletal parameters should ascertain these at many time-points and implement statistical methods to characterise change which use all repeated measurements.

More information

Identifiers

ORCID for Holly Syddall: orcid.org/0000-0003-0171-0306

Catalogue record

Export record