The Association of Dementia With Upper Arm and Waist Circumference in Seven Low- and Middle-Income Countries: The 10/66 Cross-Sectional Surveys

Weight loss is recognized in dementia and is associated with increased morbidity and mortality (1–3); however, the relationship is complex. Although midlife obesity is a risk factor for dementia (4), prospective studies have found that weight loss is already occurring several years prior to the clinical onset (5), accelerating in the 3–6 years leading up to diagnosis (6) and throughout the course of the disorder (7). Higher weight in older age is therefore protective for dementia as an outcome (8–10). Most studies have analyzed change in total body weight or body mass index, and trajectories of fat or lean mass are less clear, although loss of lean mass, sarcopenia, is related to loss of mobility and autonomy (11) and increased mortality (12). A case–control study of Chinese older adults (13) found waist circumference to be progressively lower from cognitively normal participants through those with mild cognitive impairment to those with Alzheimer’s disease (AD). Lean mass, measured using dual-energy x-ray absorptiometry, was reduced in early AD compared with controls and associated with whole-brain and white matter volume in a recent cross-sectional study (14). However, longitudinal evidence is conflicting. In Japanese-Americans (8), no association was found between late-life waist circumference and dementia risk, despite associations with lower or declining body mass index. Higher waist circumference was associated with stroke-related dementia but not AD in another study (9), and higher dual-energy x-ray absorptiometry-assessed total and regional adiposity were associated with cognitive decline in the Health ABC study, but only in men (15). Most research of this nature has been carried out in relatively homogeneous populations and to our knowledge, evidence from low- and middle-income countries is minimal. Taking advantage of a large data set derived from multiple surveys carried out using common methodology in a variety of low- and middle-income nations, the aim of this analysis was to investigate the association between body composition and dementia using waist circumference as a proxy for body fat (16) and upper arm circumference as a proxy for lean mass (17,18). Testing the hypotheses that lower waist and arm circumference would be associated with the presence of dementia and with increasing severity of dementia (as quantified by the clinical dementia rating (CDR) scale (19), we sought to describe the level of heterogeneity of these associations between survey sites.