Discriminative performance of anthropometric variables in the identification of frailty in the elderly.

Vera Elizabeth Closs, Carla Helena Augustin Schwanke


Population ageing has been the focus of many studies as it reflects an important transformation of society that creates a reality with new challenges to be faced. The ageing process can be categorized in two forms: normal ageing or ageing per se or senescence (with associated molecular, anatomical and physiological alterations) and usual ageing or senility (associated with diseases and disabilities). The frailty syndrome is among the most discussed current themes. Frailty is a condition in which the individual presents increased vulnerability to the development of dependency, hospitalization, institutionalization, fall-risk and mortality, among other adverse outcomes, when exposed to a stressor. Various pathophysiological processes involved in frailty are closely related to nutritional and body composition aspects, and anthropometry presents as a useful technique for evaluating these conditions. Given that frailty in its early stages is commonly seen in the elderly community and that available instruments for its identification (such as the widely used Fried Phenotype) are complex and difficult to apply in primary care, the search for simple tools becomes important. In this thesis, three original articles are presented. In the first, article, AGING INDEX DEVELOPMENT IN BRAZIL, REGIONS, AND FEDERATIVE UNITS FROM 1970 TO 2010, we sought to address the ageing index, the parameter through which to evaluate the expansion process of the elderly segment of total population in relation to the relative change in young age group. To this end, data from the Demographic Census and Social Indicators of the Brazilian Institute of Geography and Statistics (IGBE) were analysed. It was evidenced that the Brazilian population is undergoing a rapid ageing process. For the second article, DESCRIPTION OF ANTHROPOMETRIC MEASUREMENTS AND ANALYSIS OF THEIR ASSOCIATION WITH GENDER, AGE, AND FRAILTY SYNDROME IN ELDERLY ASSISTED AT THE FAMILY HEALTH STRATEGY: DATA FROM THE EMI-SUS, data from 583 elderly participants of a cross-sectional study “Clinical and Epidemiological Study of the Elderly Attended at the Family Health Strategy (FHS) in the City of Porto Alegre (EMI-SUS)”, were analyzed. The frequency of frailty (diagnosed by Fried Phenotype) was observed to be 21.5% and pre-frailty 51.1%, and that only one of the 41 evaluated anthropometric measures was not associated with at least one of the investigated dependent variables (gender, age group or frailty). In the third article, PERFORMANCE OF ANTHROPOMETRIC MEASURES IN PREDICTING FRAILTY IN THE ELDERLY: LOGISTIC REGRESSION AND ARTIFICIAL NEURAL NETWORK MODELS, we sought to investigate the performance of 26 anthropometric measurements in predicting frailty and to develop a tool for their identification. Data from 583 participants of the EMI-SUS were analyzed, randomly divided into a learning sample (n=439) and test sample (n=144). Logistic regression and artificial neural network models were used. The neural network demonstrated a superior predictive performance to the logistic regression model and resulted in a reliable tool for identifying frailty in the elderly. The anthropometric measures of weight, biceps skinfold, sagittal abdominal diameter and waist circumference, grouped, proved to be good predictors of frailty in the elderly. 


Ageing; Demographic Ageing; Indexes; Anthropometry; Frail Elderly; Primary Health Care; Neural Networks.


CLOSS, Vera Elizabeth. Desempenho discriminatório de variáveis antropométricas na identificação de fragilidade em idosos. 2015. Tese (Doutorado em Gerontologia Biomédica) – Instituto de Geriatria e Gerontologia da Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre, 27.08.2015.

DOI: http://dx.doi.org/10.15448/2357-9641.2015.2.22393

ISSN-L: 2357-9641

Except where otherwise specified, material published in this journal is licensed under a Creative Commons Attribution 4.0 International license, which allows unrestricted use, distribution and reproduction in any medium, provided the original publication is correctly cited.