Anthropometric indicators as a screening tool for sarcopenia in older adults from Florianópolis, Santa Catarina: EpiFloripa Ageing study

1 Universidade Federal de Santa Catarina, Centro de Ciências da Saúde, Programa de Pós-Graduação em Saúde Coletiva. Florianópolis, SC, Brasil. 2 Universidade Federal de Santa Catarina, Centro de Desportos, Programa de Pós-Graduação em Educação Física. Florianópolis, SC, Brasil. 3 Universidade Federal de Santa Catarina, Centro de Ciências da Saúde, Programa de Pós-Graduação em Ciências Médicas. Florianópolis, SC, Brasil. 4 Universidade Federal de Santa Catarina, Centro de Ciências da Saúde, Departamento de Ciências da Saúde. Rod. Governador Jorge Lacerda, km 35,4, Jd. das Avenidas, 3201, 8890-6072, Araranguá, SC, Brasil. Correspondência para/Correspondence to: IJC SCHNEIDER. E-mail: <ione.schneider@ufsc.br>.


I N T R O D U C T I O N
Sarcopenia is characterized by the relative loss of muscle mass. It is considered a multifaceted geriatric syndrome because it often has more than one cause [1]. Muscle involvement may lead to health problems, such as falls [2], physical disability [3], and higher mortality [4]. Geriatric syndromes are common conditions in the ageing process. They are characterized by the interaction between their causes, their complexity, their impact on quality of life, and the resultant health system expenditures [5].
A recent review on the prevalence of sarcopenia in older Brazilians estimated that 12.0% of men and 20.0% of women aged 60 years or more had sarcopenia [6]. Among adults aged 80 years or more, its prevalence is even higher (31.0% and 52.9% in women and men, respectively) [7]. Sarcopenia can be identified by different means, but there is no consensus on the most appropriate method [8,9]. In primary health care, sarcopenia is usually not assessed given the complexity of assessing it, which generally requires a Dual-Energy X-Ray Absorptiometry (DXA) device or Bioelectrical Impedance Analysis (BIA) [10]. Both methods are expensive and hard to access by most individuals who require the diagnosis in the Brazilian health care units and hospitals. Hence, less expensive methods to identify sarcopenia may increase their use as a screening tool, accelerate treatment, direct interventions, and avoid greater complications.
Handgrip strength and anthropometric measurements can be used for diagnosing sarcopenia, but few studies have validated these measurements [10]. These methods are easy to use and inexpensive, allowing them to be widely used. Body Mass Index (BMI) is an indicator often used for identifying underweight in older adults, and it is sensitive to the amount of body muscle mass [7]. The inflammatory characteristics associated with abdominal fat may promote the loss of skeletal muscle mass, increasing susceptibility to sarcopenia [11].
Therefore, the objective of this study was to determine the cut-off points for BMI, Waistto-Height Ratio (WHR), Waist Circumference (WC), and handgrip strength to screen for sarcopenia and determine the association between sarcopenia and these anthropometric indicators.

M E T H O D S
This is a cross-sectional, population-based study which included older adults aged 60 years or more who participated in the second wave of the cohort study called EpiFloripa Ageing (Universidade Federal de Santa Catarina [UFSC], Florianópolis, Santa Catarina, Brazil). EpiFloripa Ageing investigated the health conditions of older adults living in the urban area of the municipality of Florianópolis.
The following two-stage sampling strategy was used for selecting the baseline sample: systematic random selection of 80 tracts from the 420 urban census tracts of the municipality, and random selection of the households (units of the second stage). All individuals aged 60 years or more living in the randomly selected households were invited to participate in the study. A total of 1,705 older adults were interviewed in 2009/2010 (baseline), with a response rate of 89.1%. The second wave of the study, conducted in 2013/2014, included 1,488 eligible older adults because 217 had died. All individuals interviewed in 2013/2014 (n=1,197; response rate of 80.6%) were asked to visit the UFSC to undergo clinical and laboratory tests (n=604). Thus, the study sample consisted of older adults who were interviewed at home in 2013/2014, and underwent clinical and laboratory tests. Sarcopenia was defined as ASMI ≤2 Standard Deviations (SD) of the mean ASMI of a reference population (young adults from the Rosetta Study) by sex, as recommended by Baumgartner et al. [1]. Thus, ASMI (kg/m 2 ) was considered low (loss of muscle mass) when <7.26kg/m² for men and <5.5kg/m² for women.
Body mass index (kg/m 2 ) was calculated by dividing the body mass (kg) by height (m) squared. Height and weight were measured according to standard procedures [12]. Waist Circumference (WC) in centimeters was measured as recommended by Callaway [13]. Waist-toheight ratio was calculated by dividing WC (cm) by height (cm). Handgrip strength of the selfreported dominant arm was measured by a mechanical dynamometer (Takei Kiki Kogyio ® TK 1201, Nagoya, Japan). The individual was asked to squeeze the dynamometer as hard as possible while sitting upright, letting arms hang freely, and having the palms face the body.
Based on the literature [1], the adjustment variables were: age (60-69 years, 70-79 years, and 80 years or more), smoking status (never smoked, ex-smoker, and current smoker), alcohol intake (does not consume, moderate intake, high intake) investigated by the three questions of the Alcohol Use Disorders Identification Test questionnaire [14]. Regular leisure-time physical activity, investigated by the International Physical Activity Questionnaire [15] long version [no (<150 minutes of leisure-time physical activity per week) and yes (≥150 minutes of leisure-time physical activity per week)], and daily intake of fruits and vegetables [16] (<5 daily servings and ≥5 daily servings).
In the descriptive analysis, the prevalence of all the variables included in the study was calculated, with their respective 95% Confidence Intervals (95%CI), for the categorical variables, and mean, standard deviation, minimum, and maximum for the continuous variables, stratified by sex.
The best cut-off points for BMI, WC, WHR, and handgrip strength to screen for sarcopenia were estimated by calculating the sensitivity, specificity, positive and negative predictive values, and area under the Receiver Operating Characteristic (ROC) curve. The validity of an indicator to predict sarcopenia increases with ROC curve. The significance level was set at 5% (α=0.05).
Each variable (BMI, WC, WHR, and handgrip strength) was dichotomized at their cut-off points given by the ROC curve, and then principal component analysis identified the variables related to the outcome (sarcopenia). Only one factor with three variables (WHR, BMI, and WC) was generated for each sex. Next, an indicator of sarcopenia was created using  the three anthropometric variables (WHR, BMI, and WC) that best confirmed the diagnosis of sarcopenia. Therefore, the indicator requires that the three variables be either above or below their respective cut-off points.
The association between sarcopenia and the three anthropometric variables was verified by crude and adjusted logistic regression analyses, estimating the Odds Ratio (OR) and the 95%CI. Five adjustment models were considered for this association:

R E S U L T S
Of the 604 participants who underwent the clinical tests (response rate of 50.38%), 601 were tested for sarcopenia (3 were excluded because of technical limitations), and these 601 individuals were considered the analytical sample of the study. Table 1 shows the mean, standard deviation, minimum, maximum, percentage, and confidence interval of the study variables.  13) in women and men, respectively. Table 3 shows that the indicator (WHR, BMI, and WC) was associated with sarcopenia even when adjusted for sex, age, smoking status, and alcohol intake.

D I S C U S S I O N
This population-based study with older adults from Florianópolis (SC) found prevalence of sarcopenia of 16.03% in women and 28.85% in men. All study markers were considered appropriate to screen for sarcopenia as their sensitivity and specificity values exceeded 60.0%, except WHR, which had a sensitivity of 52.0%. BMI and WC were the indicators best capable of distinguishing between women and men, respectively, with and without sarcopenia. The screening indicator based on three anthropometric variables (WHR, BMI, and WC) was also associated with sarcopenia.
Studies that used the criteria and cut-off points recommended by Baumgartner et al.  [1], as did the present study, found prevalences of sarcopenia ranging from 8.9% to 31.0% in women and from 19.0% to 52.9% in men [1,7,[17][18][19]. However, Bijlsma et al. [9] found much lower prevalences than those found in the study population, ranging from 0 to 4.5% in men and from 0 to 3.0% in women. In addition to the diagnostic criteria, differences between the prevalences may be related to sampling differences and sample characteristics.
Besides age [20] as a factor that influences the prevalence of sarcopenia, other factors are also related and can be involved in the differences found, such as low level of physical activity, peripheral vascular disease, hormonal influences like low vitamin D, peripheral neuropathy, and weight loss secondary to malabsorption [21]. The study prevalence of sarcopenia based on the screening indicator did not differ significantly from the study prevalence of sarcopenia based on ASMI. Hence, in the present study, when combined, anthropometric measurements were capable of predicting sarcopenia.
Body mass index was considered the best variable to screen for sarcopenia in women and the second best in men. Body mass and body composition influence sarcopenia [22]. Although BMI is not capable of distinguishing between body tissues, older adults with lower BMI were more likely to have sarcopenia than those with higher BMI, corroborating other studies [23][24][25].
Body mass index cut-off points for older adults are controversial [26]. Nonetheless, the differences between the BMI cut-off point for overweight identified by the present study and that provided by the Nutrition Screening Initiative, used by the Brazilian Food and Nutrition Surveillance System (>27kg/m 2 ), are small [27]. The study results confirm that BMI is a good variable to screen for sarcopenia, especially in women, and also show the importance of using BMI when screening older adults as it is simple and easy to determine.
Waist circunference is also a good variable to screen men and women for sarcopenia, with cut-off points of ≤92.0cm for men and ≤88.0cm for women. Men and women with low WC were more likely to have sarcopenia. In men this variable was the best predictor of sarcopenia. The cut-off points identified by the present study are similar to those found in the literature [28] as an indicator of risk for metabolic complications. WC is easy to measure and usually used in body composition and central obesity assessments [29]. The present study suggests that WC can be used for identifying older adults with sarcopenia, especially men.
In the present study, WHR too was capable of discriminating sarcopenia satisfactorily. Nevertheless, other studies that used WHR for this purpose were not found, making this the first study to document that WHR is a good variable to screen for sarcopenia. This relationship can be explained as WHR is an anthropometric indicator of abdominal fat and metabolic risk, as well as sarcopenia [11]. As age increases, Handgrip strength is considered a good indicator of total muscle strength in older adults [31]. It is widely used by epidemiological studies [3,4,25,32] as it predicts poor health and functional limitation [33]. The present study found that handgrip strength is a good predictor of sarcopenia. The study handgrip strength cut-off points (<18.5kgf for women and <26kgf for men) were very close to those proposed by the Foundation for the National Institutes of Health [34] (<16.0kgf for women and <26.0kgf for men) and by the European Consensus [10] (<20.0kgf for women and <30.0kgf for men) as indicators of muscle strength loss. Yet, when associated with anthropometric indicators, it was not related.
Although the screening indicator based on anthropometric variables is itself associated with sarcopenia, BMI, WHR, WC, and handgrip strength in both sexes were also associated with the outcome. These results reinforce the possibility of using these measurements, which, together or individually, are important screening tools, capable of being implemented by primary care as they require relatively easy and inexpensive techniques.
The strengths of the study include the fact that this study tried to estimate the pretest diagnostic properties (sensitivity and specificity) of many indicators with respect to sarcopenia. The indicators may be used by health professionals in clinical practice and by other epidemiological studies that do not have access to DXA. This is also the first study that used WHR for this purpose. Moreover, the study used data from a population-based study of older adults from Florianópolis (SC), using validated instruments and standardized procedures, which enabled comparison with other studies. Investment in indicators with high sensitivity is appropriate when monitoring health with the objective of screening the population.
The limitations include the use of different measurements to identify sarcopenia or its absence by many studies on the subject, which may lead to invalid comparisons, hindering the understanding of the factors involved in the genesis and discrimination of sarcopenia; the restriction that stems from the cross-sectional design, which prevents the making of inferences about the outcome. Additionally, the study of sarcopenia-related complications, such as fractures, hospitalizations, institutionalization, and even death, cannot be omitted; and the third limitation regards the losses associated with clinical tests, which may lead to selection bias: only older adults in better health conditions managed to visit the clinical test site, which may result in an underestimation of the prevalence of sarcopenia. In the present study, muscle strength indicators and motor performance were not considered for the definition of sarcopenia as they were not available for analysis. Furthermore, caution is recommended when using anthropometric indicators and handgrip strength to identify sarcopenia as these measurements were developed with other objectives in mind.
The present study tried to find inexpensive and easy-to-use screening instruments to identify older adults more likely to have sarcopenia. BMI was a good predictor in females, and WC was a good predictor in males. Therefore, the screening indicator may be used in large scale, both by epidemiological studies and clinical practitioners.

C O N T R I B U T O R S
SC CONFORTIN and V MENEGHINI contributed to study conception and design, and data analysis and interpretation. LM ONO and IJC SCHNEIDER made a substantial contribution to data interpretation and manuscript preparation. AR BARBOSA and E D'ORSI contributed to study conception and design, and data interpretation. All authors contributed substantially to the critical review and approved the final version of the article.