INTRODUCTION

Functional limitation caused by reduced mobility is an age-related phenomenon that negatively affects the physical independence of older adults (¹). One of the main consequences of this limitation is sarcopenia, a disease characterized by low muscle strength and reduced amount of skeletal muscle mass (SMM) (²). Sarcopenia affects 10 percent of older adults worldwide (³) and 60 percent of older women in Brazil (⁴). When untreated, sarcopenia exposes older adults to functional impairment, mobility disorders, increased risk of falls, loss of functional independence, physical frailty, increases in hospitalization, decreased quality of life, and increased risk of premature death (⁵). Therefore, the early identification of the disease considerably reduces the economic impact on health systems and on the personal and social burdens of care for older adults (⁶).

For the diagnostic confirmation of sarcopenia two conditions are required: reduced muscle strength and decreased SMM (²). SMM parameters can be obtained from dual-energy X-ray absorptiometry (DXA), bioelectrical impedance analysis, magnetic resonance imaging, computed tomography, muscle biopsy, or magnetic resonance spectroscopy (²). However, these require skilled health professionals, and are high-cost and difficult to access in the context of regular clinical practice, particularly in low- and middle-income countries (⁷). Although the measurement of SMM by DXA is one of the most widely used procedures (⁴), this test lacks a good relationship with reduced mobility as associated with sarcopenia (⁸), and thus has limited diagnostic value (⁸). On the other hand, calf circumference (CC) is a cost-effective and expedient alternative for estimating SMM, a clinically relevant outcome in community-dwelling older adults (²,⁹).

Although CC has a good association with mobility in older adults (¹⁰), this relationship is characterized by an inverted U-shaped form (⁹). Normalizing CC by body size, as recommended, will correct this non-linear relationship (²). Normalization of SMM is already performed on selected DXA variables, such as appendicular skeletal muscle mass (ASMM), usually corrected relative to body size (ASMM·height^-2, ASMM·body mass^-1 or ASMM·BMI^-1) (¹¹). Unfortunately, these variables have weak associations with clinically relevant outcomes in older adults (⁸,¹²). Moreover, the identification of a low SMM index (ASMM·height^-2) includes the absolute value of CC (¹³-²²) (Table I) for generating cutoff points, which will penalize older adults with a smaller body size.

Table I.  Studies that proposed cutoff points to identify low skeletal muscle mass (SMM) by calf circumference (CC), identified from the Receiver Operating Characteristic (ROC) curve analysis 

ASMM: appendicular skeletal muscle mass; SMM: skeletal muscle mass; Sens.: sensitivity; Spec.: specificity. *Calculated from the fat-free mass obtained by bioelectrical impedance analysis. ^†Serum albumin < 30 g·L^-1 or BMI < 19 kg·m^-².

Given the limitations evident in quantifying SMM, the objective of this study was to propose cutoff points for CC normalized by body size to identify low SMM in older women with regard to functional limitation. Our hypothesis is that normalization of CC provides a pragmatic indicator of functional limitation that, in turn, indicates risk of sarcopenia. Therefore, monitoring functional limitation could alert to the degenerative consequences resulting from the ageing process in women at a public healthcare level.

MATERIALS AND METHODS

RESULTS

Figure 1 shows the flowchart for recruiting study participants. One hundred and forty-eight women were initially eligible for the studies. After applying exclusion criteria, the analyses were carried out in 145 women (78 young).

Figure 1.  Flow chart of women recruited by age groups. 

Descriptive statistics for MMSE score, chronological age, anthropometric variables, and walking distance are shown in table II. Older women were older, and had greater body mass, BMI, and absolute CC. Younger women were taller (10 cm) and weighed less (6 kg) than their older counterparts. The higher BMI found in older (≅ 27 kg·m^-2) compared to younger women (≅ 21 kg·m^-2) illustrates a greater weight gain with age. Likewise, the highest normalized CC values observed in young women showed an inverse relationship with absolute CC according to age. From a functional perspective nearly two thirds of the older women (39 %; n = 26) did not achieve a good functional score on the walk test.

Table II.  A descriptive analysis of young and older women, body measurements, and calf-circumference (CC) cutoff points, absolute and normalized, to identify low skeletal muscle mass (SMM) and functional performance in older women 

CI: confidence interval; SD: standard deviation; MMSE: Mini Mental State Examination; BMI: body mass index; CC: calf circumference. ^*In the six-minute walk test; ^†Walking distance ≤ 400 m.

Normalized CC cutoff points showed an inverse relationship with age when compared to absolute CC (Table II). CC values normalized by body mass (0.40 cm·kg^-1), height squared (8.52 cm·m^-2) or BMI (1.10 cm·kg^-1·m^-2) also had a different relationship for each age group. All mean CC values were within the 95 % confidence threshold, which suggests high reliability even when replicated for other samples of the same population.

The accuracy of detecting a functional limitation using different CC expressions is shown in table III. The p-values of the regressions originating from CC as normalized by body mass (p = 0.027) and BMI (p = 0.001) demonstrate the likelihood of significantly accounting for the occurrence of functional limitation. In fact, only CC·BMI^-1 presented an acceptable accuracy (AUC > 0.70), as shown in figure 2. There is statistical significance in the chi-quadratic distribution (Wald = 8.72; p = 0.001). From the beta exponent (b = 0.002) of the CC·BMI^-1 variable (Table III) it was possible to calculate the likelihood of decreasing functional limitation (i.e., 0.002 - 1 = -0.998). That is, for every increased tenth of unit (0.1 cm·kg^-1·m^-2) of CC·BMI^-1 there is a reduction in the chances of functional limitation by ≅ 10 %. As an example from our findings (Table III), mean CC·BMI^-1 in young women (1.6 cm·kg^-1·m^-2) compared to older women (1.3 cm·kg^-1·m^-2) has a difference of 0.3 cm·kg^-1·m^-2. This represents a reduction in the chances of functional limitation for young women by around 30 %.

Table III.  Likelihood of the occurrence of a functional limitation* explained by logistic regression from absolute and normalized calf circumference (CC) 

*Walking distance ≤ 400 m in the six-minute walk test; BMI: body mass index; χ²: chi-square; OR: odds ratio.

Figure 2.  Accuracy of absolute and normalized calf circumference (CC) to detect functional limitation (≤ 400 m in six-minute walk test) in older women, represented by the ROC curve and area under the curve (AUC). 

Although the adjustment of CC by body mass (Table III) was significant (χ² = 4.89; p = 0.027), the extremely low odds ratio value (< 0.001) did not sufficiently explain the occurrence of functional limitation. This finding was confirmed when the analysis of the ROC curve for CC·body mass^-1 showed insufficient sensitivity (27 %), specificity (98 %), and AUC (0.63) values. On the other hand, the CC·BMI^-1 cutoff point presented an acceptable AUC (0.73), even though its sensitivity (85 %) was greater than its specificity (54 %).

After applying the cutoff point of the CC·BMI^-1 (≤ 1.10 cm·kg^-1·m^-2) in our sample, about 12 % of older women were classified with low SMM (n = 7/60). Figure 3 further illustrates the relationship between CC and mobility in older women, as expressed by the distance covered in the six-minute walk test. The inverted U-shaped relationship expressed by absolute CC (a) versus the resulting linearity when normalized by BMI (b) may be seen.

Figure 3.  Comparison of the inverted U-shaped relationship (A) between absolute CC and mobility (six-minute walk test) with the linear (B) one of the indicator proposed in this study (CC·BMI^-1) in older women (n = 67). 

Our model (CC·BMI^-1), in addition to correcting the non-linear relationship between CC and mobility, showed a positive and significant correlation between adjusted CC and mobility (r = 0.48; p < 0.001).

DISCUSSION

Our aim was to develop cutoff points for CC normalized by body size to identify low SMM in older women with respect to functional limitation. The cutoff points of CC when normalized by body size were more effective than absolute CC in identifying low SMM in older women. CC·body mass^-1 and CC·BMI^-1 explained (p < 0.05) the likelihood of occurrence of functional limitation (walking ≤ 400 m in six minutes). But only CC normalized by BMI (1.10 cm·kg^-1·m^-2) achieved an acceptable sensitivity (85 %) and accuracy (AUC > 0.70) (²⁸) to identify the low SMM associated with functional limitation. Furthermore, this model (CC·BMI^-1) was able to linearize the inverted U-shaped relationship usually observed in absolute CC expressions.

Absolute CC is accurate to estimate and identify a low level of SMM in older women (¹⁵); however, when normalized by body size, particularly BMI (²), it avoids false negatives for functional limitation and sarcopenia. An absolute CC value below 27 cm alerts to care needs for older women to perform their daily activities (²⁹). However, when absolute CC is high (i.e., > 38 cm), it also predicts the risk of impaired mobility (¹⁰), suggesting the need to create double cutoff points for the absolute CC (one for high values and another for low values), due to a non-linear relationship between absolute CC and mobility (¹⁰). This was confirmed in this study when all of our older women with elevated CC (> 38 cm) had functional limitations (walking distance ≤ 400 m), as seen in figure 3A.

A multiple linear regression model proposed from NHANES data (³⁰), including CC in combination with sex, race, and age, was able to predict up to 90 % of ASMM as measured by DXA. This approach could be adequate to identify SMM deficit in older women were it not for its weak relationship with functional limitation (⁸). Absolute CC values could result in prediction errors (Fig. 3A) over time, since higher CCs are generally related to overweight and obesity (BMI > 25 kg·m^-2) characterizing a negative impact on functional performance (¹⁰).

One of the strengths of this study involves a normalization of CC that had not yet been proposed as predictor of SMM, although it has long been recommended (²). Another positive finding was that alternatively to most studies that used ASMM·height^-2 derived from DXA as an SMM indicator, our model was able to discriminate functional limitation using normalized CC in an linear expression approach. The procedure to generate a cutoff point at -2 SD of the mean value for young women is a recommended method to approach SMM parameters (²). This study also has its limitations: for example, the sample was not random or stratified. Another limitation is that muscle function and quality were not analyzed in the present study. These limitations, as well as others such as perceptual-motor factors, affect the relationship between CC and performance, and thus can be aims for future studies. Nevertheless, the application of the model to other populations has an unconfirmed predictive value and requires further development before being adopted into clinical practice.

One implication of our findings for clinical practice supports the use of normalized CC by BMI as a simple and inexpensive indicator for monitoring SMM losses associated with functional capacity in older women. Another implication involves choosing variables that predict incident adverse health-related outcomes in older adults to identify potential sarcopenia by CC, allowing intervention or prophylactic decisions to be made when monitoring the health of older women. Our CC·BMI^-1 model classified 12 % of older women with reduced SMM, and 39 % of older women as having functional limitation (walking ≤ 400 m). The cutoff point at -2 SD of mean CC·BMI^-1 for young women (≤ 1.10 cm·kg^-1·m^-2) was lower than the cutoff point based on functional limitation (≤ 1.33 cm·kg^-1·m^-2) as estimated by the ROC curve (data not shown), which would classify older women with mobility problems. This corrects the distortion of overestimating functional limitation (39 %) present in the current classification of SMM deficit (12 %), closer to the worldwide prevalence of sarcopenia (≅ 10 %) (³) but still below the prevalence of Brazilian older women (≅ 20 %) (⁴). But we emphasize that we did not consider muscle strength here, the first criterion for the diagnosis of sarcopenia, but only the second criterion (low SMM) (²). Maybe this should be recommended for further researches.

In conclusion, our hypothesis that normalized CC helps to identify functional limitation as a prognostic tool to estimate the risk of sarcopenia has been confirmed. Our findings support the use of CC normalized by BMI to identify low SMM as an expression of functional limitation in older women, without the inverted U-shaped relationship bias. Monitoring the functional limitations resulting from ageing in women may be feasible using this simple and inexpensive approach, particularly given the high prevalence of sarcopenia in older women in low- to middle-income countries. Its use can be a viable alternative in clinical practice when sophisticated methods for identifying the low SMM associated with functional limitation are not available.