INTRODUCTION

Skinfold thickness is often used for body composition assessment and widely used in the clinical practice due to its accessibility, non-invasive features and the ability to measure subcutaneous adiposity ^{1) (2}. In skinfold thickness measurement with a skinfold calliper, constant pressure is applied for a defined period of time ^{3) (4}. The tissue's dynamic response to this pressure is defined as compressibility ^{1) (5}. This characteristic has been studied by comparing skinfold calliper measurements and subcutaneous fat thickness assessed by coarse methods such as imaging methods, cadaver studies and empiric comparisons ^{1) (5}.

There are underlying suppositions on the estimation from skinfold measurement: skin thickness is negligible, adipose tissue has constant characteristics, and also that proportion of subcutaneous to visceral fat is equivalent in all subjects ¹. Notwithstanding this, it has been previously shown that compressibility varies according to the sites of measurement and between individuals, influencing the relation between the measurement and the actual adipose tissue thickness, introducing error in the estimation of body fatness ^{1) (5}.

Gender ⁵, age ⁶, hydration status ⁶, skin thickness ⁷, subcutaneous tissue pressure ⁷ and site of measurement ⁸ have been previously described as factors associated with compressibility. Nevertheless, over the past few years, knowledge in regards to compressibility has not increased significantly.

An integrated system, Lipotool^(r), was recently developed. This equipment consists of a digital skinfold calliper and a software application. This system registers 60 measurements per second ⁹. Thus, this novel methodology firstly permits the study of dynamic tissue's response evolution during the measurement ⁹.

From all skinfold thickness sites, triceps skinfold (TSF) is the most widely used in clinical practice, as, along with mid-arm circumference, it integrates mid-arm muscle circumference formula, a simple method that allows for the estimation of muscle mass ¹⁰.

Regarding the wide use of TSF, the minimization of error is of utmost importance in order to provide an adequate use and interpretation for clinical practice. Nonetheless, as far as we are concerned, skinfolds compressibility has not been explored yet in a clinical setting. Therefore, the present study aims to explore, through an innovative technique, TSF compressibility and its associated factors among a sample of hospitalized patients.

MATERIALS AND METHODS

ETHICS

This research was carried out according to the recommendations established by the Declaration of Helsinki and approved by institutional ethics and review boards. All study participants provided a written informed consent.

DATA COLLECTION

Demographical data were obtained by a trained registered nutritionist through a structured questionnaire within 72 hours of admission to hospital.

Education was evaluated by the number of completed school years and the following categories were created: 0-4, 5-12 and over 12 years. Marital status was categorized as single, married or in a civil partnership, divorced and widowed. Independence in activities of daily living was assessed with the Katz index ¹².

Patients' nutritional status was evaluated with Nutritional Risk Screening (NRS) 2002 ¹³. Standing height (cm) was measured with a metal tape (Rosscraft, Innovations Incorporated, Surrey, Canada) with a 0.1 cm resolution and with a headboard. Body weight (kg) was assessed with a calibrated portable beam scale with 0.5 kg resolution.

Triceps skinfold thickness (mm) was obtained with the Lipotool^(r) digital calliper after performing the measurement during two seconds, as established by the International Society for the Advancement of Kinanthropometry (ISAK) protocol ^3).

All measurements were performed by the same trained registered nutritionist. Intra-observer error ranged from 0.2% to 1.8%. These values are considered as acceptable for a trained anthropometrist ^{14) (15}.

Body mass index (BMI) was determined through the standard formula (weight [kg]/height² [m]) ¹⁶, and BMI categories were created according to the World Health Organization cutoffs ¹⁷.

STATISTICS

Results were described as mean and standard deviation (SD) or as median and interquartile range (IQR) according to normality of distribution, assessed with the Kolmogorov-Smirnov test.

Data on TSF measurements were provided by LipoTool^(r) software and the evolution of tissue compressibility during two seconds was registered, as this method registers 60 values per second. Thus, at the end of the measurement, 120 values were obtained.

Compressibility was determined according to a method based on the difference computed between the initial value and the final value, from the 120 TSF measurements acquired by the digital caliper ¹⁸. Thus, high difference between the initial and final TSF value corresponds to high compressibility ¹⁸.

Another method was used to define compressibility. This method was based on τ, tau, a measurement of time expressed in seconds, that reflects adipose tissue dynamic response to compression, being an individual characteristic ¹⁹. Thus, lower τ values mean that the skinfold compresses faster, and, therefore, presents higher compressibility. τ value was obtained after computing the inverse of the exponent of a regression equation displayed for the 120 measurement sets of each patient ^{19) (20}.

Data set was divided into tertiles of TSF, tertiles of τ and tertiles of difference between the TSF initial and final values (TSF difference). In order to select variables associated to compressibility, patients' baseline characteristics were compared across τ tertiles and TSF difference tertiles. Patients' baseline characteristics were also compared across TSF tertiles.

All the comparisons were computed by the one-way ANOVA test if distribution was normal, or by the Kruskal-Wallis test in case of non-normal distribution. Categorical variables were reported as frequencies. Differences between proportions were assessed with the Pearson's ² test or Fisher's exact test.

Furthermore, multivariable linear regression models were built in order to identify the independent variables associated with compressibility, assessed by τ or as TSF difference. The following variables were included in the models: TSF value (continuous), age (continuous), nutritional status (categorical; normal nutritional status used as reference) and gender (categorical; women used as reference), as these variables were considered to be potential confounders or covariates.

Statistical significance was set at p < 0.05. All analyses were conducted with MATLAB (MathWorks, Inc., Natick, MA) and the Software Package for Social Sciences (SPSS) for Windows (version 20.0; SPSS, Inc., Chicago, IL).

RESULTS

The characteristics of the 106 patients enrolled in the present study are displayed in table I, for the entire sample and stratified by TSF tertiles. Mean age (SD) was 53.1 ^{15) (8}, and 30.2% patients were aged ≥ 65 years. There were 56.6% overweight or obese patients and 11.3% patients were at undernutrition risk (Table I). The highest and the lowest time of compressibility (τ) were observed for patients in the 2^nd TSF tertile and in the 1^st TSF tertile, respectively. The highest TSF difference was observed for patients in the 3^rd TSF tertile (Table I).

Table I Patients' characteristics for the entire sample and according to triceps skinfold tertiles (mm) 

TSF: Triceps skinfold; SD: Standard deviation; IQR: Interquartile range; BMI: Body mass index; NRS-2002: Nutritional Risk Screening-2002. *One-way ANOVA. ^†Pearson Chi-square test or Fisher's exact test. ^‡Kruskal-Wallis test. ^¶Triceps skinfold difference: Initial value - Final value, across a set of 120 measurements.

As shown in table II, BMI, TSF thickness and τ value increased from the 1^st to the 3^rd TSF difference tertiles. Otherwise, patients' characteristics did not differ across τ tertiles, with the exception of the TSF difference, which was higher in the 2^nd and 3^rd τ tertiles than in the 1^st τ tertile (Table III).

Table II Patients' characteristics according to triceps skinfold difference *tertiles of sample distribution (mm) 

TSF: Triceps skinfold; SD: Standard deviaton; IQR: Interquartile range; BMI: Body mass index; NRS-2002: Nutritional Risk Screening-2002. *Triceps skinfold difference: Initial value - Final value, across a set of 120 measurements. ^†One-way ANOVA. ^‡Pearson Chi-square test or Fisher's exact test. ^¶Kruskall-Walis test.

Table III Patients' characteristics according to time of compressibility (τ) tertiles of sample distribution (second) 

TSF: Triceps skinfold; SD: Standard deviation; IQR: Interquartile range; BMI: Body mass index; NRS-2002: Nutritional Risk Screening-2002. *One-way ANOVA. ^†Pearson Chi-square test or Fisher's exact test. ^‡Kruskall-Walis test. ^¶Triceps skinfold difference: Initial value - Final value, across a set of 120 measurements.

Results from the multivariable linear regression models are presented in table IV. As shown in model 1, compressibility based on the TSF difference was associated with TSF magnitude (regression coefficient = 0.38 [0.01-0.05], p = 0.002) and nutritional status (regression coefficient = 0.23 (0.12-1.23), p = 0.018), after adjusting for age and gender. Thus, presenting a higher TSF value, i.e., a thicker TSF, and being at risk of undernutrition are factors apparently related to an increase in the difference between TSF initial and final values, meaning that the skinfold was more compressed and, therefore, presents higher compressibility.

In contrast, as displayed in model 2, time of compressibility (τ) was not significantly associated with any of the included variables.

Table IV Multivariable linear regression models for prediction of triceps skinfold (TSF) compressibility 

CI: Confidence interval; TSF: Triceps skinfold thickness (mm); NRS-2002: Nutrition Risk Screening 2002. Variables included: Age (years; continuous), nutrition status according to NRS-2002 (normal used as reference), gender (women used as reference) and TSF value (mm; continuous). *Dependent variable: TSF compressibility computed as TSF initial value - TSF final value, across a set 120 measurements. ^†Dependent variable: TSF compressibility defined as time (τ).

DISCUSSION

The present study results show that quantification of compressibility and its associated factors is dependent on the method used to analyze this adipose tissue feature.

When compressibility was defined as the difference between initial and final TSF values obtained by the digital calliper, only BMI and time of compressibility differed between the tertiles of this variable. Nevertheless, after adjustment for potential confounders, such as gender and age, results from the multivariable linear regression model showed that undernutrition risk and higher TSF thickness were factors associated with higher compressibility.

In contrast, when compressibility was defined as time, i.e., the time taken by the adipose tissue to respond to the pressure exerted by the calliper, differences were observed between TSF thickness tertiles, which could indicate that the skinfold magnitude was associated with compressibility. In addition, it is worth noting that patients in the 2^nd TSF tertile presented higher τ than patients in the 3^rd TSF tertile. However, after performing a multivariable linear regression model, no independent association was found for any of the included variables, showing that, apparently, time of compressibility was not influenced by any of the studied factors.

Explanations for these associations can be formulated, although only in a theoretical perspective as, with the present data, it is not possible to confirm them. Therefore, a thicker TSF presents a larger area of adipose tissue, and this increases the potential of being compressed. Otherwise, an individual classified as being at risk of undernutrition is potentially likely to present more laxity in skin and adipose tissue, which can influence skinfolds compressibility towards higher values.

Transposing the present results for clinical practice, TSF thickness and undernutrition risk are characteristics susceptible of affecting the association between the actual value and the calliper reading, potentially introducing error by an increase in adipose tissue compressibility. Thus, by causing more compression in the skinfold, this error can lead to an underestimation of TSF thickness, i.e., to a lower value reading and, therefore, to a misinterpretation of the measurement.

Moreover, once τ indicates skinfolds compressibility, as time of response to a constant pressure, a higher time of response is expected to be associated to lower compressibility, as the tissues compress slowly. In contrast, a higher difference between the initial and final TSF values means that the tissue went through more compression, and is, therefore, associated with higher compressibility. Notwithstanding this, our results show that τ and TSF difference vary in the same direction, as τ values are higher in TSF difference 2^nd and 3^rd tertiles and TSF difference values are higher in the 2^nd and 3^rd tertiles of τ.

Considering the aforementioned methods for evaluating compressibility and the results actually obtained, there is an apparent counterintuitive observation. Nevertheless, it is worth noting that these two methods are related with two different aspects of compressibility, time of response and the skinfold dimension. Thus, a skinfold that takes more time to be compressed is, therefore, less compressible according to this definition. It may also simultaneously present a higher difference between the value at the beginning of the measurement and the value attained when the process is complete.

Although this novel methodology has been previously used in other settings ²¹, as far as we are concerned, this is the first report on the exploration of TSF compressibility as a quantifiable variable and its associated factors in a clinical setting. Consequently, there are no previous results to which our findings can be compared. Even though one approach detected consistent associations and the other one did not, they cannot be compared in terms of accuracy as these two methods assess different features.

The absence of association with other factors found for compressibility defined as time is not sufficient to conclude that there are no differences or even that compressibility did not affect measurements performed in the present sample. We can further hypothesize that, in the two seconds the measurement is performed, τ may be related to an earlier moment of the process than the TSF difference. Thus, it is not known if in a larger period of measurement these results could be different.

Present results concern TSF only. As it has been already documented through results from studies ^5),(8),(22 using different methodologies, adipose tissue compressibility varies according to the site of measurement. Thus, it is not known whether these results would be different if other skinfolds were evaluated.

In order to comply with the inclusion criteria, no critically ill or functionally impaired patients were enrolled. Moreover, the majority of the participants were independent in activities of daily living and there was a small proportion of patients at nutritional risk. Thus, the present sample can be considered as homogenous and this feature may have influenced the results obtained. Therefore, it is not known if present results would be different in a wider sample of hospitalized patients or, even, among critically ill or bedridden patients.

In the future, it would be important to further explore compressibility through the present methodologies in other settings, such as in community-dwelling adults and older adults and different ethnic groups. The application of the present methods in different settings could allow for both testing their reproducibility and improving the techniques used.

In conclusion, among a sample of hospitalized patients, undernutrition risk and higher TSF thickness were factors independently associated with higher compressibility assessed by the difference between the initial and final TSF values. Time of compressibility (τ) was not affected by any of the studied factors. Although the present study is merely an exploratory attempt to describe compressibility and its effects, our results emphasize the need for further research in order to determine the most accurate method to quantify compressibility, to infer on the associated factors and to control its effect.