Role of Skinfold Thickness and MUAC in Predicting Insulin Resistance: ROC And Regression Analysis

ABSTRACT

Introduction

Early detection of insulin resistance (IR) in children is essential to prevent long‐term metabolic complications. Simple anthropometric measures—mid‐upper arm circumference (MUAC) and skinfold thickness—are widely used, but their predictive value, especially when age‐standardized, remains unclear. This study aimed to evaluate the performance of raw, SDS, and percentile MUAC and skinfold measurements in predicting IR in patients with obesity and adolescents.

Methods

A cross‐sectional study was conducted with 171 participants with obesity aged 6–18 years. Anthropometric measurements were recorded and converted to SDS and percentile values. Insulin resistance was evaluated using HOMA‐IR with pubertal‐stage–adjusted cutoffs. Logistic regression and ROC analyses were applied.

Results

Insulin resistance was present in 56.1% of participants. Both MUAC and skinfold thickness showed significant positive associations with IR; however, standardized measures (SDS and percentiles) demonstrated markedly stronger predictive value than raw measurements. Skinfold thickness SDS emerged as the strongest independent predictor of IR (OR = 12.04, p < 0.001), followed by MUAC SDS (OR = 7.28, p < 0.001). ROC analysis showed excellent discriminatory power for skinfold SDS (AUC = 0.82), with an optimal cutoff of 1.48 (76% sensitivity, 80% specificity). MUAC SDS also performed well (AUC = 0.76; cutoff = 1.10).

Conclusion

Skinfold thickness SDS and MUAC SDS are strong, practical, and non‐invasive markers for predicting insulin resistance in paediatric patients with obesity. Standardized measurements clearly outperform raw values and should be integrated into routine clinical assessment to enhance early detection of metabolic risk.

1. Introduction

Childhood obesity has emerged as a significant public health concern, increasing rapidly on a global scale [1]. According to a global survey, approximately 22.2% of children and adolescents aged 5 to 19 are affected by excess weight, including overweight and obesity, with this rate having shown a significant increase in recent years [2]. Obesity during childhood is not merely a cosmetic issue; it is strongly associated with serious health conditions such as insulin resistance, metabolic syndrome, type 2 diabetes, and cardiovascular diseases [3].

Insulin resistance (IR) is a metabolic disorder characterized by a diminished cellular response to insulin, leading to impaired glucose uptake. When present during childhood, IR significantly increases the likelihood of developing type 2 diabetes and other metabolic diseases later in life [3]. Therefore, utilizing practical and cost‐effective methods to predict IR at an early stage in children is of critical importance for preventive healthcare [4].

In this context, anthropometric measurements used to assess body composition and fat distribution have gained prominence. Among these, mid‐upper arm circumference (MUAC) and skinfold thickness measurements are frequently preferred due to their ease of application and their ability to reflect adipose tissue [5, 6]. Several studies have reported that these anthropometric parameters are significantly associated with IR and metabolic risk indicators in children [7]. Recent research has suggested that increased skinfold thickness and arm circumference in children may be related to IR and metabolic disturbances [8, 9]. However, the degree and clinical relevance of this association, particularly among patients with obesity, have not yet been clearly established.

Accordingly, this study aims to evaluate the association between MUAC, skinfold thickness, and IR in patients with obesity. Establishing the predictive value of these anthropometric measures may contribute to the development of simple, accessible tools for early detection of metabolic risk.

2. Methods

2.1. Research Design and Sample

This is a cross‐sectional (descriptive) study conducted to evaluate the relationship between obesity and insulin resistance in children and MUAC and skinfold thickness. The study was conducted between January 2024 and May 2025 among children aged 6–18 years who attended a tertiary paediatric gastroenterology and paediatric endocrinology outpatient clinic.

2.2. Inclusion and Exclusion Criteria

2.2.1. Inclusion Criteria

• Age 6–18 years

• Clinical diagnosis of obesity

• Availability of fasting glucose and insulin values

• Completed MUAC and skinfold thickness measurements

2.2.2. Exclusion Criteria

• Known endocrine or metabolic disorders (e.g., diabetes, thyroid disease)

• Use of corticosteroids or metabolism‐altering medications

• Incomplete anthropometric or laboratory data

2.3. Anthropometric Measurements and Definition of Obesity

Children's height and weight were measured in the morning, on an empty stomach, wearing light clothing, using a precision electronic scale and stadiometer. Body mass index (BMI) was calculated using the formula kg/m². Obesity was classified according to the World Health Organization BMI percentile tables for age and gender, with those with a BMI > the 95th percentile being considered obese. Body mass index, BMI percentiles for age, and standard deviation scores for BMI (BMI SDs) were determined based upon the Centre for Disease Control normative curves [10].

2.4. Assessment of Insulin Resistance

The Homeostasis Model Assessment of Insulin Resistance (HOMA‐IR) index was used to assess insulin resistance. Morning venous blood samples were collected from all participants after a fasting period of at least 8 h.

• HOMA‐IR was calculated as follows:

HOMA‐IR = (Fasting insulin (μU/mL) × Fasting glucose (mg/dL))/405.

According to the literature, a prepubertal HOMA‐IR of ≥ 2.5 and a pubertal HOMA‐IR of > 4 are considered insulin resistant in children [11].

Pubertal development was evaluated using the Tanner staging system [12, 13]. Participants were classified into prepubertal (Tanner stage 1) and pubertal (Tanner stages 2–5) groups based on their secondary sexual characteristics. The assessment was conducted by a trained clinician through physical examination to ensure accurate staging.

2.5. Mid‐Upper Arm Circumference and Skinfold Thickness Measurement

Mid‐upper arm circumference was measured at the acromial process of the scapula and the olecranon process of the ulna. The exact center of these two points was measured in centimetres with a tape measure. Skinfold thickness was measured at the upper arm's posterior aspect, midway between the acromion and olecranon processes over the triceps, using a Harpenden skin‐fold calliper. All anthropometric measurements were performed by a single trained physician using standardized procedures. MUAC percentiles for age, and standard deviation scores for MUAC (MUAC SDs) and skinfold thickness percentiles for age, and standard deviation scores for skinfold thickness were determined based upon the Centre for Disease Control normative curves [14].

2.6. Other Parameters

Patients' fasting lipids were measured for obesity. Hepatic steatosis was assessed using ultrasound, and patients were divided into two groups: those absent and without hepatic steatosis. MAFLD was defined by the presence of hepatic steatosis in addition to at least one of the following conditions: overweight/obesity, type 2 diabetes mellitus, or metabolic dysregulation. Metabolic dysregulation was assessed using anthropometric indices and insulin resistance (HOMA‐IR), in accordance with established consensus criteria [15].

2.7. Statistical Analysis

Statistical analyses were performed using SPSS version 16.0 (SPSS Inc., Chicago, IL, USA). Descriptive statistics were presented as means ± standard deviations for continuous variables and as frequencies and percentages for categorical variables. The normality of the data distribution was assessed prior to further analyses. Comparisons between groups were conducted using the Student's t‐test for normally distributed variables and the Chi‐square test for categorical variables. Intra‐observer reliability was assessed using intraclass correlation coefficients (ICC). Satistical normality was assessed using the Shapiro–Wilk test. As the data were not normally distributed, non‐parametric methods were used for analysis. Associations between variables were evaluated using Spearman's rank correlation coefficient. A two‐tailed p < 0.05 was considered statistically significant. Logistic regression analysis was performed to identify independent predictors of metabolic outcomes. Receiver Operating Characteristic (ROC) curve analysis was subsequently applied to evaluate the diagnostic performance of anthropometric measurements in predicting insulin resistance. A p value of less than 0.05 was considered statistically significant.

3. Results

A total of 171 patients with obesity were included, 52% of whom were female. The mean age was 13.6 ± 3.01 years. The distribution of pubertal stages was as follows: stage 1 in 39 patients (22.8%), stage 2 in 13 (7.6%), stage 3 in 17 (9.9%), stage 4 in 19 (11.1%), and stage 5 in 83 (48.5%).

The mean weight SDS was 2.74 ± 0.97, mean height SDS was 1.30 ± 1.39, and mean body mass index (BMI) SDS was 2.57 ± 0.67. The mean MUAC SDS was 1.10 ± 0.49, and the mean skinfold thickness SDS was 1.44 ± 0.57. Laboratory findings are summarized in Table 1.

TABLE 1.

Laboratory findings of patients.

Analyses	Mean ± standard deviation
AST	26.7 ± 15.1
ALT	29.5 ± 27.0
GGT	21.3 ± 14.1
Glucose	90.9 ± 10.3
Insulin	20.0 ± 11,4
HBA1C	5.3 ± 0.2
Total cholesterol	166 ± 33.3
Triglicerides	131 ± 75.5
LDL	96.4 ± 30.6
HDL	43.6 ± 8.9

Abbreviations: ALT: alanine aminotransferase; AST: aspartate aminotransferase; GGT: gamma glutamyl transferase; HDL: High‐density lipoprotein; LDL: Low‐density lipoprotein.

Hepatic steatosis was absent in 46.1% of patients; 29.8% had grade 1, 21.1% had grade 2, and 2.9% had grade 3 steatosis. Insulin resistance was present in 96 patients (56.1%).

The ICC values indicated excellent reliability for MUAC (ICC = 0.91) and good reliability for skinfold thickness (ICC = 0.88).

3.1. Logistic Regression Outcomes

According to logistic regression analysis, a positive and statistically significant relationship between Skinfold thickness and insulin resistance is possible (B = 0.104, OR = 1.110, p < 0.001). The explanatory power of the model is 12.0% (Nagelkerke R ² = 0.120), and the Hosmer‐Lemeshow test shows acceptable model fit with p = 0.063.

The Skinfold thickness SDS is a stronger indicator of this relationship (B = 2.489, OR = 12.044, p < 0.001), and the explanatory power of the model is 37.1% (Nagelkerke R ² = 0.371). The Hosmer‐Lemeshow test (p = 0.464) shows that this model fits the data quite well. Similarly, a higher skinfold thickness percentile is also significantly associated with increased risk (B = 0.127, OR = 1.136, p < 0.001), the explanatory rate is 31.4% (Nagelkerke R ² = 0.314), and the fit of the model is provided with p = 0.311.

It shows a significant association with insulin resistance in the mid‐upper arm circumference (B = 0.107, OR = 1.113, p = 0.011), but the explanatory power of the model is relatively low (Nagelkerke R ² = 0.058). However, the model's fit is acceptable with the Hosmer‐Lemeshow test (p = 0.423).

The MUAC SDS value increases the risk of insulin resistance by 7.28 times (B = 1.985, OR = 7.281, p < 0.001), making this variable a strong predictor. The explanatory power is 22.1% (Nagelkerke R ² = 0.221), and the model's fit is achieved with p = 0.215. The percentile value of the mid‐upper arm circumference is also significant (B = 0.720, OR = 1.075, p < 0.001), the explanatory power is 17.6% (Nagelkerke R ² = 0.176), and the model fit is adequate (p = 0.068) (Table 2).

TABLE 2.

Logistic regression analysis of predictors of insulin resistance.

Variable	B	OR (95% CI)	p	Nagelkerke R 2	Hosmer–lemeshow p
Skinfold thickness	0.104	1.110 (−)	< 0.001	0.120	0.063
Skinfold thickness SDS	2.489	12.044 (−)	< 0.001	0.371	0.464
Skinfold percentile	0.127	1.136 (−)	< 0.001	0.314	0.311
MUAC	0.107	1.113 (−)	0.011	0.058	0.423
MUAC SDS	1.985	7.281 (−)	< 0.001	0.221	0.215
MUAC percentile	0.720	1.075 (−)	< 0.001	0.176	0.068

Abbreviations: CI: confidence interval; MUAC: mid‐upper arm circumference; R: odds ratio; SDS: standard deviation score.

The analysis results reveal significant differences between the measurements of insulin resistance and body fat status and muscle mass. In particular, the SDS and percentile values normalized for age and development appear to have higher odds ratios and better model fit than the raw measurements. This suggests that the use of SDS and percentile values is more appropriate for the evaluation of anthropometric data. The strongest products are found in the skinfold thickness SDS (OR = 12.04) and MUAC SDS (OR = 7.28) variables, respectively.

3.2. ROC Analysis

As a result of the ROC analysis performed to evaluate the performance of the skinfold thickness SDS in distinguishing the presence of insulin resistance, the area under the curve (AUC) was found to be 0.82 and p < 0.001. This value indicates that the skinfold thickness SDS has a good level of discriminatory power in distinguishing insulin resistance. Based on the ROC analysis, the optimal cut‐off value for skinfold thickness SDS to detect insulin resistance was determined as 1.48. At this threshold, the sensitivity and specificity were 76% and 80%, respectively, indicating that skinfold thickness SDS is a reliable marker with good diagnostic accuracy for identifying insulin resistance.

MUAC SDS AUC: 0, 76 and p < 0.00. Cut off.1.10 sensitivity and specificity were 69% and 79%, respectively (Table 3).

TABLE 3.

ROC analysis of anthropometric SDS variables for predicting insulin resistance.

Variable	AUC	95% CI	p	Cut‐off	Sensitivity (%)	Specificity (%)
Skinfold thickness SDS	0.82	—	< 0.001	1.48	76	80
MUAC SDS	0.76	—	< 0.001	1.10	69	79

Abbreviations: AUC: area under the curve; CI: confidence interval; MUAC: mid‐upper arm circumference; SDS: standard deviation score.

3.3. Correlation Between MUAC SDS, Skinfold Thickness SDS and Biochemical Parameters

Spearman's correlation analysis demonstrated a significant positive correlation between MUAC SDS and AST (p = 0.022) and ALT (p = 0.001). However, no significant correlation was observed between MUAC SDS and triglyceride levels (p = 0.312).

A significant correlation was found between skinfold thickness SDS and AST (p < 0.001) as well as ALT (p = 0.003). In contrast, no significant correlation was observed between skinfold thickness SDS and triglyceride levels (p = 0.15).

Overall, anthropometric SDS measures (MUAC SDS and skinfold thickness SDS) showed significant associations with liver transaminases, whereas no significant relationship was observed with triglyceride levels.

3.4. Hepatic Steatosis

When patients were categorized based on the presence or absence of hepatic steatosis, no significant associations were found with skinfold thickness, MUAC, MUAC SDS, or MUAC percentile. However, significant relationships were observed between hepatic steatosis and both skinfold thickness SDS and skinfold thickness percentile (p = 0.008, p < 0.001, respectively). Furthermore, a significant association was identified between hepatic steatosis and insulin resistance (p < 0.001).

4. Discussion

Obesity is associated with serious complications such as type 2 diabetes mellitus, hypertension, and hepatic steatosis. Early detection and prevention of these complications are crucial. This study aimed to predict insulin resistance using a non‐invasive method. Mid‐upper arm circumference was found to be positively correlated with insulin resistance. Furthermore, skinfold thickness SDS, skinfold thickness percentile, MUAC SDS, and MUAC percentile were all positively associated with insulin resistance and served as reliable predictors. Notably, a skinfold thickness cutoff value of 1.48 yielded a sensitivity of 76% and a specificity of 80% for detecting insulin resistance.

Mid‐upper arm circumference is a valuable tool for assessing body composition. Originally, MUAC was primarily used to detect malnutrition due to its simplicity and ease of measurement [16]. Subsequently, numerous studies have demonstrated that MUAC is a suitable tool for assessing obesity and predicting its related outcomes [17]. For instance, a large study involving 6287 Chinese adults reported that a higher MUAC values were associated with an increased risk of several cardiometabolic disorders and subclinical atherosclerosis [18]. Similarly, Zhu et al. demonstrated that MUAC is a simple and effective tool for identifying central obesity and insulin resistance [19]. In addition, a study conducted on 93 pubertal adolescents with obesity in Brazil found that higher HOMA‐IR levels and increased cardiometabolic risk scores were associated with larger MUAC measurements [20]. Conversely, a study from rural Bangladesh found no association between MUAC and carotid intima‐media thickness, although a significant relationship with insulin resistance was still observed [21]. In line with these findings, our study also demonstrated a significant association between MUAC and insulin resistance. Notably, MUAC SDS and percentile values showed stronger predictive performance compared to raw MUAC measurements. Using a MUAC SDS cut‐off value of 1.10, sensitivity and specificity were 69% and 79%, respectively. These results suggest that standardized anthropometric measures may provide a more accurate assessment of metabolic risk. One of the key findings of our study is that SDS‐ and percentile‐based evaluations outperform raw measurements, supporting their use in paediatric populations where age‐ and development‐adjusted indices are essential.

Various calculation methods, including skinfold thickness measurements, have been developed to assess body composition. Although body mass index (BMI) is widely used to evaluate obesity, it does not adequately reflect body fat distribution, particularly central adiposity, which is more strongly associated with metabolic risk. In contrast, skinfold thickness provides a more direct estimation of body fat composition [22, 23]. Previous studies have demonstrated the clinical relevance of skinfold measurements in predicting metabolic risk. For example, Ali et al. found that the accumulation of subcutaneous adiposity was a better indicator of cardiometabolic risk factors than visceral fat and served as a strong predictor of insulin resistance and hypertriglyceridemia in children and adolescents [24]. Similarly, Soundararajan et al. demonstrated that skinfold thickness serves as an early indicator of metabolic syndrome in overweight and adolescents with obesity, making it a valuable tool for assessing body composition and related health risks [9]. Andaki et al. aimed to establish skinfold thickness cutoff points for predicting the risk of metabolic syndrome in children aged 6 to 10 years. They demonstrated that skinfold thickness measurements are reliable indicators of metabolic syndrome in this age group. Furthermore, age‐ and gender‐specific smoothed percentile curves for skinfold thickness serve as useful tools for identifying children at risk for metabolic syndrome [25]. In addition, Addo et al. investigated the association between adiposity measured through triceps and subscapular skinfold thickness and total body fat quantified by dual‐energy X‐ray absorptiometry (DXA) in relation to HOMA insulin resistance. Their results indicated that skinfold measurements at both sites reliably estimate HOMA‐IR and effectively identify individuals at high risk for insulin resistance, with accuracy comparable to total body fat determined by DXA. Due to its affordability and widespread applicability, skinfold thickness is a valuable tool for assessing adiposity in research focused on insulin resistance and other metabolic parameters in adolescents [7]. In line with these findings, our study demonstrated that skinfold thickness is significantly associated with insulin resistance. Notably, standardized measures such as skinfold thickness SDS and percentile values showed stronger predictive performance compared to raw measurements. As these values increased, the likelihood of insulin resistance also increased. ROC analysis further supported these findings, with a skinfold thickness SDS cut‐off value of 1.48 yielding a sensitivity of 76% and specificity of 80%. These results highlight the clinical utility of skinfold thickness, particularly when expressed as age‐ and development‐adjusted indices. One of the key findings of our study is that SDS‐ and percentile‐based evaluations provide superior diagnostic performance, suggesting that standardized anthropometric measures may be more appropriate for assessing metabolic risk in paediatric populations.

Hepatic steatosis, particularly as a common complication of obesity and metabolic syndrome, is closely associated with insulin resistance and increased cardiometabolic risk [26]. Therefore, simple anthropometric measurements have gained attention as potential tools for early detection and risk stratification. Recent studies have reported a positive association between MUAC and hepatic steatosis in individuals with MAFLD. For example, higher MUAC values have been shown to correlate with liver fat accumulation and fibrosis, suggesting its utility as a simple and non‐invasive marker of liver disease severity [27]. Similarly, another study in children and adolescents demonstrated a significant positive association between MUAC and both the presence of MAFLD and the severity of hepatic steatosis, with a threshold effect observed at specific MUAC values [28]. In addition, longitudinal cohort data have shown that increasing adiposity trajectories, including mid‐upper arm circumference and skinfold thickness from early childhood, are associated with the presence and severity of hepatic steatosis in adolescence [29]. In contrast to these findings, our study did not demonstrate a significant association between hepatic steatosis and MUAC, MUAC SDS, or MUAC percentiles. This discrepancy may be explained by differences in study design, population characteristics, and analytical approaches. Notably, our study used a binary classification of hepatic steatosis, which may have limited the ability to detect associations with disease severity. In addition, MUAC reflects not only adipose tissue but also muscle mass, which may reduce its specificity as a marker of fat accumulation.

Although previous studies have primarily focused on MUAC, limited data exist regarding the relationship between skinfold thickness and hepatic steatosis. In our study, no significant association was observed between hepatic steatosis and raw skinfold thickness measurements. However, skinfold thickness SDS and percentile values showed a significant association with hepatic steatosis. This suggests that standardized, age‐adjusted measures may be more sensitive indicators of adiposity‐related risk in paediatric populations. Taken together, these findings indicate that while MUAC may provide a general assessment of body composition, skinfold thickness—particularly when expressed as SDS or percentiles—may offer a more precise evaluation of fat accumulation and its metabolic consequences.

Importantly, the present study adds to the existing literature by demonstrating that standardized anthropometric measures, particularly skinfold thickness SDS, provide superior predictive performance for insulin resistance and hepatic steatosis compared to raw measurements in paediatric populations.

This study has several limitations that should be acknowledged. First, its cross‐sectional design precludes establishing a causal relationship between anthropometric measurements and insulin resistance. Second, the study was conducted in a single tertiary care center, which may limit the generalizability of the findings to broader paediatric populations. In addition, the relatively small sample size may reduce the statistical power of the study. Therefore, larger, multicenter, and longitudinal studies are needed to confirm and extend these findings. Third, insulin resistance was assessed using HOMA‐IR rather than the hyperinsulinemic–euglycemic clamp, which is considered the gold standard, although HOMA‐IR is widely accepted for clinical and epidemiological studies. Additionally, body composition was evaluated using anthropometric measures without direct imaging‐based assessment of fat distribution. Despite these limitations, the use of standardized SDS and percentile values and the relatively large sample size strengthen the clinical relevance of the findings. A further limitation of this study is the small number of participants with severe obesity (BMI > +3 SDS), which precluded a reliable subgroup analysis and limits the ability to draw robust conclusions regarding the accuracy of anthropometric measurements in this high‐risk subgroup where altered body fat distribution may affect measurement validity. The assessment of hepatic steatosis in this study was based on ultrasonography (USG), which is not considered the gold standard method. Although USG is widely used in clinical practice due to its non‐invasive nature and accessibility, it has limited sensitivity, particularly in detecting mild steatosis. Therefore, some cases may have been underdiagnosed. More accurate modalities such as magnetic resonance imaging or liver biopsy were not utilized, which should be considered a limitation of the present study. The lack of waist circumference and waist‐to‐height ratio measurements may be considered a limitation, as these parameters are important indicators of central adiposity.

In conclusion, our study highlights that skinfold thickness SDS and percentiles are reliable indicators for predicting insulin resistance in children and adolescents, while MUAC alone may not fully capture adipose tissue accumulation. The lack of association between raw MUAC values and hepatic steatosis further emphasizes the need for standardized, age‐adjusted anthropometric measurements in paediatric populations. These findings support the integration of skinfold thickness assessments alongside MUAC in clinical settings to improve early detection of metabolic risks. Further research is needed to confirm these associations and to explore their potential for predicting long‐term metabolic outcomes.

Funding

The authors have nothing to report.

Ethics Statement

The methodology for this study was approved by the Ankara Training and Research Hospital Human Research Ethics Committee (number: 738/2025).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: jpc70456‐sup‐0001‐Supinfo.docx.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.