The skinfold method, the measurement of subcutaneous fat folds, is the most widely adopted field method for the assessment of body fat, especially in children.
It is based on the principle that fat is of a known density and by “summing” measurements of subcutaneous fat thickness across the body, total and regional fat can be estimated.
Skinfold thickness measurements are typically used to rank individuals in terms of relative total “fatness”, or to assess subcutaneous fat at various regions of the body.
Population specific equations are used to derive estimates of percent body fat.
In infancy, it might be the sole tool available for assessing body composition longitudinally as other methods may not be feasible, or may only be suitable for use at body sizes e.g. PEA POD, can only measure infants up to 10kg.
The skinfold method involves measuring the skinfold (subcutaneous fat) thickness at specific sites of the body using a skinfold caliper and a nonstretchable measuring tape to correctly locate the measurement area.
Equipment
Caliper
The cost of calipers ranges from £9 to approximately £300. For research purposes, calipers with a more refined scale (e.g. 0.1 mm intervals) and constant pressure of 10 g/cm^{3} between the jaws are desirable. Examples include the Holtain (see Figure 1), Lange and Harpenden calipers (see instrument library for more details). The Lange and Harpenden calipers have been used in developing prediction equations and reference values (Lee 1996 [20]). The Lange is most popular in the US, and the Harpenden and Holtain in Europe.
Figure 1 Example of skinfold caliper typically used in children and infants.
Source: https://holtain.co.uk/tw.php
Measuring tape
Typically a nonstretch fibreglass or plastic measuring tape (such as those used in circumference measurements) is used to locate the anatomical midpoints on the body where the skinfold measurement is taken.
Protocol
Skinfold measurement can be obtained from 2 to 9 different standard anatomical sites around the body using a caliper, as shown in Figure 2. The subscapular and triceps skinfolds are the most commonly used.
Figure 2 Anatomical sites for skinfold thickness measurement taken at the left side.
Source: MRC Epidemiology Unit.
The following are the nine anatomical sites (as illustrated in Figure 2) that are most commonly used in the assessment of skinfold thickness:
Figure 3 Quadriceps skinfold thickness in an infant to the left and triceps skinfold thickness in an adult to the right.
Source: MRC Epidemiology Unit.
An example of a calibration block with known thicknesses Figure 4 is used to calibrate skinfold calipers. Typically, calibrations are carried out on a monthly basis.
Figure 4 An example of a calibration block.
Source: MRC Epidemiology Unit.
Skinfold thickness are typically recorded in mm. Some calipers record in both mm and cm. The skinfold thickness values should be quality checked during data processing in the same manner as other health related variables, for example by checking for outliers and data entry errors.
Raw skinfold thickness values are often used and they act as reliable indicators of regional fatness. In a similar way to body mass index (BMI), they can be converted into standard deviation scores (SDS) for longitudinal evaluations.
The triceps site is the most commonly used singlesite skinfold measurement as it is easy to measure and reference data (e.g. WHO triceps skinfold thickness for age) are available for comparison. However, no equations are available for estimating body fat from a singlesite skinfold measurement. Triceps measurement is also used to derive indices of body composition using arm anthropometry.
To convert raw skinfold thickness values into a percent of body fat, populationspecific or generalised equations are used. These equations are derived from empirical relationships between skinfold thickness and body density. Many equations firstly calculate body density and require an additional calculation to estimate percent body fat. The Brozek et al (1963) and the Siri (1961) equations can be used for this step:
Body fat values should be generated from published equations which closely match the study population. It is critical that the equation selected for estimating body fat is appropriate to the demographics of the cohort under investigation (e.g. race, age, and gender).
Several equations are available. The most commonly used equations are listed below:
Adults
Jackson & Pollock (1985)
% Body Fat = (0.29288 * sum of skinfolds) – (0.0005 * square of the sum of skinfolds) + (0.15845 * age) – 5.76377, where the skinfold sites (mm) are abdominal, triceps, thigh and suprailiac
% Body Fat = (0.29669 * sum of skinfolds) – (0.00043 * square of the sum of skinfolds) + (0.02963 * age) + 1.4072, where the skinfold sites (mm) are abdominal, triceps, thigh and suprailiac
% Body Fat = (0.41563 * sum of skinfolds) – (0.00112 * square of the sum of skinfolds) + (0.03661 * age) + 4.03653, where the skinfold sites (mm) are abdominal, triceps and suprailiac
Durnin Womersley (1974)
Durnin Womersley (1974) developed general equations from a heterogeneous group of varying ages. The calculation of body fat % involves measuring 4 skinfold sites, triceps, biceps, subscapular and suprailiac, and substituting the log of their sum into one of the following equations (Table 1), depending on the participant’s age and sex. The density value can then be converted to percent body fat (%BF) using Siri 1961 or Brozek 1963 equations described above.
Table 1 Durnin Womersley equations for the estimation of body density using 4 skinfold sites.
Age (years) 



< 17 


1719 


2029 


3039 


40 49 


> 50 


D = predicted density of the body (g/ml), and L = log of the total of the 4 skinfolds (mm).
Source [14]
Estimates derived using these equations have been compared to those from the criterion 4component model (see Figures 5 and 6). The Durnin and Wormersley (1974) equation showed significant mean difference/bias of 2%, while the Jackson and Pollock (1985) equation showed mean bias of 6.6%. Both equations tend to underestimate body fat especially in larger individuals. Similar results have also been observed in men (Peterson et al., 2003).
Figure 5 BlandAltman plot showing the limits of agreement between percentage body fat calculated with the 4compartmentmodel equation (%BF_{4C}; the reference equation) and percentage body fat calculated with the Durnin and Wormersley
equation (%BF
_{DW}) in women.
Source: Peterson et al. (2003).
Figure 6 BlandAltman plot showing the limits of agreement between percentage body fat calculated with the 4compartmentmodel equation (%BF_{4C}; the reference equation) and percentage body fat calculated with the Jackson and Pollock
equation (%BF
_{JP}) in women.
Source: Peterson et al. (2003) [23].
Children and young adolescents
A priori, age and sexdependent regression equations published by Weststrate and Deurenberg (1989) that were derived from modification of the Siri (1961) equation (% fat = 495/density450), are the most suitable for the estimation of % FM in children and adolescents:
However, Slaughter et al. (1988) equations proposed for prepubertal, pubertal and postpubertal males and females are the most commonly used.
Table 2 lists equations used to determine body composition values in children and adolescents using skinfold measurement.
Table 2 Published equations used to estimate body fat in children and adolescents from skinfolds.
Author(s)  Population  Equation(s) 

Lohman et al. (1984)  Prepubescent children  M and F: Fat (%) = 530 / D − 489 
Weststrate and Deurenberg (1989)

10−18 y (modification of Siri equation)

F: Fat (%) = [553−7.3 (Age−10)] / D − [514−8 (Age−10)] M: Fat (%) = [562−4.2 (Age−2)] / D − [525−4.7 (Age−2)] 
Brook (1971)  1−11 y (predicted from equations for adolescents) 
F: D = 1.2063−0.0999 (LOG sum of 4 skinfolds) M: D = 1.1690−0.0788 (LOG sum of 4 skinfolds) 
Durnin and Rahaman (1967); Durnin and Womersley (1974) 
13−15.9 y 16−19.9 y 
F (13−15.9 y): D = 1.1369−0.0598 (LOG sum of 4 skinfolds) M (13−15.9 y): D = 1.1533−0.0643 (LOG sum of 4 skinfolds) F (16−19.9 y): D = 1.1549−0.0678 (LOG sum of 4 skinfolds) M (16−19.9 y): D = 1.162−0.063 (LOG sum of 4 skinfolds) 
Johnston et al. (1988)  8−14 y 
F: D = 1.144−0.06 (LOG sum of 4 skinfolds) M: D = 1.166−0.07 (LOG sum of 4 skinfolds) 
Deurenberg et al. (1990) 
Pubertal F: 13.1 ± 0.15 y Pubertal M: 13.8 ± 0.21 y Postpubertal F: 16.8 ± 0.36 y Postpubertal F: 17.5 ± 0.39 y 
F pubertal: D = 1.1074 − 0.0504 (LOG sum of 4 skinfolds) + 1.6 (age 10^{3}) M pubertal: D = 1.0555 − 0.0352 (LOG sum of 4 skinfolds) + 3.8 (age 10^{3}) F postpubertal: D = 1.183 − 0.0813 (LOG sum of 4 skinfolds) M postpubertal: D = 1.1324 − 0.0429 (LOG sum of 4 skinfolds) 
Sarría et al. (1998)  11−16.9 y 
M (11−13.9): D = 1.1516 − 0.0658 (LOG sum of 4 skinfolds) M (14−16.9): D = 1.169 − 0.0693 (LOG sum of 4 skinfolds) 
Sloan et al. (1962)  Young women  F: D = 1.0764 − 0.00081 suprai − 0.00088 tric 
Wilmore and Behnke (1970)  Young women  F: D = 1.06234 − 0.00068 subsc − 0.00039 tric − 0.00025 thigh 
Slaughter et al. (1988) 
Prepubertal F: 10.0 ± 1.0 y Prepubertal M: 9.8 ± 1.3 y Pubertal F: 11.4 ± 1.9 y Pubertal M: 12.2 ± 1.4 y Postpubertal F: 15.3 ± 1.6 y Postpubertal M: 15.8 ± 1.6 y 
All F: Fat (%)= 1.33 (tric+subsc) − 0.013 (tric+subsc)^{2} − 2.5 Prepubertal M: Fat (%) =1.21 (tric+subsc) − 0.008 (tric+subsc)^{2} − 1.7 Pubertal M: Fat (%) = 1.21 (tric+subsc) − 0.008 (tric+subsc)^{2} − 3.4 Postpubertal M: Fat (%)= 1.21 (tric+subsc) − 0.008 (tric+subsc)^{2} − 5.5 All F when (tric+subsc) > 35 mm: Fat (%) = 0.546 (tric+subsc) + 9.7 All M when (tric+subsc) > 35 mm: Fat (%) = 0.783 (tric+subsc) + 1.7 F: Fat (%)= 0.61 (tric+calf) +5.1 M: Fat (%)= 0.735 (tric+calf) + 1 
Lean et al. (1996)  18−64.3 y 
F: Fat (%)= 0.730 BMI + 0.548 tric + 0.270 Age − 5.9 M: Fat (%)= 0.742 BMI + 0.95 tric + 0.335 Age − 20 
Bray et al. (2001)  10y 
M and F: Fat (%)=7.66 + 0.22 subsc + 0.21 thigh + 0.64 biceps + 0.31 calf M and F: Fat (%)=8.71 + 0.19 subsc + 0.76 biceps + 0.18 suprai + 0.33 tric 
F: females, M: males, y: years, D: density (kg/l), BMI: body mass index (kg/m^{2}), sum of 4 skinfolds: biceps + triceps+ subscapular + suprailiac (mm), age (years), tric: triceps skinfold (mm), biceps: biceps skinfold (mm), subsc: subscapular
skinfold (mm), suprai: suprailiac skinfold (mm), thigh: thigh skinfold (mm), calf: calf skinfold (mm).
Source: Rodriguez et al. (2005).
Some equations for children and adolescents have been compared with the criterion 4component model, see Table 3. Significant bias for percentage body fat and fat free mass was observed for the equations by Slaughter et al. (1988), Johnston et al. (1988) and Brook (1971). No significant mean bias was shown by the equation by Deurenberg et al. (1990), but the bias was correlated significantly to fatness and limits of agreements were wide indicating that individual values are not accurate. This may affect the evaluation of body composition changes within individuals overtime.
Table 3 Bias and 95% limits of agreement for % body fat predicted using skinfoldthickness equations against measurements made with the 4component model.
Equation 




Slaughter et al. (1988) 



Percentage body fat (%) 



Johnston et al. (1988) 



Percentage body fat (%) 



Deurenberg et al. (1990) 



Percentage body fat (%) 



Brook (1971) 



Percentage body fat (%) 



^{1}Bias was calculated as skinfoldthickness values minus values from use of the 4C model. Correlations were calculated as the correlation between the difference and mean. 95% limits of agreement calculated as ± 2SD of the difference between
techniques. FFM values were log transformed to express the difference as a percentage of the mean. Values for percentage body fat are expressed as a percentage of body weight.
^{2} P < 0.0001.
^{3} P < 0.005.
Adapted from: Wells et al. (1999).
Infancy
There are limited equations to use in infancy to derive % body fat and they tend to be population or age specific (e.g. first 10 days of life) and based on different skinfold thickness measuring sites.
The Deierlein et al. (2012), Catalano et al. (1995) and Aris et al. (2013) equations have been evaluated using air displacement plethysmography (PEABOD) at birth and 3 months, demonstrating significant bias for body fat in the equation by Catalano et al. (1995) at birth, and significant bias for body fat at 3 months for all the equations.
Table 4 Bias and 95% limits of agreement for % body fat predicted using skinfoldthickness equations against measurements made with the 4component model.
Equation 














Birth 








Deierlein 








Catalano 








Aris 








3 months 








Deierlein 








Catalano 








Aris 








*Significance for the correlation of the strength for the relationship between the mean of the criterion and each equation correlated to the difference between the equations estimated infant fat mass and the criterion measured fat mass. A nonsignificant
correlation suggests no bias in the technique across the range of fatness.
Source: Clauble et al. (2016).
Equations derived in children have also been used to estimate % body fat in infancy, such as Deurenberg et al. (1989) and Slaughter et al. (1988). However, the relationship between total body density and skinfold thickness varies with age and those equations may not be applicable in younger groups.
Estimates derived using the Slaughter et al. (1988) equation have been compared to those from air displacement plethysmography (PEABOD) at 6 weeks and at 4.5 months of age, and to those from DEXA at 4 months. Agreement analysis showed significant bias at 6 weeks, underestimating percentage body fat by 2.4–8.9%. At 4.5 months, the underestimation was greater in infants with the highest body fat. The limits of agreement were wide; error ranged from 18% fat below to 9% fat above the PEA POD measurement. The agreement analysis between Slaughter et al. (1988) and DEXA shows a considerable systematic bias which increases with increasing % body fat. The equation overestimated % body fat by 10.7% when compared to the PEAPOD measurement (Lingwood et al., 2012; Schmelze et al., 2002).
Estimates derived from the Deurenberg et al. (1989) equation have been compared to those from DEXA at 4 months of age. The equation overestimated % body fat by 6.27% when compared to the DEXA measurement (Schmelze et al., 2002).
When analysing data in infancy, often the raw thickness data are used. The sum of the thicknesses is determined and internal standard deviation score (Zscore) are derived. Internal Zscores can be generated by regressing skinfolds on age (and using the saved residuals), and then adjusting for sex in the analyses.
Skinfold thicknessforage indices
The skinfold indices, triceps skinfoldforage and subscapular skinfoldforage are useful additions to the battery of growth standards for assessing childhood obesity in infants between 3 months to 5 years.
These indices are expressed in percentiles (percentage of median) and can be assessed by the percentile point achieved by a child relative to the healthy children of that age and gender in the same population. Median is regarded as a reference value, and 3^{rd} and 97^{th} percentiles as thresholds to indicate abnormally low or abnormally high values.
The indices can also be expressed as Zscore derived by using the formula:
(Measured value – Average value in the reference population) / Standard deviation of the reference population
The WHO growth standard (2006) for triceps skinfoldforage and subscapular skinfoldforage are used for interpretation.
An overview of skinfold thickness methods is outlined in Table 5.
Strengths
Limitations
Table 5 Characteristics of skinfold thickness methods.
Consideration  Comment 

Number of participants  Large 
Relative cost  Low 
Participant burden  Low 
Researcher burden of data collection  Medium as method requires highly trained observers 
Researcher burden of coding and data analysis  Low 
Risk of reactivity bias  No 
Risk of recall bias  No 
Risk of social desirability bias  No 
Risk of observer bias  Yes 
Space required  Low 
Availability  High 
Suitability for field use  High 
Participant literacy required  No 
Cognitively demanding  No 
Considerations relating to the use of skinfold thickness methods in specific populations are described in Table 6.
Table 6 Use of skinfold thickness methods in different populations.
Population  Comment 

Pregnancy  Suitable, but estimates of body fat changes derived from skinfolds are prone to measurement error, especially during pregnancy due to hydration level. Rapid decreases in measurement occur postpartum that are likely attributable to changes in hydration following delivery rather than marked changes in subcutaneous fat 
Infancy and lactation  Suitable 
Toddlers and young children  Suitable 
Adolescents  Suitable 
Adults  Suitable 
Older Adults  Suitable, but presence of oedema may affect estimates 
Ethnic groups  Suitable 
Other (obesity)  Suitable, but difficult to get reliable measurements, especially in those cases in which skinfold thickness approach the upper limit of the measurement range of the caliper 
To obtain reliable data from this method it is essential to standardize the procedure, train the participating staff and assess inter and intra observer reliability to monitor measurement error.
Refer to section: practical considerations for objective anthropometry