In epidemiologic studies of children, well-trained research staff can accurately and precisely assess weight, height, sitting height, skinfold thicknesses, and body circumferences.
A high proportion of children have overweight and obesity in the United States and other countries. Accurate assessment of anthropometry is essential to understand health effects of child growth and adiposity. Gold standard methods of measuring adiposity, such as dual X-ray absorptiometry (DXA), may not be feasible in large field studies. Research staff can, however, complete anthropometric measurements, such as body circumferences and skinfold measurements, using inexpensive portable equipment. In this protocol we detail how to obtain manual anthropometric measurements from children, including standing and sitting height, weight, waist circumference, hip circumference, mid-upper arm circumference, triceps skinfold thickness, and subscapular skinfold thickness, and procedures to assess the quality of these measurements. To demonstrate accuracy of these measurements, among 1,110 school-aged children in the pre-birth cohort Project Viva we calculated Spearman correlation coefficients comparing manual anthropometric measurements with a gold standard measure of body fat, DXA fat mass1. To address reliability, we evaluate intra-rater technical error of measurement at a quality control session conducted on adult female volunteers.
Overweight and obesity remain at epidemic levels, with approximately one-third of US children and two-thirds of adults having overweight or obesity, according to 2011 – 2012 estimates2. Overweight, obesity, and excess body fat confer greater risk of adverse cardiometabolic outcomes, including Type 2 diabetes and cardiovascular diseases, as well as other adverse physical and psychological health outcomes, including asthma and depression3,4,5,6.
Most studies that examine associations between obesity and later-life health outcomes assume accurate measurements of weight and length/height. Categories of weight status in adults and children include underweight (body mass index (BMI) < 18.0 kg/m2 for adults and < 5th age-sex-specific percentile for children), normal weight (BMI 18.0 to < 25.0 kg/m2 for adults and 5th to < 85th percentile for children), overweight (BMI 25.0 to < 30.0 kg/m2 for adults and 85th to < 95th percentile for children), and obesity (BMI ≥ 30 kg/m2 for adults and ≥ 95th percentile for children). Even minor measurement errors can influence these categorizations, especially in children for whom errors that appear small on an absolute scale can represent a large error relative to the child's size7. For example, in a prior study of children under 2 years of age, comparisons of length measured by the conventional clinical paper-and-pencil method with the recumbent length-board method indicated that paper-and-pencil method systematically overestimated length by an average of 1.3 (1.5) cm — an error that results in substantial misclassification7.
Using BMI to estimate adiposity offers many advantages for research, including the low equipment cost and minimal burden of height and weight measurement, as well as the opportunity to leverage self-report and clinical measures. However, even with accurate measurement of height and weight, variation in BMI does not necessarily reflect variation in adiposity, since BMI incorporates both lean and fat mass1. Thus, methods that directly measure adiposity are also important for understanding relationships with health outcomes.
Gold standard methods of adiposity and body composition measurement generally rely on technological methods, including air displacement plethysmography (ADP), hydrostatic weighing, magnetic resonance imaging (MRI), and computed tomography (CT), as well as dual X-ray absorptiometry (DXA)8,9,10. While these methods provide some of the most accurate measures of body composition, many of them are not practical in pediatric research studies, especially those that are field-based. For example, hydrostatic weighing requires that individuals be completely submerged in water. ADP equipment has, until quite recently, been available only to measure infants up to 8 kg or children and adults over the age of 6 years, but not toddlers or preschool-age children. CT scans emit a large amount of radiation compared with the other techniques, and the long acquisition time for MRI makes it impractical for many studies8. DXA emits about 1/500 the radiation dose of a CT scan, approximately the equivalent of one day of natural background radiation11, making it an attractive option for research studies involving children. All of these methods, however, are expensive to purchase and none are portable, making them infeasible for field-based studies with limited funding. Bioelectrical impedance analysis (BIA), which measures the impedance of a minute electrical signal sent through the body to estimate body composition, can be less expensive and more portable, but assumptions underlying the calculation of body fat are not applicable to small children10.
In contrast to these technology-based measures, manual anthropometric measures can be performed by a trained observer in most field settings and at a substantially lower equipment cost. Manual anthropometry includes measurements of height, weight, circumferences, and skinfold thicknesses8. Other advantages of manual anthropometry are that it involves no unnecessary radiation exposure, and skilled staff can obtain them efficiently. However, a common concern about manual anthropometric measurements is that they may be both inaccurate and imprecise12.
Obtaining accurate and precise measurements is possible with standardized procedures, adequate training, and sufficient attention to quality control (QC) procedures. The Project Viva team has developed a manual anthropometry training protocol that can yield high-quality, reproducible measures of stature, circumferences, and skinfold thickness. Over more than a decade, we have applied this training and QC protocol to mothers and children in Project Viva, a longitudinal, pre-birth cohort study13. Project Viva staff collected anthropometric measures on child during visits at birth (0 – 3 days), and with both the mother and child at the following time points: infancy (4.9 – 10.6 months), early childhood (2.8 – 6.3 years), mid-childhood (6.6 – 10.9 years), and early teen (11.5 – 16.5 years) 13.
This paper describes the protocol we developed and refined for measurement of height, weight, skinfold thicknesses (triceps and subscapular skinfolds), and body circumferences (waist, hip, and mid-upper arm circumferences [MUAC]) in Project Viva. We also describe how we have assessed both manual anthropometric measurement precision by means of technical error of measurement (TEM) calculations and accuracy in comparison to gold standard DXA measurements.
All the procedures are approved by the Harvard Pilgrim Health Care Institutional Review Board.
1. Training Procedures
2. Preparation of Subjects
3. Height
Figure 1: Location of the Mid-axillary Line. A coronal line halfway between the anterior and posterior axillary lines15. Please click here to view a larger version of this figure.
Figure 2: The Frankfort Plane. A horizontal plane that passes through the inferior margin of the orbit and the tragion (notch above the tragus of the ear) 16. Please click here to view a larger version of this figure.
4. Sitting Height
5. Weight
6. Waist Circumference
7. Hip Circumference
8. Mid-Upper Arm Circumference (MUAC)
Figure 3: Anatomy of the Shoulder. Shoulder anatomy includes the acromion process, identified in red17. Please click here to view a larger version of this figure.
9. Triceps Skin Fold Thickness
10. Subscapular Skinfold Thickness
11. Quality Control (QC) Procedures
This analysis addresses precision of the manual anthropometric measurements using data generated from quality control (QC) procedures, and evaluates intra-rater measurement error by Technical Error of Measurement (TEM) 12. TEM ranges of acceptability are based on calculations of repeated intra-rater anthropometric measures, where 95% of measurement discrepancy is due to factors other than rater imprecision12,14. A higher TEM indicates greater variability among measurements. Acceptable measurements, as analyzed in relation to TEM ranges of acceptability, fall below or within TEM ranges of acceptability. Large TEM values, that is TEM values above the range, indicate unreliability and indicate need for additional training. In this analysis, we present TEM values both in native units and in percent TEM, calculated by (average TEM/mean of the measure)*100, to compare across multiple anthropometric measures with different units.
Table 1 shows intra-rater TEM values from a QC session the Project Viva research team conducted on five healthy adult female volunteers. Six research assistants measured the volunteers repeatedly for a maximum of 60 times per anthropometric component. Each woman provided 55 measures, on average. TEM values fell within the range of acceptability for each measure, indicating low variability between measurements and thus accurate technique12,14. Research assistants most precisely measured height (% TEM = 0.2) followed by hip circumference (% TEM = 0.7). Research assistants were least precise at measuring subscapular skinfold thickness (% TEM = 7.4) and triceps skinfold thickness (% TEM = 6.9). Two members of the research staff achieved TEM values for waist circumference that were above the range of acceptability (TEM = 2.1; TEM = 3.0), signaling need for further training.
Data used to calculate TEM values are presented in Figure 4 and Figure 5. Figure 4 shows all height measurements obtained at the QC session described above. Each volunteer provided an average of 11 height measurements. Height measurements ranged from 151.4 cm to 166.4 cm amongst the five volunteers. Individual volunteer variation ranged from 1.3 cm (for the volunteer represented in red) to 4.1 cm (for the volunteer represented in orange) across all six raters. Most height measurements were within 1.0 cm of each other. The research assistant represented by circles was least accurate in measuring height, with several outlying measurements. This signals need for more training. Figure 5 shows all subscapular skinfold thickness measurements obtained at the QC session. Skinfold thickness measurements ranged from 8.6 cm to 33.8 cm amongst the five volunteers. Volunteer variation ranged from 1.6 cm (for the volunteer represented in blue) to 9.6 cm (for the volunteer represented in green). Research staff members experienced greater inter-rater reliability for subscapular skinfold thickness measurements than for height measurements.
Table 2 is adapted from a previous analysis by Boeke et al.1 on 1,110 Project Viva participants measured during mid-childhood (age 6.5 – 10.9 years). This table addresses accuracy of manual anthropometric measures of body fat, expressed as Spearman correlations with gold standard DXA measures. DXA fat was highly correlated with all manual anthropometric measures including weight (rs= 0.80), waist circumference (rs= 0.81), and sum of triceps + subscapular skinfold thickness (rs= 0.90) but was less strongly correlated with height (rs= 0.47).
Figure 4: Scatterplot of Height Measurements. Height measurements (cm) taken by six research assistants during QC session on five adult female volunteers. Each volunteer is represented by a different color. Each research assistant is represented by a different shape. Please click here to view a larger version of this figure.
Figure 5: Scatterplot of Subscapular Skinfold Thickness Measurements. Subscapular skinfold thickness measurements (cm) taken by six research assistants during QC session on five adult female volunteers. Each volunteer is represented by a different color. Each research assistant is represented by a different shape. Please click here to view a larger version of this figure.
No. of measures | Mean | TEM for each of the 6 research staff | Mean TEM | Acceptable | % TEM | ||||||||
TEM range[1] | |||||||||||||
1 | 2 | 3 | 4 | 5 | 6 | ||||||||
Height (cm) | 55 | 160.4 | 0.2 | 0.3 | 0.4 | 0.4 | 0.1 | 0.5 | 0.3 | 0.1 – 1.3 | 0.2 | ||
Waist circumference (cm) | 54 | 77.1 | 2.1 | 3.0 | 0.4 | 0.4 | 0.4 | 1.5 | 1.3 | 1.0 – 1.6 | 1.9 | ||
Hip circumference (cm) | 54 | 99.2 | 0.5 | 1.1 | 0.8 | 0.2 | 0.6 | 0.6 | 0.6 | 1.2 – 1.4 | 0.7 | ||
Mid upper arm circumference (cm) | 56 | 27.9 | 0.3 | 0.4 | 0.4 | 0.2 | 0.2 | 0.3 | 0.3 | 0.1 – 0.6 | 1.1 | ||
Subscapular skinfold thickness (mm) | 56 | 14.5 | 0.8 | 0.9 | 0.7 | 0.4 | 0.1 | 0.9 | 0.6 | 0.1 – 7.4 | 7.4 | ||
Triceps skinfold thickness (mm) | 55 | 16.7 | 0.7 | 0.7 | 1.2 | 0.9 | 0.1 | 1.9 | 0.9 | 0.1 – 3.7 | 6.9 |
Table 1: Intra-rater Reliability (within Measurer). Technical error of measurement (TEM) for each of the anthropometric measures, within each individual measurer. Data from six Project Viva research assistants performing repeated measures on five adult women. TEM calculated as , where d2 is the difference between repeated measurements by each research staff member (intra-rater reliability)12. A higher TEM indicates greater variability within the repeat measurements collected by each individual. %TEM calculated as (average TEM/mean of the measure)*100.
Height | Weight | Weight: | BMI | Waist circ | SS+TR | DXA fat | |
(cm) | (kg) | Height | (kg/m2) | (cm) | (mm) | (kg) | |
N | 1110 | 1110 | 1110 | 1110 | 1106 | 1103 | 875 |
Mean (SD) | 128.8 (7.8) | 29.0 (7.9) | 0.22 (0.05) | 17.2 (3.1) | 60.0 (8.3) | 19.9 (9.8) | 7.5 (3.9) |
Spearman correlation coefficient | |||||||
Height | 1.00 | 0.80 | 0.66 | 0.38 | 0.56 | 0.33 | 0.47 |
Weight | 1.00 | 0.98 | 0.84 | 0.87 | 0.69 | 0.80 | |
Weight: Height | 1.00 | 0.93 | 0.90 | 0.75 | 0.84 | ||
BMI | 1.00 | 0.86 | 0.79 | 0.83 | |||
Waist circumference | 1.00 | 0.73 | 0.81 | ||||
SS+TR | 1.00 | 0.90 | |||||
DXA fat | 1.00 |
Table 2: Correlations between Each of Several Anthropometric Measures and with DXA Body Fat among 1110 Project Viva Children at 6.5 – 10.9 Years. Adapted from Boeke et al1. BMI = body mass index; SS = subscapular skinfold; TR = triceps skinfold; DXA = dual X-ray absorptiometry.
Accurate body composition measures are critical for properly assessing childhood growth in research studies. Researchers widely accept DXA as a gold standard method, and many criticize manual anthropometric measures as being imprecise and inaccurate. However, this analysis of anthropometric techniques to estimate body fat suggests that well-trained research assistants who follow a standardized protocol can conduct manual anthropometric measures with excellent accuracy, yielding adiposity estimates that are highly correlated with DXA1. In addition to individual measures, combinations of manual anthropometric measures, such as the sum of skinfold thickness and weight to height ratio, are highly correlated with measures of DXA body fat. The purpose of this protocol is to standardize processes for eight commonly used anthropometric measures, to improve accuracy and allow comparison between research studies and pooling of results.
Critical Steps within the Protocol
Accurate assessment of body composition with manual anthropometry requires sufficient time for training and the conduct of quality control procedures to ensure precision and accuracy. Given the lead time required, research assistants optimally should be available for a minimum of 24 months. Equipment should be sturdy and regularly checked for calibration. To achieve high levels of reliability, raters must follow all steps of the anthropometric protocol precisely, as even minor alterations affect accuracy. Anecdotal evidence suggests that improper identification of anatomical measurement sites, hand placement, and tautness of measurement equipment cause the greatest variation between measures. With attention to detail, the enclosed protocol provides a clear method of collecting precise measurements yet it also shows that protocol alone is insufficient to achieve universally accurate results in the field.
Modifications and Troubleshooting
Low intra-rater TEM values achieved by research staff members who completed brief but rigorous training suggest a high level of repeatability. Trainers should provide additional training, however, to research assistants with TEM values outside of the range of acceptability for any measurement. To ensure that research assistants achieve precise measures in the field, all trainees undergo a certification process. Trainees must pass two field QC assessments in order to be fully certified as anthropometric raters. In this analysis, two raters achieved waist circumference measures outside of the TEM range of acceptability (as shown in Table 1), thus did not pass the certification process. These trainees received additional supervision and training prior to repeating their QC assessments and independent field data collection. While the outlined QC procedures provide overall confidence in a research assistant's ability, they do not produce immediate feedback at the time of the field measurement. One approach to overcoming this limitation is to have two observers each perform the measurement on the same subject. If numbers differ, the research assistants can take additional measurements; data analysts may use the mean of the two research assistants' measures.
Limitations of the Technique
Manual anthropometric assessment requires time and training, with ongoing monitoring of quality. However, other methods also may require substantial training or other start-up costs in addition to the costs of the equipment. For example, Massachusetts (where Project Viva is conducted) requires that anyone conducting a DXA scan be certified as a radiologic technologist or licensed physician. Project Viva research assistants studied for an average of 60 h for the 3-h long test, for which registration cost $425. Additionally, manual anthropometry cannot directly assess visceral fat, in contrast to some imaging techniques.
Significance of the Technique with Respect to Existing/Alternative Methods
All approaches to measuring body composition have advantages and disadvantages. Manual anthropometry can be used at all ages, confers no risks, and has minimal costs. However, the success of these methods depends on the availability of a stable staff able to spend weeks or months completing the training procedures and to follow protocols precisely.
Future Applications or Directions
With minor alterations, these techniques can also be adapted for other anthropometric measures, such as recumbent length, and chest and thigh skinfolds. In summary, this paper demonstrates that with training and QC, research assistants can perform manual anthropometric methods for assessment of adiposity in children with precision and accuracy. These methods are safe, low-cost, and require minimal, portable equipment and are therefore suitable for field studies among children.
The authors have nothing to disclose.
We greatly appreciate the contributions of our expert anthropometry trainers Irwin Shorr and Jorge Chavarro; the many volunteers who have allowed themselves to be pinched and measured in our anthropometry workshops, and the Project Viva mothers and children for their invaluable contributions. We’d like to extend a special thanks to members of the Project Viva research staff, past and present, especially to Nicole Witham and Marleny Ortega, for their contribution to the video accompanying this manuscript. Funding from the National Institutes of Health supported this work (R01 HD 034568, K24 HD069408).
Stadiometer | Weigh and Measure, LLC | SSWM-1 | Basic Shorr board (without smooth slide features) can also be used. In order to accommodate the width of children's hips during sitting height, the base of a stadiometer should be approximately 60 cm wide or larger. |
Bioimpedance scale | Tanita Coporation of America | TBF 300A (model is discontinued), DC-430U is comparable | Scale is used for weight and bioimpedance. Any digital, standardized scale can be used for weight only. |
Skinfold Caliper | Holtain Limited | n/a | This model uses a dial gauge in graduations of 0.2 mm. Models with a linear gauge are also acceptable. |
Hip/waist tape measure | Gulick II Plus Measuring Tape | 67019 | This model uses compression bands, which makes it easier to identify how firmly the tape measure is being pulled. The compression band is not necessary, but highly recommended. |
MUAC measuring tape (ShorrTape© Measuring Tape) | Weigh and Measure, LLC | STape | The tape measure should be flexible with a single or double slotted insertion window. |