Intrarater Reliability of Tonometry and Bioimpedance Spectroscopy to Measure Tissue Compressibility and Extracellular Fluid in the Legs of Healthy Young People in Australia and Myanmar

Introduction

Lymphedema is the external manifestation of an overwhelmed lymphatic system. Extracellular fluid (ECF) and circulating proteins are not adequately removed from the tissue spaces, and over time, the excess protein-rich fluid is replaced by an accumulation of fibrosis and fatty tissue, particularly in the subcutaneous compartment. As lymphedema progresses, the affected body part becomes enlarged, the skin thickens (hyperkeratosis), and eventually takes on the characteristic appearance of elephantiasis. ¹

Globally, the greatest cause of lymphedema is lymphatic filariasis (LF), a nematode worm transmitted by mosquitoes in 73 tropical countries. It is estimated that 16 million of the world's poorest people have LF-related lymphedema (LFRL). ² Lymphedema also forms after treatment for some cancers and breast cancer-related lymphedema (BCRL) affects between 17% and 28% of breast cancer survivors. ³ Whether LF or cancer related, progression of lymphedema can be slow and persistent or may advance rapidly as a consequence of repeated bacterial or fungal infections. During the onset and progression of lymphedema, tissue compressibility and ECF loads change along a predictive course. ⁴ Devices commonly used to measure changes related to BCRL include tissue tonometry to quantify compressibility of affected tissue and bioimpedance spectroscopy (BIS) to quantify ECF fluctuations. ⁵ Assessment of LFRL more frequently depends on clinical staging by presentation and history with historically limited use of tools to quantify changes. ⁶

Use of tissue tonometry to provide an objective measure of compressibility in the skin and subcutis was first reported in 1976. ⁷ It provides an objective measure of the changes in tissue composition, which occur over the course of lymphedema. In the early fluid-rich stage, the tissue is soft and easily compressible, but as lymphedema progresses, the skin and underlying tissues become progressively thicker and harder and compressibility will decrease. Several versions of the mechanical Tonometer and recently developed digital devices (the digital Indurometer and the SkinFibroMeter) are now available. Increased ECF loads in lymphedema have been measured using BIS since 1996⁸ and several devices are in regular use in both clinical and research settings. ⁹

Previous studies have shown tonometry and BIS to be reliable in measuring changes associated with lymphedema, but most were in middle-aged women with unilateral BCRL of the arm using the contralateral arm as a control.^10–14 More recently, attention has been directed toward detecting subclinical changes, which occur before lymphedema can be observed clinically, and BIS has been shown to be a reliable tool for this purpose in women at risk of BCRL. ¹⁵ To date, only BIS has been used on infants. ¹⁶ Few lymphedema studies have included people with leg lymphedema ¹⁷ and although human growth and aging may influence measures, no reliability studies have been conducted on young people. ¹⁸

Confident and defensible use of these tools to diagnose and monitor lymphedema is dependent on establishing whether they are reliable for use with populations of need. While lymphedema is essentially the same disease regardless of the cause, LFRL affects a much larger and demographically different population than women in developed countries with BCRL. Most commonly leading to lymphedema of the legs, LFRL can affect both men and women at any age, young, middle aged, and older. ⁶

The aim of this study was to identify the intrarater reliability of three different tissue tonometers and one BIS device to assess the lower limbs of young people. Populations from two different ethnic backgrounds were included to inform the methodology of a forthcoming study to detect subclinical lymphatic change in young people living in a filariasis endemic region in Myanmar.

Materials and Methods

The Guidelines for Reporting Reliability and Agreement Studies were used to inform the study design and reporting. ¹⁹ A convenience sample of young people from Townsville, Australia, and Amarapura Township, Mandalay Region, Myanmar, were invited to participate in the study. In Australia, people aged 8–21 years were recruited through institutional emails and local notice boards. In Myanmar, young people aged 10–21 years were recruited with the assistance of local health workers and residential block leaders. Participants were excluded if they had any clinical sign of lymphedema or any injuries to their legs. The Myanmar cohort was screened for LF infection using immunochromatographic test cards, and young people who tested positive were excluded from the reliability analysis. Ethical approval for the study was provided by the James Cook University Human Research Ethics Committee (approval number H5497) and by the Myanmar Ministry of Health and in accordance with the Helsinki Declaration as revised in 2008. Consent to participate in the study was given by young people aged 18–21 years and by a parent or guardian of minors aged 8–17 years. The parent or guardian also accompanied the child during the measurement session.

Height was recorded in centimeters and weight was recorded in kilograms. All tissue tonometry measures were taken at the midpoint of the anterior thigh, posterior thigh, and calf of each lower limb. A tape measure was used to obtain three length measurements: the calf between the base of the heel and the crease behind the knee, the posterior thigh between the crease behind the knee and the gluteal fold, and the anterior thigh between the superior border of the patella and the crease of the groin. The midpoint of each segment was then calculated by halving the full-length measures and a washable skin marker was used to mark each midpoint on the leg. All tonometry measures were taken at the marked points by a qualified lymphedema practitioner experienced in the use of tonometry (JD). In Myanmar, local health assistants collected informed consent, recorded height, weight, infection status, and interview responses. Verbal assent to take the measures was reiterated before any leg measurements were taken. Limb dominance, which has been reported to alter BIS, ¹¹ was determined by asking the following question: Which foot do you use to kick a ball? Blinding was not possible as all measures were taken by a single operator, but a sequential order of measures was followed in both groups to minimize the risk of bias.

The mechanical Tonometer (Flinders Biomedical Engineering, Australia) is a handheld device with a 275 g mass above a 1 cm diameter indenter, which extends through an opening in a 6 cm diameter reference plate and applies a differential pressure of 263 g/cm². With the reference plate resting lightly on the skin, the weighted indenter presses into the skin and underlying tissue and the degree of tissue resistance is shown on an analog displacement gauge in 0.1-mm increments. Higher values indicate softer tissue as the indenter is able to press deeper into the tissue.

The Indurometer (Flinders Biomedical Engineering, Australia) is a digital version of the mechanical Tonometer utilizing a constant force spring to deliver a 200 gm pressure. The reference plate rests evenly on the surface of the skin and the device is pressed into the tissue. A beep is emitted once the 200 gm force has been achieved. The result is displayed digitally in 0.01-mm increments, which can be read from the digital display after the measure is completed; higher values indicate softer tissue.

The SkinFibroMeter (Delfin Technologies, Finland) has a smaller reference plate (diameter 2 cm) and a 1-mm-long indenter and records the resistance to 50 gm of pressure. The device is applied gently five times in the same location and the digital readout displays the average measure of resistance in Newtons (N), a lower value indicating less resistance or softer tissue.

Bioimpedance was measured using the SFB7 (Impedimed, Queensland), a portable, battery-operated multifrequency analyzer, which delivers 256 frequencies over the range 3–1000 KHz. Low frequencies pass through the ECF, while higher frequencies pass through both the ECF and the intracellular fluid (ICF). An increase in the Ri:Re (intracellular impedance: extracellular impedance) ratio indicates an increase in ECF load. ²⁰

Although persistent indentation should not occur in healthy skin, the tonometry instruments were used in the following order to avoid any residual pitting from the previous device: (1) The SkinFibroMeter, which applies only 50 gms of pressure; (2) the Indurometer, which requires 200 gms of pressure and is applied only briefly; and (3) the mechanical Tonometer, which also applies 200 gms, but must remain on the skin while the operator reads two analog scales. All measures were taken three times with each device in the order right calf, left calf, right posterior thigh, left posterior thigh, right anterior thigh, and left anterior thigh. The SkinFibroMeter and Indurometer were used with both groups and the mechanical Tonometer was used with the Myanmar group only.

The BIS measures were collected last. Self-adhesive electrodes were placed on the skin at the ankles and the hands according to the manufacturer's instructions for whole leg analysis. Both legs were measured three times beginning with the right leg and the Ri (ICF resistance) and Re (ECF resistance) values for each measure were recorded.

All analyses were performed using STATA 12 (StataCorp.). A t-test was used to compare group characteristics such as height and weight. As the data were continuous, an intraclass correlation coefficient (ICC) was used to determine the intrarater reliability or test–retest reliability. This test indicates the reliability of each measurement device to differentiate participants under repeated similar assessment conditions. ¹⁹ A one-way repeated measures analysis of variance was used to obtain within-subject variance and hence overall average ICC for the four devices at each measurement location. Higher ICC represents better intrarater reliability, hence an average ICC score of 0 represents no correlation and 1 equals absolute agreement. An ICC of less than 0.4 is considered to indicate poor agreement between measures, between 0.4 and 0.75 is fair to good agreement, and greater than 0.75 is excellent. ²⁰ A sample size estimation of 62 subjects was based on an expected correlation coefficient of 0.4 with a 95% confidence interval using an online calculator (http://biomath.info/power/corr.htm).

Each device uses a different scale, which prevents direct comparisons of the overall means and variance of the values recorded; so, to determine the agreement between devices at each measuring location, a coefficient of variation (CoV) [19] was calculated using the following formula: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \rm CoV} = \mathop {\hbox{standard deviation} \ ( { \rm SD} )\over{\hbox{Overall mean for the device}}} {{\times 100}} \end{align*} \end{document}

The CoV is reported as a percentage, a lower score indicating less variation (more agreement) between devices.

Results

Thirty-four young Australians and 38 young Myanmar people were included in the study. Despite including younger children in the Australian group, there was no significant difference in the mean age, gender, or leg dominance between the two groups. The majority of participants were female (60.4%), the mean age (range) was 15 years (8–21), and 95.6% were right leg dominant. The Australian group was significantly taller and heavier than the Myanmar group, but there was no statistically significant difference in BMI (Table 1).

Indurometry of the anterior thigh and BIS measurements were not recorded for one Australian participant. Poor electrode contact affected a small number of BIS measures, and if any single measure was affected, the remaining two measures in that measurement set were also discarded. This meant that there were bilateral BIS measures available for 32 Australian and 36 Myanmar participants, with dominant leg values available for all Myanmar people and nondominant leg values available for 33 Australians.

The midpoint of the anterior thigh recorded the softest measures by all three devices (highest mean values by mechanical Tonometer and Indurometer, lowest mean values by SkinFibroMeter). The hardest tissue was consistently recorded at the midpoint of the calf, and the mean values for the posterior thigh fell between these two ranges for all devices. The BIS ratio of resistance, ICF:ECF, was slightly higher in the nondominant leg in both populations, indicating that this leg had slightly more ECF than the dominant leg, but this was not significant. Mean values and SDs for each device at each measuring point are provided as supplementary material (Supplementary Table S1; Supplementary Data is available in the online article at www.liebertpub.com/lrb).

Of the tissue tonometers, the mechanical version (used only in the Myanmar group) had the best ICC (0.893–0.964) and CoV (10.8%–20.4%) scores. Excellent ICC scores were also recorded for the Indurometer (0.792–0.956) with low (good) CoV of 14.8%–32.2%. The SkinFibroMeter had good to excellent ICC scores (0.565–0.877); however, this device scored the highest (poorest) CoV (18.1%–43.1%). The ICC scores for BIS approached 1.0 for all measures (ICC = 0.912 to >0.999) and the CoV was very low (2.3%–20.8%). The ICC, estimated reliability, and CoV scores for each device and group are reported in Table 2.

Both the mechanical Tonometer and the Indurometer had better ICC scores when used to measure the posterior thighs and calves than when used to measure the anterior thighs in both populations. ICC scores for the SkinFibroMeter were better in the Myanmar group than the Australian group, but in contrast to the other devices, the best ICC scores were on the anterior thigh. The relationship between anatomical location of the measuring points and the ICC score is demonstrated in the radar graph in Figure 1. The data closer to the outside border indicate a higher ICC score (excellent agreement between repeated measures on a single measurement point) on a scale of 0–1.

OPEN IN VIEWER

FIG. 1.

Radar Graph of the ICC scores for the three tissue tonometers at six measuring points. ICC, intraclass correlation coefficient. A color version of this figure is available in the online article at www.liebertpub.com/lrb.

The converse was true for the CoV. All devices recorded the lowest CoV (greatest agreement) at the anterior thigh. The relationship between agreements as represented by the CoV is demonstrated in Figure 2. Data points closer to the center represent less difference between the CoV (standardized SD as % of mean) of any device's measure at each point, on a scale between 0% and 40%.

OPEN IN VIEWER

FIG. 2.

Radar Graph of the CoV scores for the three tissue tonometers at six measuring points. CoV, coefficient of variation. A color version of this figure is available in the online article at www.liebertpub.com/lrb.

Discussion

This study compared intrarater reliability of three tissue tonometers and one BIS device on the legs of young people in two tropical locations. It also compared agreement in the measurement rating scales between the three tissue tonometers: mechanical Tonometer, digital Indurometer, and SkinFibroMeter. BIS is an important tool in both clinical practice and research, is subject to very little operator influence, and, in this study, showed excellent reliability in keeping with multiple previous studies.^{5,11,12,18,21} Mechanical tissue tonometers have also been important in quantifying fibrous changes associated with lymphedema that occur in the skin and underlying tissue, especially in research settings,^5,10 but their use in clinical settings has been hampered by problems with the device design, which can affect the measures. ¹³ The mechanism driving the indenter relies on gravity, so if the device is placed on a skin surface, which is not horizontal to the ground, the force of gravity will be changed and affect the true measure. Additionally, the analog readout must be viewed in situ and involves reading two dials, which record increments in different directions, that is, the whole millimeters are recorded on a small anticlockwise dial and the micrometer increments are recorded on a larger clockwise dial. Keeping the device horizontal and correctly reading the dials while also trying to reduce any parallax error can be cumbersome at some measurement sites. Furthermore, when soft edematous tissue is measured, the indenter continues to move slowly into the tissue and the dials may not come to a complete stop. The operator must then decide at what point to read the dials and this introduces a subjective component to the measure. This can be compounded when the first application leaves a visible indentation in the skin, thereby influencing subsequent measures at the same point. ¹³ Although the digital Indurometer was designed to overcome some of the limitations of the mechanical Tonometer, the comparative study by Vanderstelt et al. ¹⁴ found the CoV (%range) of measures with the newer device to be significantly higher (28.7%–33.6%) than the original mechanical unit (13.0%–20.2%) among participants with stage 1 BCRL. The differences were not significant at other measuring points (13.0%–33.6%) or when comparing measures on the control arms. The authors noted that although the Indurometer may not be as reliable as the mechanical Tonometer in the early fluid-rich stage of lymphedema, the ease of operation is a significant improvement and makes the device more accessible to a variety of operators. The SkinFibroMeter introduces a new set of parameters for tissue assessment and there are no previous reliability studies for comparison.

In our study on young healthy legs, there was little difference in ICC score between the mechanical Tonometer and Indurometer at any measuring point, and both devices demonstrated excellent reliability and low (good) CoV values. This is similar to the earlier study by Pallotta et al. ¹³ who reported a CoV (% range) for the mechanical Tonometer (8%–21%) and the Indurometer (7%–16%). Better scores in our study for the mechanical Tonometer (10.8%–20.4%) compared with the Indurometer (14.8%–32.3%) may be because the single operator had extensive experience using the mechanical Tonometer in research settings, whereas the previous studies used three ¹³ and seven operators. ¹⁴ However, both the digital devices have inbuilt sensors, which prevent any single measure from being recorded if the application is erratic, too fast/slow, or too heavy/light. They are designed to be used with minimal training and any influence of inexperience should have been quickly corrected during preliminary practice and familiarization sessions with negligible effect on the study results. It is more likely that the significant differences between devices found by Vanderstelt et al. ¹⁴ in stage 1 BCRL were related to the complex changes that occur during the onset and development of manifest lymphedema, particularly in the early fluid-rich stage when tissue changes are still in a state of flux and issues of indentation can confound accurate readings. This hypothesis is supported by the same study, ¹⁴ which found the lowest CoV values (greatest agreement) for measures of late-stage BCRL when fibrosis and thickened skin will be patently overt and less susceptible to fluctuation.

Conversely, in our study of healthy participants, greater variability (lower ICC score and higher %CoV) was found in measures of the harder skin of the calf rather than the softer tissue of the anterior thigh, so a fluid-rich tissue state is not the cause of variability in this cohort. Rather, it is likely that the poorer CoV values in calf measures were due to the size of the reference plate in relation to the anatomical location of the measure. The relatively large reference plates of both the mechanical Tonometer and the Indurometer often extended beyond the curve of the narrow calves of the children, whereas on the broader, softer thigh area, the reference plate could generally rest fully on the skin, even in smaller participants. Similar issues with positioning the reference plate to fully contact the skin have also been reported during measures of the anterior chest wall. ¹³ Designers and manufacturers may wish to consider modification to a smaller reference plate for increased usability in nonuniform areas such as the anterior chest and for use with children and smaller limbs.

In the current study, the SkinFibroMeter showed excellent reliability in three of the measurement points in the Myanmar group and good reliability in all other measurement points in both groups, but when assessing agreement, the CoV was higher (poorer) for this device than the other tissue tonometers and also higher than the CoV reported in previous studies of tonometry. Additionally, the lowest (best) CoV for the SkinFibroMeter was for the anterior thighs (18.1%–26.1%) with higher (worse) CoV for the calves (25.8%–38.2%), which is opposite to the other two devices. It is difficult to understand what might account for the increased variability with this device since the smaller reference plate and inbuilt error rejection should make it more reliable in narrow limbs. The very light technique may make it more difficult to apply the reference plate evenly, or it may be more susceptible to pressure variation by the operator, and the range at which the error measures are rejected is not known.

All three tissue tonometers recorded the softest absolute measures at the midpoint of the anterior thigh where the fatty layer over the muscle would naturally be thicker than the skin over the musculotendinous junction, which lies at the midpoint of the calf. The latter showed the hardest measures, with values over the more muscular posterior thigh falling between the two. BIS measures indicated a slightly higher fluid load in the nondominant leg in both young populations, which is opposite to reports of slightly increased fluid loads in the dominant arms of normal adults.^8,15

Although all four devices were found to be reliable, the level of reliability as indicated by the ICC scores was related to the type or composition of the underlying tissue at the measurement point. These differences may be small and not a direct indication of reliability in lymphedema, but they have the potential to impact results. Limb dominance and the anatomical features of the measurement location should be taken into account when choosing the best tool for the measure and when interpreting measures that compare affected and unaffected limbs in lymphedema.

Conclusions

The findings of this study that all four devices have good to excellent reliability have widened the demographic in which they can be used to include the lower limbs of a young nonlymphedema population. Given the operational issues associated with the mechanical Tonometer and based on usability, reliability, and agreement between devices, the newer digital Indurometer is an acceptable replacement for the mechanical Tonometer to measure tissue compressibility in young people without lymphedema. The SkinFibroMeter is also reliable and shows good agreement between tissue tonometers in this group, although to a slightly lesser degree than the other two tonometers. All four devices are appropriate for use in future studies on the legs of young people without clinical signs of lymphedema. Further evaluation of the newer digital devices in lymphedema populations is needed and in particular studies on leg lymphedema are missing from the literature.