Electrode Equivalence for Use in Bioimpedance Spectroscopy Assessment of Lymphedema

Abstract

Background:

Bioimpedance spectroscopy (BIS) is commonly used in the assessment and monitoring of lymphedema. This study investigated electrodes as a source of variability that could impact the accuracy of BIS in the clinic and determined if Ag/AgCl electrocardiograph (ECG) electrodes could be used as an alternative to instrument-specific electrodes.

Methods and Results:

Two types of Ag/AgCl electrodes were studied: instrument-specific bioimpedance electrodes (bioimpedance) and single tab ECG electrodes ( cardiac ). Six areas of investigation were addressed: intrinsic electrode resistance; electrode age; drive electrode position; electrode width/surface area; concordance between cardiac and bioimpedance electrodes; and mixing electrode types and batches. Participants included women (n = 26) and men (n = 8), both with (n = 4) and without lymphedema (n = 30). Resistance (R₀) of the limbs was measured and used to calculate interlimb BIS ratios. Intrinsic electrode resistance varied between batches (p ≤ 0.001), with cardiac electrodes recording higher resistance. Electrode age had no impact on limb resistance (p = 0.85). Drive electrode position biased limb resistance (0.1%–2.3%) and electrode size/surface area had a small (≤1%), but significant effect on limb resistance (p ≤ 0.001). However, calculation of interlimb BIS ratios negated the impact of these as well as any effect of mixing electrode batches and types (p = 0.15–0.96). Electrode type had no impact on arm and leg resistance, or interlimb BIS ratios (p = 0.173–0.289).

Conclusion:

Calculation of interlimb BIS ratios improves accuracy of clinical BIS. Ag/AgCl cardiac electrodes can be used as an alternative to device-specific electrodes to measure limb resistance.

Background

Bioimpedance spectroscopy (BIS) is a measurement technique that is growing in popularity for the detection and monitoring of lymphedema. It is used in both research^1–4 and clinical practice^5,6 to quantify tissue extracellular fluid (ECF). BIS passes a small alternating electrical current, from electrodes placed on the skin's surface, through the tissues in the body or a body region of interest, to determine the resistance or opposition to flow of the current through the ECF.⁷ ECF volume is inversely related to resistance^8,9 and as lymph is a major component of ECF, BIS is particularly well suited for detection of overall changes in volumes of lymphatic fluid.^7,8

BIS assessment for lymphedema uses a tetrapolar electrode arrangement in which the voltage measurement and drive electrodes are separated, thereby minimizing skin contact resistance errors.^10,11 The two measurement electrodes define the BIS measurement region to be quantified and the two distally located drive electrodes span the two measurement electrodes. The current is passed between two drive electrodes and the resistance at zero frequency (R₀) is derived according to the Cole Model for the region defined by the measurement electrodes.¹² R₀ represents the resistance of ECF, including lymph. Accurate placement of the measurement electrodes is critical as small placement errors can affect the measurement as resistance varies proportionately to interelectrode distance.^13,14 Furthermore, a minimum distance between the measurement and drive electrodes has been suggested to reduce interference between electrodes. There is little research confirming an optimal or minimal distance for measurement and drive electrode spacing, although 5 cm has been suggested.^6,13 While a 5 cm spacing is readily achievable when measuring whole limbs, it may be difficult or impossible to achieve this spacing when measuring short limb segments such as the hand¹⁵ or when performing resistance measurements in babies or infants.¹⁶ It is unclear what effect a smaller or greater distance between the drive and measurement electrode may have on measurement outcomes. While it is unknown if it is essential to standardize electrode placement and the spacing between the drive and measurement electrodes, some manufacturers have introduced electrodes that preset the separation distance (dual-tab electrodes).

The intrinsic electrical characteristics of the electrodes are an additional potential source of measurement variability. To mitigate this variability, some manufacturers of impedance analyzers may specify particular electrodes of a certain size and type for use with their instruments. Typical impedance electrodes for lymphedema assessment are Ag/AgCl gel electrodes and are similar to those commonly used for electrocardiograph (ECG) measurements, and in many situations, ECG electrodes are frequently used for impedance measurements. A wide range of electrode types are commercially available, which potentially adds further variability to resistance measurements. Nescolarde et al.¹⁷ found significant variation in bioelectrical impedance vectors when using different electrode types to measure whole body resistance in 35 healthy subjects. It is unknown if this same variation is found measuring body segments such as what occurs in lymphedema. Other potential sources of variation are electrode age and size. Electrode gel is susceptible to changes in hydration over time and this may affect its electrical characteristics. A wide range of electrode sizes are available and in an attempt to provide sufficient space between electrodes¹⁵ or to reduce costs, electrodes are sometimes cut in half. It is unknown whether electrode age and size affect resistance measurement in humans. It is therefore important to understand and quantify the impact of these potential sources of error in the clinical environment.

The aims of this technical report were therefore as follows: (i)

to explore factors related to electrodes that could impact impedance measurement in the clinical environment, and

(ii)

to determine whether ECG electrodes could be used as an alternative to instrument-specific impedance electrodes in the measurement of unilateral lymphedema.

Materials and Methods

Participants were a sample of convenience from staff and clients associated with a lymphedema clinic in rural New South Wales, Australia. Any person who was pregnant or had cardiac (pacemaker or defibrillator) or metal implants in any limb was excluded from the study. Participants attended one or more measurement sessions of ∼10-minute duration. Ethics approval was obtained from the Human Research Ethics Committee of the University of Sydney (Project No. 2016/450). All participants provided written informed consent.

Equipment

Two different devices were used to address the aims of this study. A single low-frequency (10 kHz)¹⁸ tetrapolar bioimpedance device (XCA ImpediMed Imp™) was used to investigate the resistance characteristics of the electrodes themselves. For the studies on human participants, a single-channel, tetrapolar BIS device that scans between 4 and 1000 kHz¹⁹ (SFB7 ImpediMed Imp™) was used. Human BIS files were processed using the Bioimp software (vs 5.2.4.0) from which the resistance at zero frequency (R₀) was derived. R₀ was used as the variable for electrode comparisons since it is the variable used in the BIS assessment of lymphedema. Resistance is the opposition to current flow in the body conductors (the fluids), while reactance is the opposition to current flow due to the body's tissues acting as capacitors. Impedance is the total opposition to current flow through the body that includes both resistance and reactance.⁷ As resistance and impedance values are identical at zero frequency, both terms can be used interchangeably in human studies, in which R₀ is the parameter of interest; however, in this study, the term resistance has been used to reduce confusion. Integrity of the BIS hardware was assessed daily using the test cell supplied by the manufacturer.

Two types of electrodes were used: single-tab electrodes distributed specifically for bioimpedance measurements ( bioimpedance ) and single-tab disposable resting ECG style electrodes ( cardiac ). Both electrodes were Ag/AgCl gel and 33 mm long. However, the bioimpedance electrodes were 22 mm wide whereas the cardiac electrodes were 29 mm wide, providing a 32% larger contact area. The type of cardiac electrodes used in the study was determined by those that were readily available at the health service where the project was completed. The cardiac electrodes were cut in half across the tab¹⁵ for some studies as they were larger than the bioimpedance electrodes.

Protocol

Participant height, to the nearest 0.5 cm, and weight, to the nearest 0.1 kg, were measured to allow the calculation of body mass index (kg/m²). Resistance of participant limbs was measured using the principle of equipotentials.¹³ Participants were positioned in supine for between 3 and 10 minutes, with the arms and legs extended for the assessment. Electrode placement sites were cleansed with an alcohol wipe and then marked with indelible ink to reduce positioning variability introduced by test–retest design. Sites marked for electrode placement on the hand were as follows: the base of the nail on the third finger¹⁵; metacarpalphalangeal (MCP) joint of the third finger⁷; mid-wrist in line with the ulnar styloid¹³; and 46 mm distal from the mid-wrist position to replicate dual-tab electrode positions (Fig. 1). Electrode sites on the foot included the following: the base of the second and third toe²⁰; between the malleolus^13,20; 46 mm distal from the malleolus position to replicate dual-tab electrode positions; and midway between the 46 mm and base of the second and third toe positions (to replicate the MCP position on the hand) (Fig. 2). Electrodes were centered over these sites.

FIG. 1.

Position of drive electrodes on the hand. Electrodes labeled as follows: measurement electrode placed over ulnar styloid (M), drive electrode at 46 mm from ulnar styloid (D1), drive electrode over MCP at base of third finger, the currently used standard electrode position (D2), and drive electrode at base of nail of the third finger (D3). MCP, metacarpalphalangeal.

FIG. 2.

Position of drive electrodes on the foot. Electrodes labeled as follows: measurement electrode placed between malleoli (M), drive electrode at 46 mm from malleolus (D1), drive electrode midway between D1 and D2 to replicate MCP on hand (D2), and drive electrode at base of the second and third toe, the currently used standard electrode position (D3).

Six studies were undertaken, investigating effects of the following: (i) intrinsic electrode resistance; (ii) electrode age; (iii) drive electrode position in relation to the measurement electrode; (iv) electrode width/surface area; (v) concordance between resistance measured by cardiac and bioimpedance electrodes; and (vi) effect of mixing electrode batches and types on interlimb BIS ratios.

Participants included persons with and without lymphedema to provide a range of resistance values and were from a population of convenience. Participant characteristics for each of the six studies are shown in Table 1. In total, 34 participants, including 26 women, 3 of whom presented with upper limb lymphedema secondary to treatment for cancer, as well as 8 men, 1 of whom presented with lymphedema of the upper limb secondary to cancer, were included in this study. Of the participants without lymphedema, 12 persons participated in multiple studies. None of the participants without lymphedema (n = 30) exceeded the whole arm BIS ratio thresholds (3SD) for lymphedema as determined by Cornish et al.⁵ In contrast, all participants with arm lymphedema exceeded these thresholds.

Table 1.

Participant Characteristics from All Studies

Study	A, Electrode characteristics	B, Electrode age	C, Drive position	D, Width of electrodes	E, Comparison between types of electrodes	F, Mixing batches and types	Lymphedema participants
N	NA	5	6	5	30	6	4
Age, years, mean (SD)	NA	24.6 (21.5)	29 (4.8)	31.8 (11.9)	47.6 (17.9)	47.2 (9.2)	66.8 (10.1)
BMI,^a mean (SD)	NA	25.4 (2.7)	26.6 (3.1)	27.9 (4.6)	26.9 (4.14)	26.2 (3.4)	26.9 (4.14)
Male:Female	NA	0:5	0:5	0:5	7:23	2:4	1:3

BMI, body mass index (kg/m²).

NA, not applicable, nonhuman study.

Studies

Determination of intrinsic electrode resistance

Electrodes from four different production batches of both the cardiac and bioimpedance electrodes were procured for testing. A batch of electrodes is defined as having the same manufacturer's production batch number and expiration date. Six electrodes from each production batch were sandwiched, gel sides together (tabs facing opposite directions) to form three test units for each batch. Drive and measurement leads were attached to each test unit (Fig. 3). Two measurements using the XCA machine were completed to determine the resistance and reactance of each electrode type. This resulted in six measurements per batch of electrodes (two measurements from three test units from each batch).

FIG. 3.

Two electrodes were sandwiched together to form a test unit. Drive (D) and measurement (M) leads were attached to each electrode test unit and two resistance measurements taken of each unit. Three electrode test units from each batch were used to determine intrinsic electrode resistance.

Effect of electrode age on measurements of limb resistance

Two different production batches of bioimpedance electrodes were used to determine if electrode age had an impact on limb resistance measurements. The first batch was ∼44 months old, and therefore 20 months past their use-by date (Expired). In contrast, the second batch was ∼7 months old, and within 17 months of their use-by date (Current). The electrodes had been stored in their original unopened manufacturer's packaging in an air-conditioned building. Whole arm and leg resistance measurements were taken using one batch of electrodes (Table 1); these electrodes were removed and replaced with the other batch of electrodes and measurements completed again. The electrodes were placed over the third finger MCP, in line with the ulnar styloids, between the malleolus and base of third toes. Six women participated, providing 24 limbs for measurement. The age of the electrode used first was alternated to reduce the biological impact of lying in supine, specifically to account for the rises in limb resistance that occurs as ECF levels lower in the limbs, as fluid flows toward the trunk when a person assumes a supine (horizontal) position.^21,22

Effect of drive electrode position on limb resistance

Electrodes from a single current production batch of bioimpedance electrodes were used to determine the effect of drive electrode position on limb resistance. Three different drive electrode positions were used on the hand (for arm resistance) and foot (for leg resistance) of the limb being measured. The measuring electrode remained the same at the right ulnar styloid or between the right malleoli for all measurements. The three drive electrode positions for the arm were as follows: (i) 46 mm distal to the center or the ulnar styloid to simulate dual-tab electrodes, (ii) over the MCP of the third finger, and (iii) at the base of the third finger nail bed (Fig. 1).¹⁵ The three drive electrode positions for the leg were as follows: (i) 46 mm distal from the center of the malleolus; (ii) mid-foot; and (iii) base of second and third toe (Fig. 2).^7,13 For measurements of the arm, the distal drive electrode was placed at the base of the second and third toe on the foot; for leg measurements, the second drive electrode was placed over the MCP of the third finger on the hand.

Five female participants without a history of lymphedema (Table 1) underwent resistance measures of their right arm and leg using the 3 different drive electrode positions on 2 occasions, ∼1 month apart, providing measurements from 20 limbs. The order of measurements taken was proximal to distal on the first occasion and then distal to proximal on the second occasion to reduce the biological impact of lying in supine.^21,22 The women were invited back for a second occasion to confirm that order of the measurements did not impact the resistance measures.

Three men and three women without history of lymphedema underwent resistance measurements of both arms and legs using the three different drive electrode positions to determine the effect of drive electrode position on interlimb BIS ratios. The order of the measurements was alternated to reduce the impact of lying in supine.^21,22 These men and women also participated in study F (Table 1).

Effect of electrode width and surface contact area on limb resistance

The resistances of the arms and legs of five participants without lymphedema (Table 1) were measured with three electrodes of different widths: bioimpedance electrodes (22 mm width); full-width cardiac electrodes (29 mm width); and half-width cardiac electrodes (14 mm width). The contact areas of the different width electrodes were as follows: bioimpedance 726 mm², full-width cardiac 957 mm², and half-width cardiac 462 mm². Drive electrodes were placed over the MCP on the hand and base of second and third toe on the foot, and measurement electrodes were positioned on the midpoint of ulnar styloid at the wrist and malleolus of the ankle for all 20 limbs measured.

Comparison of the limb resistance measured by bioimpedance and cardiac electrodes

Concordance between bioimpedance and half-width cardiac electrodes was examined in 30 participants, including those with a previous diagnosis of lymphedema (Table 1). Half-width cardiac electrodes were used to ensure at least 5 cm separation between the measurement and drive electrodes, and as a result of the findings of study D. The drive electrodes were placed over the MCP and base of the second and third toe. The measurement electrodes were placed in line with the ulnar styloid and between the malleolus. The electrodes of one type (i.e., bioimpedance or cardiac ) were attached and resistance of each of the four limbs measured. The electrodes were then replaced with the second electrode type and resistance was measured again. The type of electrode used first was alternated to reduce the biological impact of lying in supine.^21,22

Effect of mixing electrode batches and types on interlimb BIS ratios

Three men and three women without history of lymphedema underwent resistance measurements of both arms and legs using three combinations of electrode batches and types to determine effect of mismatched electrodes on interlimb BIS ratios (Table 1). Participant limb resistance was measured three times: once with electrodes from a single batch of bioimpedance electrodes, once when the electrodes on only the right hand and foot were replaced with electrodes from a different batch of bioimpedance electrodes, and once when the electrodes on only the right hand and foot were replaced with half-width cardiac electrodes, thus providing 12 interlimb BIS ratios for each electrode combination. The electrodes on the left hand and foot were from the initial single batch of bioimpedance electrodes and were not changed during the three measurements. The electrodes used first to replace the original electrodes on the right side were alternated to reduce the impact of lying in supine.^21,22

Data analysis

Resistance and reactance provided by the XCA device were used for comparison of electrode characteristics. For the BIS measurements in human participants, limb resistances, extrapolated from measured data to zero frequency (R₀) according to the Cole model,¹² were used. In accordance with measurement protocols for unilateral lymphedema assessment, interlimb BIS ratios were calculated in those without lymphedema: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \frac { { { \rm { R } } _0 } { \rm { non } } - { \rm { dom \;arm } } } { { { \rm { R } } _0 } { \rm { dom \;arm } } } } \,\, { \rm { and } } \,\, { \frac { { { \rm { R } } _ { 0 { \rm { \; } } } } { \rm { Right \;leg } } } { { { \rm { R } } _0 } { \rm { \;Left \;leg } } } } \end{align*} \end{document}

and in those with lymphedema (arms affected): \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \frac { { { \rm { R } } _0 } { \rm { unaffected \;arm } } } { { { \rm { R } } _0 } { \rm { affected \;arm } } } } \,\, { \rm { and } } \,\, { \frac { { { \rm { R } } _0 } { \rm { \;Right \;leg } } } { { { \rm { R } } _0 } { \rm { Left \;leg } } } } \end{align*} \end{document}

Technical error of measurement (TEM) was used to determine measurement imprecision resulting from intrinsic electrode resistance within electrode batches. Human BIS data were checked for normality using the Shapiro–Wilk test. Where data were normally distributed, a paired t-test determined differences among the data; Wilcoxon matched-pairs signed-rank test or Mann–Whitney was used where distribution was not normal. ANOVA, Kruskal–Wallis, and Friedmans tests were used to determine differences between related measures within measurement sets in each group of participants. Bland–Altman plots were used to evaluate the agreement among the different electrodes and width of electrodes; nonparametric tests were used when there were less than 10 measures, regardless of normality. Lin's concordance correlation coefficient was used to determine the extent to which data were in agreement. Data were analyzed using Prism 7 for Windows (version 7.01) and significance was set at p < 0.05.

Results

Determination of electrode resistance

Three test units from four production batches of both the cardiac and bioimpedance electrodes were tested twice to determine intrinsic electrode resistance. No significant difference in resistance was found between the first and second measurement on each test unit for both the bioimpedance (t₁₁ = 1.203 p = 0.25) and cardiac (Wilcoxon p = 0.87) electrodes. The within production batch TEM was clinically acceptable for all test units of electrodes, ranging between 0.06 and 1.09 Ω (0.1%–1.2%, Table 2). However, the electrode resistance varied markedly between production batches in both the cardiac and bioimpedance electrodes (p < 0.001). The cardiac electrode test units had higher resistance (median = 172.5 Ω, IQR = 160.9–252.8) than bioimpedance electrode test units (median = 98.0 Ω, IQR = 86.3–110.3; Mann–Whitney, p < 0.001). Nonelectrode instrumental error for impedance devices has previously been determined to be less than 0.5%.²³

Table 2.

Electrode Resistance (Ω) Stratified for Production Batch

	Bioimpedance Electrodes			Cardiac Electrodes
	Resistance	Reactance	TEM ^a	Resistance	Reactance	TEM ^a
Production batch 1, mean	84.9	1.37	1.046 (1.2)	177.6	1.83	0.158 (0.1)
Production batch 2, mean	90.5	1.67	1.087 (1.2)	158.6	2.20	0.372 (0.2)
Production batch 3, mean	101.7	1.80	0.058 (0.1)	279.1	2.97	0.261 (0.1)
Production batch 4, mean	114.1	2.90	0.119 (0.1)	155.8	2.37	0.204 (0.1)

Measurements made with the Impedimed XCA at 10 kHz. TEM for resistance was calculated using the data from the two tests completed on each of the three test units in each batch.

TEM, technical error of measurement, values in Ω (% of resistance).

Effect of electrode age on limb resistance

One production batch each of current and expired bioimpedance electrodes was used to measure limb resistance of six women. There was no difference in the limb resistance measured by the expired [mean (SD) = 331.0 (42.83) Ω] and current [331.1 (43.43) Ω] production batches of electrodes [t₍₂₃₎ = 0.193, p = 0.85].

Effect of drive electrode position on limb resistance

The three drive electrode positions on the foot provided significantly different limb resistance readings. Similar findings were found with the different drive positions on the hand with one exception: those completed with the drive electrode at the MCP and nail bed of the third finger had a 0.3 Ω nonsignificant difference in mean resistance (Table 3). Limb resistance was highest when using the most proximal drive electrode position and lowest at the most distal position for both the arm and foot (Table 3). Measurement bias caused by drive electrode position ranged from 0.5% to 2.3% in the leg and 0.1% to 0.9% in the arm (Table 3). Changing the position of the drive electrode had a greater effect on resistance in the leg than in the arm (p ≤ 0.001 for all drive electrode positions). Conversion of the raw resistance measures into clinically relevant BIS ratios removed the bias effect of drive electrode position in both arms (p = 0.15) and legs (p = 0.96). Drive electrodes were positioned over the MCP of the third finger and base of second and third toe for the remaining studies.

Table 3.

Comparison of Limb Resistance Determined Using Different Drive Positions on the Hand and Foot

Limb resistance obtained at three drive electrode positions using bioimpedance* electrodes*
	Arm			Leg
Drive electrode position	Hand 46 mm	MCP	Nail bed third finger	Foot 46 mm	Mid-foot	Base of third toe
Resistance (Ω), mean (SD)	339.3 (42.67)	336.6 (42.64)	336.3 (42.83)	288.0 (39.71)	281.4 (39.03)	281.36 (39.03)

Comparison of two drive electrode positions
Two position comparison	Difference between mean (SD) resistance (Ω) obtained at two positions	p (ANOVA)	Bias (%) from drive position (limits of agreement)	Concordance (confidence limits)
Hand 46 mm^a vs. MCP	2.69 (0.52)	<0.001	0.8 (1.66–3.71)	0.998 (1.00–1.00)
Hand 46 mm^a vs. Nail bed third finger	2.97 (0.80)	<0.001	0.9 (1.40–4.54)	0.997 (0.99–1.00)
MCP vs. Nail bed third finger	0.29 (0.64)	0.37	0.1 (−0.96 to 1.54)	1.00 (1.00–1.00)
Foot 46 mm^a vs. mid-foot	5.18 (0.94)	<0.001	1.8 (3.35–7.02)	0.991 (0.97–1.00)
Foot 46 mm^a vs. Base of third toe	6.66 (1.29)	<0.001	2.3 (4.13–9.19)	0.985 (0.95–1.00)
Mid-foot vs. Base of third toe	1.48 (0.54)	<0.001	0.5 (0.42–2.54)	0.999 (1.00–1.00)

Forty-six millimeter between center of the measurement and drive electrodes to simulate dual-tab electrode spacing.

MCP, metacarpal phalangeal joint.

Effect of electrode width and surface contact area on limb resistance

Full-width bioimpedance electrodes, and full-width cardiac and half-width cardiac electrodes were used to measure limb resistance. Repeated-measures ANOVA showed a small, but significant difference between the limb resistance measured by the three different width electrodes (p ≤ 0.001). The resistance differences were greatest when the full-width cardiac electrode was compared with the half-width cardiac electrode and the bioimpedance electrode (Table 4). Concordance between electrodes was highest between the bioimpedance and half-width cardiac electrodes, with no statistical difference found between resistance measured by these two widths (p = 0.10). Calculation of interlimb BIS ratios from the limb resistances removed the effect of electrode size as there was no difference between the interlimb BIS ratios of any size electrode (p = 0.60). The bioimpedance and half-width cardiac electrodes were therefore used in study E.

Table 4.

Comparison of Resistance Determined from Different Width Electrodes

	Mean resistance in Ω (SD)	Mean resistance in Ω (SD)
N	Larger electrode	Smaller electrode	Width difference (mm)	p	Bias (%) (limits of agreement)	Concordance (confidence limits)
	Full-width cardiac	Half-width cardiac
20	327.8 (42.4)	325.2 (42.7)	15	<0.001	−0.8 (−7.00 to 1.88)	0.997 (0.99–1.00)
	Full-width cardiac	bioimpedance
20	327.8 (42.4)	324.4 (43.5)	7	<0.001	−1.0 (−9.79 to 3.01)	0.994 (0.98–1.00)
	bioimpedance	Half-width cardiac
20	324.4 (43.5)	325.2 (42.7)	8	0.10	−0.3 (−4.16 to 2.50)	0.999 (1.00–1.00)

Electrode width: bioimpedance 22 mm, Full-width cardiac 29 mm, half-width cardiac 14 mm.

Electrode contact area: bioimpedance 726 mm², Full-width cardiac 957 mm² and half-width cardiac 462 mm².

Comparison of the limb resistance measured by bioimpedance and cardiac electrodes

The bioimpedance and half-width cardiac electrodes were used to measure limb resistance of 30 participants, 4 of whom had upper limb lymphedema. There was no difference between the resistance measured by the bioimpedance and half-width cardiac electrodes in either the arms (p = 0.197) or legs (p = 0.173). Concordance was high between the two types of electrodes with minimal bias (Table 5).

Table 5.

Comparison of Resistance and Bioimpedance Spectroscopy Ratio Determined from Bioimpedance and Half-Width Cardiac Electrodes from 30 Participants

	Number of measures	Bioimpedance electrodes, median (IQR)	Half-width cardiac* electrodes, median (IQR)*	p	Bias (%) (limits of agreement)	Concordance (95% confidence interval)
Arm resistance (R₀), Ω	60	304.9 (272.9–359.9)	304.2 (270.9–361.0)	0.197^a	0.1 (−4.27 to 5.07)	0.999 (1.00–1.00)
Leg resistance (R₀), Ω	60	281.4 (246.6–317.4)	282.4 (247.7–315.4)	0.173^b	0.2 (−5.28 to 6.55)	0.997 (1.00–1.00)
Arm BIS ratios	30	1.029 (1.008–1.081)	1.041 (1.005–1.080)	0.247^b	−0.2 (−0.02 to 0.02)	0.990 (0.98–1.00)
Leg BIS ratios	30	1.009 (0.981–1.041)	1.024 (0.986–1.042)	0.269^a	−0.2 (−0.02 to 0.02)	0.966 (0.93–0.98)

t-Test.

Wilcoxon.

BIS, bioimpedance spectroscopy.

The resistance data from the bioimpedance and half-width cardiac electrodes were used to calculate 30 interlimb arm and 30 interlimb leg BIS ratios. There was no difference in interlimb BIS ratios obtained from bioimpedance and half-width cardiac electrodes for either the arms or legs (p = 0.247 and p = 0.289, respectively; Table 5).

Effect of mixing electrode batches and types on interlimb BIS ratios

The effect of combining electrode batches and electrodes types on an individual participant was assessed by changing the electrodes on one limb to a different batch or type of electrode so that two different types or batches of electrodes were used to determine the interlimb BIS ratios. The use of combinations of electrode types and batches did not significantly affect the interlimb BIS ratio for either arms (p = 0.43) or legs (p = 0.74).

Discussion

The importance of electrode characteristics and placement in resistance measurements has not been widely considered.¹⁷ This study examined a number of factors that could impact bioimpedance measurement, but found only drive electrode position and electrode size made a significant difference on human absolute limb resistance measurements. The magnitude of these differences was small and could be minimized by adopting consistent BIS protocols or calculation of interlimb BIS ratios that are typically used in lymphedema assessment to reduce the impact of weight and body fluid composition changes.

The intrinsic resistance of the cardiac electrodes was significantly higher compared with the bioimpedance electrodes. This result is similar to the finding of Nescolarde et al.¹⁷ who compared nine different types of Ag/AgCl electrodes and found significant differences in electrode resistance. They found that the electrodes with the highest intrinsic resistance provided different bioelectrical impedance measures from the electrodes with the lowest intrinsic resistance when used to assess body composition.¹⁷ Ideally, intrinsic resistance should be as low as possible and the same for all electrodes. Nescolarde et al.¹⁷ have suggested that intrinsic impedance of <100 Ω at 50 kHz is acceptable for impedance measurements. In the study, the intrinsic electrode resistance for bioimpedance electrodes was less than 115 Ω measured at 10 kHz. Although the intrinsic electrode resistance was greater for cardiac electrodes, this did not affect either the limb resistance or the interlimb BIS ratios derived from the bioimpedance and half-width cardiac electrodes. The difference in these findings could be due to the measure reported: Nescolarde et al.¹⁷ used a height-normalized bioimpedance vector derived from the right half of the body, which incorporates both resistance and reactance, whereas this study measured resistance (R₀) of only the limbs as is commonly done for lymphedema assessment. In addition, the use of interlimb BIS ratio appears to account for variation in resistance between electrode batches, and in this study, between electrode types, providing internal consistency in interlimb BIS ratios regardless of electrode type.

Hypothetically, electrode age could also impact on the accuracy of resistance measures in the clinical environment. However, we found that using electrodes, which were almost 2 years past their expiration date and stored according to manufacturer's instructions did not impact on the accuracy of limb resistance measures. This finding should provide confidence to continue use of appropriately stored electrodes, at least up to twice their “lifetime,” an issue that may be of particular relevance in clinical situations with low rates of resistance measurement, which may prevent electrodes being used before the recommended use-by date.

Measurement error can be introduced if the measurement electrodes is not reliably positioned. For example, a change in position as small as 5 mm in both measurement electrodes (1 cm overall) can produce errors of 2%.¹⁴ We found that the position of the drive electrode, in relation to the measurement electrode, also impacts on limb resistance, with limb resistance decreasing as the drive electrode was moved further away from the measurement electrode. The decrease in resistance could be due to reducing interference between electrodes as the distance between the electrodes increased. The magnitude of displacement of the drive electrode in this study was greater than the displacement of the measurement electrode assessed in previous studies,¹⁴ but still produced differences in the resistance measured of up to 1% in the arm and 2% in the leg when comparing the proximal and distal drive electrode positions. This effect, while small and arguably practically insignificant compared to errors caused by measurement electrode placement, could potentially result in clinically relevant accuracy issues when accumulated with other small errors. To limit the cumulative effect of small errors in electrode placement and improve repeatability, a consistent approach to placement of all electrodes on each limb, for each measurement and over time, is recommended. Alternately, use of dual-tab electrodes that standardize this distance will minimize this error. However, assuming consistent positioning of electrodes within individual BIS measurements, any variation of position of the drive electrodes over time will be mitigated by conversion of absolute resistance measures into the clinically relevant interlimb BIS ratios, thereby contributing to the reliability of BIS for ongoing monitoring of lymphedema.

Electrode width and surface contact area impacted limb resistance. Differences in resistance were identified within (full-width and half-width cardiac ) and between electrode types (bioimpedance and cardiac ). It is unknown whether this difference was due only to electrode surface area, which has been demonstrated to impact measurement of R₀ by up to 3.5% when two different size electrodes were used at the same time,²⁴ or if distance between the measurement and drive electrodes contributed. Owing to the size of the electrodes, the distance between the full-with cardiac measurement and drive electrodes was less than the distance between either of the bioimpedance and half-width cardiac electrodes; the reduced distance between the full-width cardiac electrodes may have impacted resistance measures, as found when changing the distance between the measurement and drive electrodes. Conversion of the absolute limb resistance measurements to interlimb BIS ratios eliminated the difference between the various electrode sizes, with all three electrode sizes providing similar interlimb BIS ratio results when a consistent placement protocol was used.

Interestingly, and as highlighted above, conversion of absolute limb resistance into clinically relevant interlimb BIS ratios appears to resolve many of the issue that may impact accuracy and reliability of BIS measurements in the lymphedema clinic. Not only does the interlimb BIS ratio account for changes in body composition and weight but also negates the effect of inherent electrodes differences, electrode size, and position of the drive electrode, and allows electrodes from different batches and in this case, different types of electrodes, to be mixed during individual measurements. The utility of the interlimb BIS ratio provides confidence in the accuracy of BIS for monitoring and assessment in the lymphedema clinic.

While there are a number of factors that could impact limb resistance measurements, this study has shown that the factors considered in this study have either no effect on the measurement of human limb resistance or small effects that could be overcome with consistent BIS protocols or conversion of limb resistance measurements into more clinically useful interlimb BIS ratios. Attention to detail at a technical level, and specifically electrode placement, can prevent cumulative bias in results. A consistent and easily followed BIS protocol enables cardiac electrodes to be used as an alternative to instrument-specific bioimpedance electrodes in the device tested. It would be prudent, however, for users to confirm these findings if using a different impedance device and any nondevice-specific electrode.

Conclusion

The interlimb BIS ratio commonly used in lymphedema practice provides a robust measure of limb edema that overcomes potential sources of error in the clinic. Ag/AgCl cardiac electrodes can be used to accurately measure limb resistance and calculate interlimb BIS ratios, are commonly available in health settings, and could provide an alternative source of electrodes for use in the assessment and monitoring of persons with or at risk of developing unilateral limb lymphedema. Furthermore, development of protocols for BIS use in clinical practice, specifically regarding positioning and storage of electrodes, can only enhance accuracy of results, regardless of type of electrode used.

Footnotes

Authors' Contributions

B.J.S. contributed to the study conception and design, data collection, analysis and interpretation of data, and article preparation; E.S.D., L.C.W., and S.L.K. contributed to the study conception and design, analysis and interpretation of data, and article preparation.

Author Disclosure Statement

Authors B.J.S., E.S.D., and S.L.K. have no competing financial interests. L.C.W. consults to Impedimed Ltd. Impedimed had no involvement in the design, undertaking, or article preparation of this study.

References

Kim

, Jeon

, Sung

, Jeong

, Do

, Kim

. Prediction of treatment outcome with bioimpedance measurements in breast cancer related lymphedema patients. Ann Rehabil Med, 2011; 35:687–693.

, Cleland

, Guth

, et al. L-dex ratio in detecting breast cancer-related lymphedema: Reliability, sensitivity, and specificity. Lymphology, 2013; 46:85–96.

Dylke

, Schembri

, Bailey

, et al. Diagnosis of upper limb lymphedema: Development of an evidence-based approach. Acta Oncol, 2016; 55:1477–1483.

Kyle

, Bosaeus

, De Lorenzo

, et al. Bioelectrical impedance analysis—Part I: Review of principles and methods. Clin Nutr, 2004; 23:1226–1243.

Cornish

, Chapman

, Hirst

, et al. Early diagnosis of lymphedema using multiple frequency bioimpedance. Lymphology, 2001; 34:2–11.

Kyle

, Bosaeus

, De Lorenzo

, et al. Bioelectrical impedance analysis—Part II: Utilization in clinical practice. Clin Nutr, 2004; 23:1430–1453.

Ward

, Kilbreath

, Cornish

. Bioelectrical impedance analysis for early detection of lymphedema. In: Weissieder H, Schuchhardt C, eds. Lymphedema: Diagnosis and Therapy, 4th ed. Essen: Viavital Verlag;, 2008:502–517.

Cornish

, Chapman

, Thomas

, Ward

, Bunce

, Hirst

. Early diagnosis of lymphedema in postsurgery breast cancer patients. Ann N Y Acad Sci, 2000; 904:571–575.

Ward

, Dylke

, Czerniec

, Isenring

, Kilbreath

. Confirmation of the reference impedance ratios used for assessment of breast cancer-related lymphedema by bioelectrical impedance spectroscopy. Lymphat Res Biol, 2011; 9:47–51.

10.

Hoffer

, Meador

, Simpson

. Correlation of whole-body impedance with total body water volume. J Appl Physiol, 1969; 27:531–534.

11.

Cornish

, Thomas

, Ward

. Effect of temperature and sweating on bioimpedance measurements. Appl Radiat Isot, 1998; 49:475–476.

12.

Ward

, Essex

, Cornish

. Determination of Cole parameters in multiple frequency bioelectrical impedance analysis using only the measurement of impedances. Physiol Meas, 2006; 27:839–850.

13.

Cornish

, Jacobs

, Thomas

, Ward

. Optimizing electrode sites for segmental bioimpedance measurements. Physiol Meas, 1999; 20:241–250.

14.

Elsen

, Siu

, Pineda

, Solomons

. Sources of variability in bioelectrical impedance determinations in adults. International Symposium on In Vivo Body Composition Studies. Upton, New York, 1986.

15.

Dylke

, Alsobayel

, Ward

, Liu

, Webb

, Kilbreath

. Use of impedance ratios to assess hand swelling in lymphoedema. Phlebology, 2014; 29:83–89.

16.

Lingwood

, Coghlan

, Ward

, Charles

, Colditz

. Measurement of extracellular fluid volume in the neonate using multiple frequency bio-impedance analysis. Physiol Meas, 2000; 21:251–262.

17.

Nescolarde

, Lukaski

, de Lorenzo

, de-Mateo-Silleras

, Redondo-del-Rio

, Camina-Martin

. Different displacement of bioimpedance vector due to Ag/AgCl electrode effect. Eur J Clin Nutr, 2016; 70:1401–1407.

18.

ImpediMed Ltd. ImpediMed Imp XCA Exracellular Fluid Analysis. Available at https://www.accessdata.fda.gov/cdrh_docs/pdf5/K050415.pdf (accessed September 18, 2017).

19.

ImpediMed. Imp™ SFB7. Intelligent Impedance Instruments. Pinkenba QLD: ImpediMed Limited; 2005.

20.

Ward

, Dylke

, Czerniec

, Isenring

, Kilbreath

. Reference ranges for assessment of unilateral lymphedema in legs by bioelectrical impedance spectroscopy. Lymphat Res Biol, 2011; 9:43–46.

21.

Thomas

, Cornish

, Ward

, Patterson

. A comparison of segmental and wrist-to-ankle methodologies of bioimpedance analysis. Appl Radiat Isot, 1998; 49:477–478.

22.

Gibson

, Beam

, Alencar

, Zuhl

, Mermier

. Time course of supine and standing shifts in total body, intracellular and extracellular water for a sample of healthy adults. Eur J Clin Nutr, 2015; 69:14–19.

23.

Ward

. Inter-instrument comparision of bioimpedance spectroscopic analysers. Open Med Dev J, 2009; 1:3–10.

24.

Buendia

, Bogonez-Franco

, Nescolarde

, Seoane

. Influence of electrode mismatch on Cole parameter estimation from total right side electrical bioimpedance spectroscopy measurements. Med Eng Phys, 2012; 34:1024–1028.