Assessment of the Intraobserver and Interobserver Reliability of a Communicating Vessels Volumeter to Measure Wrist-Hand Volume
===============================================================================================================================

* Rogério Mendonca de Carvalho
* Maria del Carmen Janerio Perez
* Fausto Miranda, Jr

## Abstract

****Background**** Traditional volumetry based on Archimedes' principle is the gold standard for the measurement of limb volume, but the routine use of this technique is discouraged because of several disadvantages.

****Objective**** The purpose of this study was to evaluate intraobserver and interobserver reliability of direct measurements of wrist-hand volume using a new communicating vessels volumeter based on Pascal's law.

****Design**** A reliability study was conducted.

****Methods**** To evaluate the reliability of the communicating vessels volumeter in generating measurements, 30 hands of 15 participants (9 women, 6 men) were measured 3 times each by 3 observers, totaling 270 volumetric results.

****Results**** Measurement time was short (X̄ =3 minutes 42 seconds). The intraclass correlation coefficient (ICC) was .9977 for observer 1 and .9976 for observers 2 and 3. The interobserver ICC was .9998. The standard error of measurement was about 3 mL for all observers; the interobserver result was 1 mL. The interrater coefficient of variance (CV) was 1.15% for the series of 9 measurements collected for each segment; the intrarater CV was 1.20%.

****Limitations**** No swollen hands were measured, and measurements were not compared with the gold standard technique. Thus, accuracy of the new volumeter was not determined in this study.

****Conclusion**** A new device has been developed for plethysmography of the extremities, and the results of its use to measure the volume of the wrist-hand segment were reliable in both intraobserver and interobserver analyses.

Hand volume measurements are particularly useful for patients with burns, rheumatoid arthritis, or generalized lymphedema, or after mastectomy, soft tissue injury, or surgery.1 Also, hand volumetry can be used for patients with traumatic lesions, edemas, deformities, or anomalies. When persistent, hand edema may slow the recovery process and can be associated with different comorbidities. Moreover, it may produce a state of functional incapacity by limiting muscular elasticity, decreasing the articular range of motion, shortening the aponeurosis, or even leading to tissue necrosis. This situation requires early detection and systematic monitoring and control so that treatment approaches can be reviewed and undesirable consequences avoided.2

Several measurement techniques can be used in hand therapy (eg, circumference measurements,3–5 the figure-of-eight method,6,7 volumetry,1,8,9 bioelectrical impedance,10–13 computer programs1 that record the measurements). Use of the Perometer (Juzo, Wuppertal, Germany) recently has been suggested as a simple and quick alternative method for measuring limb volume. It is an optoelectronic imaging device that has been designed specifically for measuring limb volume, circumference, contour, and cross-sectional area.8 However, perometry measurement is based on the assumption of a circular or elliptical cross-section. The hand cannot be reliably measured because it deviates markedly from the circle or ellipse in cross-section. Other drawbacks of the device are its relatively high cost and large size, which hinder measurement outside the clinic and preclude bedside use.14–16 In relation to the traditional overflow method, the design of most volumeters, in which the outflow spout is located below the top,5 makes it difficult to immerse the hand so that the water level aligns with the line drawn at the wrist.1 Indeed, the major reason for not including hand volume as part of routine assessments is the absence of a readily available and clinically feasible method to estimate hand volume. Thus, there is a clear need for an alternate method of estimating hand volume.4

A new device, called the “communicating vessels volumeter” (CVV), has been developed for use in plethysmography.17 Its operation is based on Pascal's law and the hydrostatic paradox (Stevin's law), which associates pressures to water columns of different surfaces. The design and physical principles involved in its development, in theory, may eliminate all the inhibiting factors of the traditional water displacement method. The technical features of the CVV have already been presented in a previous publication17; therefore, the hypothesis tested in this study was that the new device would generate reliable intraobserver and interobserver measurements of the volumes of wrist-hand segments evaluated.

The purpose of this study was to evaluate intraobserver and interobserver reliability of preliminary measurements of wrist-hand volume using a small-scale prototype of a novel CVV.

## Materials and Method

### Participants

Screening interviews and physical examinations with a physical therapist were conducted for the selection of the population of volunteers according to the inclusion and exclusion criteria. Individuals who were aged 18 to 60 years, had a body mass index below 30 kg/m2, were in good health on the day of evaluations, and agreed to participate were included in the study. Individuals who were using corticosteroids or had cardiac insufficiency, any type of lymphedema in any part of the body, kidney insufficiency or any other problem that might affect circulation or blood pressure (not controlled with medication), moderate to severe mycosis of the fingernails or interdigital spaces of the hands, rheumatic or chronic inflammatory disease that could affect the acral joints, infection, or latex allergy were excluded.

Fifteen people (9 women and 6 men) were selected as study participants, and 30 nonedematous hands were evaluated 9 times each: 3 observers measured each wrist-hand segment 3 times, which generated 270 volume measurements for the study. The participants' ages ranged from 20 to 47 years (X̄=26, SD=6.9). Their height ranged from 1.55 to 1.73 m (X̄=1.63, SD=0.06) for women and from 1.69 to 1.81 m (X̄=1.73, SD=0.05) for men. Their weight ranged from 49 to 66 kg (X̄=58.6, SD=6.1) for women and from 80 to 93 kg (X̄=83.8, SD=5.1) for men. Body mass index ranged from 19.57 to 24.24 kg/m2 (X̄=21.82, SD=1.59) for women and from 24.42 to 29.76 kg/m2 (X̄=27.88, SD=1.88) for men. All participants were right-handed. All volunteers agreed to participate in the study and signed an informed consent statement approved by the Ethics in Research Committee of the Federal University of São Paulo.

Three observers were invited to participate: 2 physical therapists and a third-year undergraduate physical therapist student, all with no previous personal or academic connections with this study or the work of the main author. All observers were trained to become familiar with the procedure at 2 time points: 2 days before the study and 1 hour before the beginning of the measurements on each of the 2 examination days.

### Procedure

A glass prototype of the CVV, made to order (Hermex Indústria e Comércio de Artigos de Vidro para Laboratório Ltda, São Paulo, Brazil) according to the technical design of the patent seekers, was used for this study. It is composed of 3 communicating vessels: an immersion chamber for the segment, a volumetric column, and a balloon for the suction of air residues. The components formed a hermetic system for the evaluation of the segment, which was protected by a glove to be attached to the immersion basin; the opening to introduce the hand then was sealed.

Two rings manufactured from polyvinyl chloride (PVC) and acrylic, one chamfered for the fixation of the border of the glove and the other to restrict the upper expansion of the glove, were used as coupling devices at the upper edge of the immersion basin. A cylindrical device manufactured in aluminum with an upper winding platform elevating to 5.5 cm from the bottom of the immersion basin was fixed to the inside of this base with adhesive foam to serve as a baseline reference to limit the submersion of the wrist-hand segment. Sterilized surgical gloves (New Hand, Lemgruber, Avenida das Américas, Rio de Janeiro, Brazil) were used as the interface between the segment evaluated and the water in the immersion basin. To standardize the volume added by the gloves, all volunteers wore #8.0 gloves (medium large) regardless of the size of the hands examined. An analogical submersible thermometer was fixed with a suction cup to the internal wall of the immersion basin to monitor water temperature during the trials. The vacuum source was a 70-W portable tracheal aspirator (Aspiramax, NS Indústria de Aparelhos Médicos Ltda, São Paulo, Brazil) for home use. A level was used on the table where the volumeter was placed during the tests to ensure the horizontal position of the hydrostatic levels. To record the ending time for each measurement, a large digital clock was fixed on the wall above the table where the volumeter was placed. The water inside the immersion basin was kept within the thermal interval defined by Boland and Adams18 by using a portable electrical heater.

All of the 270 evaluations were performed on 2 days. Four volunteers (3 women and 1 man) were evaluated on the first day, and the other 11 (6 women and 5 men) were evaluated on the second day. All measurements were made in the order of volunteer arrival. The volunteers were previously instructed not to ask the evaluators about the volume of their hands during or after examination, to keep their gaze fixed ahead during evaluations, and not to look at the volumetric column. The observers were trained to perform the measurements, and each session consisted of 1 measurement for each observer for each wrist-hand segment (right and left) for the same volunteer.

At the beginning of the evaluation, the volunteer was placed in an orthostatic position next to the volumeter, without any accessories such as watches, bracelets, or rings on their hands, wrists, or fingers. The observer asked the volunteer to put the #8.0 sterile glove on the hand under examination. The baseline reference for all measurements was 7 cm.

At this point, the observer asked the volunteer to submerse the segment partially so that the edge of the glove could be fixed into the chamfer in the fixation ring previously connected to the upper edge of the immersion basin. The edge ring with its acrylic blades open then was coupled to the upper edge of the larger cylinder over the glove and above the fixation ring. The air remaining between the glove and the basin water was removed by activating the vacuum generator connected to the top of the spherical balloon by a latex tube.

The wrist-hand segment was submersed until the distal extremity of the third finger touched the center of the upper platform of the baseline reference. Figure 1 illustrates the process of immersion of the hand into the device filled with water and removal of residual air using a tracheal aspirator. Volume was read always in the same order: first observer 1, then observer 2, and finally observer 3.

![Figure 1.](http://demo.highwire.org/https://demo.highwire.org/content/hwdjpt/92/10/1329/F1.medium.gif)

[Figure 1.](http://demo.highwire.org/content/92/10/1329/F1)

Figure 1. 
Set of photographs displaying the hand inside the communicating vessels volumeter (CVV) filled with water: (A) introduction of the hand into the immersion basin, with the edge of the glove fixed into the chamfer in the fixation ring (note the upper expansion of the glove, because the edge ring has not yet been positioned); (B) positioning of the edge ring with its acrylic blades open prior to the complete submersion of the hand; (C) appearance of the glove before removal of residual air (note the unevenness of the glove); (D) process of suction of residual air; and (E) a close-fitting glove after removal of residual air (one ring was deliberately left on the third finger to better illustrate the phenomenon)

Between measurements of the same wrist-hand segment by observers 1 and 2 and by observers 2 and 3, the volunteers were asked to pull the hand up to the upper edge of the immersion basin and then move it back slowly to submerse it to the same baseline reference position and resume the same process from the “zero” point of evaluation.

### Data Analysis

The sample size of 15 participants was determined according to the total number of samples used in major studies that calculated reliability for traditional water displacement plethysmography.9,19–21 Demographic data and descriptive measurements are presented as mean and standard deviation. The standard error of measurement (SEM)22 was determined for later comparisons of variables (in milliliters). A graph for the results of measurements was built according to volume (and its respective means) collected by each observer from each wrist-hand segment per participant.

A mixed linear model23,24 was applied to test method reliability, taking into account logarithmic transformation of the data and interference from other variables (observer, repetition, sex, side, time, temperature, and sex-side interaction). In this model, the logarithm of volume was used to stabilize variance and minimize data asymmetry because volume measurements are positive variables and follow gamma distribution.25 Heteroscedasticity was tested after logarithmic transformation. All other reliability calculations—SEM, intraclass correlation coefficient (ICC), and coefficient of variance (CV)—were performed based on nonlogarithmic measurement values (ie, expressed as milliliters).

#### Intrarater reliability.

The 2-way mixed-effects model of the ICC was used to measure intraobserver reliability (intrarater ICC). Values closer to 1 indicate better results. Intrarater reliability also was analyzed according to percentage CV, namely standard deviation of the mean (intrarater CV%). Differences in measurements between volunteers for the same observer, under this perspective, were classified as a group (observer 1 group, observer 2 group, and observer 3 group). The intrarater CV% of the 3 measurements made by each observer was calculated for each of the 30 segments, and means of results represented the intrarater CV% generated by the tests conducted by the same observer.

#### Interrater reliability.

The ICC also was used to measure interobserver reliability (interrater ICC). Interrater CV% was calculated for each wrist-hand segment of each participant. The standard deviation of all volumes calculated for the same segment divided by the mean of these measurements represented the interrater CV% values found in the device tests.

## Results

The length of the first tests ranged from 1 to 7 minutes (X̄=3 minutes 42 seconds, SD=1 minute 18 seconds) from the end of the preceding procedure to the reading of the value on the volumeter column during the current measurement. This calculation also took into consideration the time to tare the device, the time to prepare the participant to put on the glove and to adjust the glove to the edge of the basin, and the time for the suction of the remaining air between the glove and the water with the aspirator and the waiting time to reestablish the hydrostatic balance among the 3 communicating vessels. The duration of the procedures that did not require taring or preparation, in which time was recorded from the end of the preceding procedure to the reading of the volumetric column during the current procedure, ranged from 0.5 to 2 minutes (X̄=42 seconds, SD=18 seconds).

Hydrostatic level was low (300 mL) for 8 volunteers, medium (350–400 mL) for 5, and high (450 mL) for 2. In all 15 volunteers, hydrostatic level was the same for the 2 sides of each volunteer. Two attempts at measuring the correct hydrostatic level were necessary only for the right hands of 2 volunteers on the first evaluation day. For all other segments, correct hydrostatic levels were achieved in the first attempt. Minimum water temperature during trials was 24°C, and maximum temperature was 28°C (X̄=26.1, SD=1.1).

Baseline references were set at 7 cm for all participants except 1 volunteer, for whom the baseline reference was 5.5 cm. For this participant, the water level of the volumetric column did not reach the graded scale when the baseline reference of 7 cm was used, even at a hydrostatic level of 300 mL, which showed that the volume of each of her hands was below 300 mL. The smallest value measured was 300 mL, and the largest value measured was 512 mL (X̄=374, SD=62). The Table shows intraobserver and interobserver mean measurement values, standard deviations, intrarater and interrater CV% values, and SEM values according to the unit of volumetric measurement.

View this table:
[Table.](http://demo.highwire.org/content/92/10/1329/T1)

Table. 
Mean Values (SD) and Intraobserver and Interobserver Values According to Percent Coefficient of Variation (%CV) and Standard Error of Measurement (SEM)

The ICCs at a 95% confidence interval (95% CI) calculated by the intrarater analysis were .9977 for observer 1 and .9976 for both observers 2 and 3. The interrater ICC was .9998.

Figure 2 is a scatter plot graph showing the participant number (1–15) (on the x-axis) and the volume measured by each observer (on the y-axis) of both wrist-hand segments (total of 6 measurements, 3 on each side) of each volunteer.

![Figure 2.](http://demo.highwire.org/https://demo.highwire.org/content/hwdjpt/92/10/1329/F2.medium.gif)

[Figure 2.](http://demo.highwire.org/content/92/10/1329/F2)

Figure 2. 
Volume measurements for each participant by observer (3 measurements per hand). Each measurement is represented by an “X,” and mean measurement values from the same participant by observer are represented by an “O” (bold circle) (generated in R environment, R free statistical software, available at [http://www.r-project.org](http://www.r-project.org)).

The application of the mixed linear model revealed no differences among the means for the observers or among repetitions when *P*<.05. The calculation of heteroscedasticity showed that the sets of volumetric measurement logarithms in this study were nonheteroscedastic (ie, their variances were statistically equal).

## Discussion

The CVV is a device whose operating principles are intended to overcome technical disadvantages related to the clinical applicability of volumetry, and the intrarater and interrater reliability results, standard deviations, and SEM values for CVV obtained in this study were comparable to those previously reported in the literature for hand volumetry using the overflow method.2,6,8,21,24

The ICC for a 95% CI was significant26 for both intrarater and interrater analyses. Interrater ICC is one of the grounds for criticism of water displacement volumeters, being lower than that for intrarater analysis according to Deltombe et al.27 The clinical relevance of this fact lies in the need to develop measurement methods that are less dependent on the rater, ideally allowing different observers to obtain very similar values for the same hand evaluated. Thus, patients with altered hand volume could have similar prognoses and readings of treatment outcomes, regardless of the attending health professional. For the CVV under study here, interrater ICC was high and as statistically significant as the individual ICC. Interrater ICC was numerically greater than intrarater ICC, which may reflect changes occurring over time in the volume of the segment evaluated due to situations such as the participant's resting position between measurement sessions. As previously reported by some authors, volume normally varies over time, increasing in hands that are resting in favor of gravity at the intervals between assessments.18,28,29 This phenomenon should be considered in future studies, and its impact may be minimized by standardizing a resting position between assessments that supports the segment under evaluation against gravity.

In a study9 in which the volume of 30 nonedematous hands was measured using the overflow method by 3 clinicians with 3 to 8 years of experience in hand volumetry, without a standardized protocol, and the findings compared with the measurements of 30 nonedematous hands by 3 undergraduates with minimal knowledge of hand volumetry but using a standardized protocol, SEM and intrarater and interrater ICC results were more consistent in the undergraduate group. In the present study, which also used raters inexperienced in the method under study and an assessment protocol and guidelines, SEM and intrarater and interrater ICC results were consistent with those of the group using a standardized protocol in the study by Farrell et al.9 Once the CVV begins to be used as a valid alternative to conventional methods in clinical practice, future research might elucidate whether experience will eliminate the use of strict standardization in assessment protocols to produce reliable measures.

Other authors,6,29 who did not perform a third measurement (retest) or used only control hands in larger samples,2 found higher SEM values and numerically lower ICC values for the overflow method than those reported for the CVV in this study. These findings probably were due to the fact that, in larger samples, statistical tests become more sensitive to volumetric differences of the method under analysis and to the need for a third measurement when a more accurate analysis of test reliability is desired.

The international standard to evaluate the reliability of anthropometric measurements recommends that 3 nonconsecutive measurements should be made by 3 different observers.30 This standard, used in the present research, has been reported as acceptable for the validation of measurements in other hand volumetry trials5,9 and should be adopted in future studies that analyze agreement and accuracy between the CVV and the overflow method.

Good reliability in similar samples also was found for hand circumference measurements6,24 using the figure-of-eight method, a technique as inexpensive as the price of a tape measure, which involves a 1-dimensional measurement of the segment under evaluation. According to Pellechia,6 the figure-of-eight method is more time-efficient than the gold standard of volumetry. The tape measure method requires less than 1 minute, whereas volumetry takes several minutes to be completed. However, hand circumference measurements, although more time-efficient and requiring less equipment, have not always produced acceptable intraobserver and interobserver reliability results.7

Evaluation times using the CVV were shorter than those reported for traditional volumetry (about 10 minutes) and close to those found for circumferential measurements.2,6,7 The short time needed for this technique was attributed, mostly, to the fact that observers did not have to worry about possible water losses due to abrupt flow and did not need to ensure slow immersion of the segment until the liquid stabilized, as recommended for the overflow method.1,3,4,8,21,28,31,32 Therefore, the hermetic CVV system allows the relatively quick introduction of the hand in the immersion basin and immediate volumetric reading because the displaced liquid does not have to be transferred to other containers.

The glove protected the hand of volunteers from direct contact with water,33,34 a factor that may favor the indication of the method for segments with open or dressed wounds and venous ulcers.4 Moreover, the technique does not require constant replenishments between assessments, as in the case of overflow methods.35 The CVV method was considered to be hygienic, without water loss between assessments, as reported for the traditional water displacement method.36,37 Initial instructions for its use were easy to follow and did not cause any embarrassment to the participants or observers. In addition, the glove provided a potential safeguard against cross-transmission of infections.31

In this study, intrarater and interrater CV% also showed low results. The low variability of intrarater measurements is well illustrated in Figure 2, with the volume of both hands of each volunteer close to the mean value (slightly greater or less than, depending on which side was evaluated). Likewise, the nearly specular heights of the points plotted for each volunteer across the 3 graphs in Figure 2 suggest very close interrater measurements. Another study27 showed that intrarater CV% was significantly lower for optoelectronic volumetry (1.5%) and the disk model method (1.9%) than for water displacement volumetry (2.9%) and the frustum sign method (3.2%). The interrater CV% values were 1.7%, 3.1%, 4.5%, and 4.8%, respectively, for these methods. A recent study2 using overflow measurements of 200 nonedematous hands demonstrated an interrater CV% of 17.65%. The principle of indirect measurement, inherent in all methods, may explain the greater intraobserver variance values, as measurements strongly depend on the observer, except in optoelectronic volumetry.

The results of the mixed linear model, with log-transformed measurement values, confirmed that there were no statistically significant differences at *P*<.05 among the repeated measurements by each observer or among the measurements made by different observers. These results indicate that CVV is a potentially reliable technique for hand volume measurement, although further tests with edematous hands should be conducted to confirm its applicability as a safe option in clinical practice.

In this study, the water temperatures used in the tests did not affect the volume of segments measured.5–9,18 Furthermore, the ages of the 15 participants with nonedematous hands were similar to those in other trials analyzing the reliability of volumetric measurements.2,6,7,9

The resolution of our CVV prototype was ±1 mL. Devices that combine weighing of the immersion system1 may ensure better resolutions at ±0.01 g, a value that a volumetric column would achieve only if it were imponderably high and thin, and its design and manufacturing would be infeasible due to its extreme fragility. In addition, the hybrid device is significantly more expensive than a typical volumetry set, which can be purchased for a few hundred dollars. A complete, commercially available system to weigh displaced volume may cost about US$2,400, which is much less than the price of an optoelectronic Perometer (about US$50,000),1,20 but more than that of a CVV, whose cost for this study, which included its individualized manufacturing out of large-scale production, was about US$700.

Open systems require that the immersion basin be replenished1–9,15,16,18–21,27,31,33–41 for the measurements of the same segment by different observers. In the case of the CVV, however, water does not flow out, but rather goes up in the graded column. Volumetric readings in the CVV are provided immediately and for each portion of the immersed segment, which is similar to what occurs in a scale to measure body mass. Therefore, it is even possible to assess, in the same procedure and while the system is hermetically closed, 2 different portions of the same limb.

A new technique for limb volume measurement has been developed using laser scanning technology.16 It provides volume calculations for any and all segments of the scanned limb. This instrument calculates the volumes from the data generated by the readings of the 3-dimensional topography of the scanned surface, which creates a digital model of the scanned object. According to its developers, the comparison of laser scan volumes proved to be extremely accurate when measuring objects of known volume. In addition, comparisons of laser scan measurements with water displacement measurements of objects of unknown volume suggest superior reliability. They also reported that water displacement consistently underestimates volume at errors that increase as the object volume increases. However, the high cost and long evaluation times (5–10 minutes per limb, from stabilization to assessment)16 are factors that discourage the use of laser scanning on a routine basis.

Deltombe et al27 stated that the assessment of reliability should be the first priority because determination of validity is necessarily biased by confounding a supposed gold standard with the reality and in practice the overflowing volumetry does not give the exact volume of the limb. In strictly technical terms, the CVV is the first volumeter that uses a truly direct technique to measure human extremities by applying Pascal's principle to Archimedes' law. Optoelectronic perometry8,14,15 and laser scanning16 involve computer-based algorithm calculations for segment scanning, and circumferential measurements3–5,27,31,38,40,42 demand the use of mathematical formulas to convert 1-dimensional measures into volume units. Likewise, overflowing is not a direct measurement technique (because it requires moving the displaced contents into other containers or weighing the overflow contents to obtain equivalent mass units), nor a dynamic method (because the final volume does not match the time of immersion).34 In view of the foregoing, we agree with Deltombe et al27 that reliability assessment is of paramount importance, especially in this study, which is the first to report the findings of the CVV. However, we acknowledge the limitation of the measurements presented here because the limits of agreement of the CVV with the gold standard were not determined and the accuracy of the new method was not assessed by comparison of hand volumes against objects of known volume (“phantoms”). In addition, it is important that future research evaluating the new method take into account the assessment of swollen hands and the device's ability to detect changes in volume.

Two conclusions can be drawn based on this study. The intraobserver and interobserver rates for a new device for volume measurement of extremities (ie, CVV), built for the evaluation of the wrist-hand segment, showed no significant differences in reliability. Also, the analysis, alone or in combination, of time, temperature, sex, side of the wrist-hand segment, and interaction between sex and side of segment did not reveal any statistical differences between intraobserver or interobserver measurements.

## Footnotes

*   All authors provided concept/idea/research design. Dr de Carvalho provided writing, data collection and analysis, and study participants. Dr Miranda provided project management and fund procurement. Dr Perez and Dr Miranda provided facilities/equipment, institutional liaisons, and consultation (including review of manuscript before submission).

*   The authors thank Cristiano Campozana de Queiroz, engineer, and Wilson Adriani Filho, engineering technician, for their invaluable technical contributions to the construction of this device, and Dr Alexandre Galvão Patriota, statistician, for the analyses and mathematical models that he prepared for the assessment of the study results.

*   The Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) awarded for the performance of this study under grant number 05/55240–0 and Núcleo de Proteção Intelectual (NUPI), Paulista School of Medicine, Federal University of São Paulo (UNIFESP), funded the preparation of the patent request for the Communicating Vessels Volumeter, filed under number PI 0502899-0.

*   Received May 24, 2011.
*   Accepted May 29, 2012.