| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Research Reports |
R Taylor, MB, BS, DTMH, PhD, FRCP, FAFPHM, is Professor, School of Population Health, Faculty of Health Sciences, University of Queensland, Herston, Queensland, Australia, and Consultant Epidemiologist, New South Wales Breast Cancer Institute, University of Sydney, Westmead Hospital, Westmead, New South Wales, Australia
UW Jayasinghe, MSc, GradDipStat, GradDipApplStat, PhD, is Consultant Statistician, New South Wales Breast Cancer Institute, University of Sydney, Westmead Hospital, Westmead, New South Wales, Australia
L Koelmeyer, BApplSc(OT), is Occupational Therapist, Occupational Therapy Department, Westmead Hospital, Westmead, New South Wales, Australia
O Ung, MB, BS, FRACS, is Clinical Associate Professor and Clinical Services Director, New South Wales Breast Cancer Institute, University of Sydney, Westmead Hospital, Westmead New South Wales, Australia
J Boyages, MB, BS, PhD, FRACR, is Associate Professor and Executive Director, New South Wales Breast Cancer Institute, University of Sydney, Westmead Hospital, PO Box 143, Westmead, New South Wales, Australia 2145
(johnb{at}bci.org.au). Address all correspondence to Dr Boyages
Submitted January 3, 2005;
Accepted August 1, 2005
 | Abstract |
|---|
Subjects. Participants included subjects who had had breast cancer surgery, including axillary dissection—19 with and 22 without a diagnosis of arm lymphedema—and 25 control subjects.
Methods. Two raters measured each subject by using circumferential tape measurements at specified distances from the fingertips and in relation to anatomic landmarks and by using water displacement. Interrater reliability was calculated by analysis of variance and multilevel modeling. Volumes from circumferential measurements were compared with those from water displacement by use of means and correlation coefficients, respectively. The standard error of measurement, minimum detectable change (MDC), and limits of agreement (LOA) for volumes also were calculated.
Results. Arm volumes obtained with these methods had high reliability. Compared with volumes from water displacement, volumes from circumferential measurements had high validity, although these volumes were slightly larger. Expected differences between subjects with and without clinical lymphedema following breast cancer were found. The MDC of volumes or the error associated with a single measure for data based on anatomic landmarks was lower than that based on distance from fingertips. The mean LOA with water displacement were lower for data based on anatomic landmarks than for data based on distance from fingertips.
Discussion and Conclusion. Volumes calculated from anatomic landmarks are reliable, valid, and more accurate than those obtained from circumferential measurements based on distance from fingertips. [Taylor R, Jayasinghe UW, Koelmeyer L, et al. Reliability and validity of arm volume measurements for assessment of lymphedema. Phys Ther. 2006;86:205–214.]
Key Words: Arm circumference • Arm volume • Axillary surgery • Breast cancer • Lymphedema • Reliability • Validity • Volume measurement
| Introduction |
|---|
 TopAbstract Introduction MethodResultsDiscussion and ConclusionsReferences |
|---|
Arm lymphedema needs to be measured quantitatively to aid in the assessment of severity at the time of diagnosis and remeasured to assess response to treatments that may be administered. Furthermore, accurate measurement of arm volume is needed in observational studies of arm lymphedema as a complication of local treatment for breast cancer and in research trials of prevention or treatment. Evaluation of the effectiveness of treatments for lymphedema requires an accurate, easy-to-use method for the calculation of arm volume. Circumferential measurements usually are made (at different distances from the fingertips) and compared with those for the other arm as a sum or average or as a computed volume of an arm segment.
The most widely accepted measure of lymphedema is limb volume compared with that of the unaffected limb or compared with that of the same limb before the interventions or events that led to lymphedema. Volumes are most accurately measured by water displacement,4,6,7 although with limbs there are difficulties in defining and implementing the upper level for immersion, and water displacement is not convenient for routine clinical use. Megens et al4 and Sander et al7 found an intraclass correlation coefficient (ICC) of .99 for interrater reliability of water displacement volumes. However, many researchers choose not to use the water displacement method because it is time-consuming, is not portable, and can be nonhygienic.4 Considering the difficulties related to the water displacement method, searching for alternative methods of volume determination remains a worthwhile pursuit. Volumes can be calculated from multiple circumferential measurements of the limb on the basis of the assumption of truncated conical segments (frustum). Intervals for circumferential tape measurements can be based on the distance from the middle fingertip or on the distance from or between anatomic landmarks.
Sander et al7 compared volumetric measurements of an upper extremity obtained by water displacement with geometric measurements. They used geometric volume formulas for a cylinder, frustum (truncated cone), rectangular solid, and trapezoidal solid to compute volumes of the arm and hand at different measurement intervals and found that the frustum assumption produced the smallest standard error of measurement (SEM). Arm circumferences also can be measured with a Perometer,8,* which involves automated readings obtained at 0.4-cm intervals along the arm while the arm is positioned in a frame with a mobile source of infrared light; volumes are computed from the summation of elliptical segment volumes by use of specially designed software. Perometer measurements are reliable, convenient, and highly reproducible, with each measurement taking only a few seconds,9 but the equipment is costly and not used widely. Limb volume also can be estimated from the anterior and lateral silhouettes with a device that uses beams of infrared light; however, this method is expensive, and its accuracy is uncertain.10
The purpose of this study was to determine whether volumes calculated from anatomic landmarks are reliable, valid, and more accurate than those obtained from circumferential measurements based on distance from fingertips. The accuracy of volume measurements of the upper limb computed from circumferential measurements based on distance from fingertips and anatomic landmarks was compared with that of measurements obtained from volume displacement (criterion validity). The hypothesis was that volumes calculated from circumferences related to anatomic landmarks would be more accurate than volumes calculated from circumferences at fixed distances from fingertips. The reasoning was that the latter method often involves a segment across the elbow joint (which is not very conical in shape) and, because the length of the arm differs from woman to woman, fixed distances from fingertips would be in different positions relative to the anatomy in different women; this situation has implications for the validity of comparisons of grouped data. Subsidiary objectives were to assess the interrater reliability of measurements of volumes of the upper limb and to examine construct validity by comparing groups. Three groups of women were included in the study: control subjects, women with a clinical diagnosis of lymphedema in the ipsilateral arm after axillary surgery for breast cancer, and women without a clinical diagnosis of lymphedema in the ipsilateral arm after axillary surgery.
| Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussion and ConclusionsReferences |
|---|
Both arms of each subject were measured twice by 2 raters using 2 methods of circumferential measurement (at fixed distances from fingertips and in relation to anatomic landmarks) and water displacement. Five different raters took measurements during the study period. All raters were either occupational therapists or nurses who had previous experience with arm measurements in patients with lymphedema and who also received special training for the study, especially in measurement by water displacement, which is not routinely used at the Lymphedema Clinic.
Measurements
Arm measurements were obtained by multiple circumference measurements with a measuring tape (at various distances from fingertips and anatomic landmarks) and calculation of volumes and by water displacement. Each subject was measured once each by 2 raters. For the comparative volume estimations, the segment of the limb between the wrist and an upper boundary on the upper arm was used. This strategy was used because calculation of volumes of the hand by circumferential measurements is imprecise and a conical assumption is obviously not valid. An upper boundary was used because women are not able to submerge their entire limb for volume displacement measurements. The upper boundary selected was the level to which all women (in a pilot study) were able to submerge their arm for the water displacement measurements, and it was expressed as a relative value to take into account variations in limb lengths. The upper boundary was 65% of the distance from the elbow (olecranon) to the shoulder tip (acromion).
The volume for a segment between 2 adjoining measured segments was computed by assuming a truncated cone. Four truncated cones were summed to measure arm volumes based on both distance from the tip of the middle finger (wrist to 30 cm, 30–40 cm, 40–50 cm, and 50 cm to the 65% mark) and anatomic landmarks (wrist to mid forearm, mid forearm to elbow, elbow to mid upper arm, and mid upper arm to the 65% mark) (Figure 
). Circumferential tape measurements were taken at each of these points using a thin, flexible plastic tape.
|
 | (1) |
For the water displacement method, volume for the hand and volume for the total limb (to the upper boundary) were recorded separately. This goal was accomplished by first measuring the water displacement for the hand (to the wrist) and then measuring the total displacement for the limb to the designated upper boundary. A special device was constructed with an overflow spout; displaced water was caught and measured in a second container. The volume of the arm segment between the wrist and the upper boundary was obtained by subtracting the hand volume recorded from the total limb volume recorded. Laterality (right, left), dominance (handedness), and whether the arm was ipsilateral (affected) or contralateral (unaffected) to the breast cancer surgery site were noted.
Data Analysis
Interrater reliability
Interrater reliability was computed by use of the intrasubject correlation, which is the correlation between measurements obtained by 2 independent raters of the same subjects across a number of different subjects. The intrasubject correlation was derived from an analysis of variance (ANOVA)12–14 and multilevel modeling.15 When different sets of raters evaluated each subject, the intrasubject correlation coefficient (
) was derived from an ANOVA model across raters, as follows:
 | (2) |
In multilevel modeling, the intrasubject correlation () is defined as the correlation between 2 independent ratings of the same subject.13 The correlation between 2 such ratings is calculated as follows15:
 | (3) |
) and the average number of raters (N) in the Spearman-Brown formula, as follows13,16:
 | (4) |
Criterion validity
The criterion validity of circumferential measurements was determined by comparing volumes computed from circumferential measurements with a reference standard of volume measurements obtained by water displacement. Mean volumes were compared by use of paired sample t tests, and Pearson correlation coefficients with confidence intervals were computed for volumes obtained from circumferential measurements compared with those obtained from water displacement. Volumes obtained from the 2 methods for the same arm were compared; therefore, the characteristics of the arm (eg, affected, dominant) were not important in this analysis.
Limits of agreement (LOA)
By computing LOA, we could determine whether volumes calculated from the methods of circumferential measurements and water displacement were interchangeable.4,18,19 The LOA calculations showed how closely the 2 methods agreed with each other in numeric values.4 The LOA procedure involves computing the mean (d) and standard deviation (s) of differences in all pairs of measurements for 2 methods. If the differences are normally distributed, then 95% of differences will lie between d–2s and d+2s. This range is considered the 95% confidence interval of the LOA. The difference between the circumferential measurement and water displacement methods can be expected to vary between d–2s and d+2s.18 The question of interchangeability should be based on a clinical criterion and the application of the measurements.
SEM and minimum detectable change (MDC)
The SEM or the error associated with a single measure was computed with the following formula20,21:
 | (5) |
The MDC was defined as follows22:
 | (6) |
Construct validity
The construct validity of volume measurements of the upper limb as an indicator of lymphedema was assessed by comparing volumes of affected and unaffected arms for women with breast cancer, with and without diagnosed lymphedema, and also by comparing left and right arms and dominant and nondominant arms. Volumes of affected and unaffected arms for women with breast cancer, with and without diagnosed lymphedema, were compared before and after adjustment for covariates (right arm and dominant arm) by use of multilevel modeling with arm volumes as response variables. Multilevel models were fitted with MLwiN (version 1.1).17,
| Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussion and ConclusionsReferences |
|---|
|
.95 for subjects with lymphedema, .98 for subjects with breast cancer and without lymphedema, and .94 for control subjects (for dominant and nondominant arms assessed separately) (Tab. 2). There were no differences between mean volumes, as indicated by confidence intervals and coefficients of variation, for different measurement methods (Tab. 3).
|
|
SEM and MDC
The MDC of volumes or the error associated with a single measure for data obtained from anatomic landmarks was lower than that for data obtained from distance from fingertips (Tab. 3). The range of the SEM for the arm data was 64.5 to 81.7 mL (Tab. 3). The volumes calculated from anatomic landmarks had the smallest SEM. A difference of up to 150 mL (MDC) is considered to be measurement error.
Criterion Validity
Volumes calculated from circumferences were higher than those obtained from water displacement by up to 5% (P<.001). Differences from water displacement volumes were smaller for volumes computed from circumferences based on anatomic landmarks (up to 2%) (P<.001) than for those calculated from distance from fingertips (Tab. 3). Comparison between methods indicated that there was a high correlation (.98) between volumes obtained from water displacement and volumes calculated from circumferential measurements (Tab. 3).
Construct Validity
The volume of the affected arm for the lymphedema group was 11% higher than that of the unaffected arm (P<.001), but the difference was slight (0.2%) and not significant for the breast cancer group without diagnosed lymphedema (Tab. 4). The results were unchanged when adjusted for arm dominance and left or right arm by multilevel modeling. Before and after adjustment for dominant arm and right arm by multilevel modeling, the differences in volumes between the breast cancer groups with and without diagnosed lymphedema (interaction) remained statistically significant (Tab. 5). The volume of the right arm was higher (
2%) than that of the left arm, although the difference did not reach statistical significance (Tab. 3). For control subjects, the volume of the dominant arm was higher than that of the nondominant arm when determined by circumferential measurements or water displacement (1.5%), but this difference did not reach statistical significance (data not shown).
|
|
| Discussion and Conclusions |
|---|
|
TopAbstractIntroductionMethodResultsDiscussion and ConclusionsReferences |
|---|
Circumferential measurements obtained with a tape measure could be affected by the positions on the arm chosen by different observers to measure circumferences, although the observers followed the same guidelines for distance from fingertips or anatomic landmarks. Nevertheless, ICCs for interrater reliability of these circumferential measurements and volumes calculated from these measurements for the arm, excluding the hand, were high (mostly .98). The ICCs for reliability of volumes determined from water displacement also were high (mostly .98). These results are similar to those of Sander et al7 (ICC=.99) and Megens et al4 (ICC=.99) for volumes calculated from distance from fingertips and from water displacement for a comparable group (affected arms in subjects with lymphedema).
Validity assessed by Pearson coefficients of correlation between volumes determined from water displacement and volumes calculated from circumferential measurements was very high (.98) and also similar to that found by Sander et al7 for a comparable group (affected arms in subjects with lymphedema). Further study of criterion validity indicated that volumes computed from circumferential measurements were higher than volumes measured by water displacement, although the difference was not large (<5%). Water displacement measurements have higher validity than circumferential measurements because the latter assume that arm segments are truncated cones, but they are not. The arm in cross section is an ellipse more than a circle, and there are surface irregularities not captured by spaced measurements. Furthermore, the thickness of the tape measure, although it is tiny, makes a measurement fractionally higher than the real circumference.
The use of anatomic landmarks to determine sites for circumferential measurements produced volumes more accurate (up to 2% larger than water displacement volumes) than those obtained from measurements at distances from fingertips (up to 5% larger than water displacement volumes). Unlike the findings of our study, Sander et al7 found volumes calculated from circumferential measurements for the arm (minus fingers) to be smaller than those determined by water displacement, as did Karges et al24 and Pani et al.25 However, in studies by Stranden26 and Megens et al,4 the volumes determined from water displacement were smaller than the volumes calculated from surface measurements for edema in the leg and upper extremity, respectively. The techniques of water displacement differed somewhat between the studies. The greater accuracy of circumferential measurements based on anatomic landmarks than on fixed distances from fingertips was hypothesized prior to data collection, because measurements related to anatomic sites take into consideration relative lengths of the arm and produce segments that correspond more to truncated cones because segments do not span the elbow. Such measurements are more comparable in different women with different arm lengths. However, obtaining measurements in relation to anatomic landmarks requires more expertise and training in observers than does using distances from fingertips because knowledge of surface anatomy is needed and therefore may be more difficult to implement in clinical practice. Nevertheless, validity, assessed by correlation coefficients, was high for both methods of circumferential measurement, compared with water displacement.
For adequate construct validity, it would be expected that the difference in arm volumes would be greater in women with clinical lymphedema following breast cancer than in those without clinical lymphedema. Women with clinical lymphedema had significantly greater differences between affected and unaffected arms than women with breast cancer but without diagnosed lymphedema. The findings of larger volumes for right arms than for left arms and dominant arms larger than nondominant arms also are consistent with expectations and confirm construct validity.
The volumes determined from anatomic landmarks had the lowest SEM (range=64.5–65.4 mL), followed by those determined from distance from fingertips (range=66.6–71.0 mL) and water displacement (range=66.5–81.7 mL) (Tab. 3). Sander et al7 reported SEMs of 116 mL for volumes determined by the 9-cm frustum method and 117 mL for volumes determined by water displacement. Clinicians require that a decision can be made concerning how much change from a previous reading is clinically significant and may require a change in treatment. Minimum detectable change is the way of determining that the change is not attributable to chance variation or measurement error. The results of our study suggest that a change of less than 150 mL (MDC) should be treated as no change by clinicians21; a change of more than 150 mL is not likely to be attributable to chance variation or measurement error.
These results indicate high reliability of the measurements of arm circumferences and volumes and are similar to those of both Sander et al7 and Megens et al.4 We also noted that volumes calculated from circumferential measurements relative to anatomic landmarks were more accurate than and different from those calculated from segments defined from distances from fingertips. However, both circumferential methods produced high coefficients of correlation with water displacement. Volumes calculated from circumferential measurements based on anatomic landmarks had smaller mean LOA with water displacement and displayed smaller MDC and SEM than did those calculated from circumferential measurements based on distances from fingertips (Tab. 3). In addition, the difference in volumes between the right arm and the left arm determined by circumferences based on anatomic landmarks was closer to that determined by water displacement than to that determined by circumferences based on distances from fingertips (data not shown). Both circumferential methods systematically overestimated the volumes in relation to water displacement and should not be used interchangeably with water displacement. However, they may be used individually.
Accurate measurement of arm volumes is required in studies of women after various types of breast cancer treatment, including axillary dissection, and for assessing arm outcomes in clinical trials of different treatment procedures. These studies also permit the examination of predictors of lymphedema. Furthermore, accurate measurement of arm volumes is required for assessing methods for the treatment of lymphedema either in observational case series or in prospective clinical trials.
Valid and repeatable measures of arm volumes are required for both studies of causes and antecedents of arm lymphedema and trials of therapy for lymphedema. Water displacement is too cumbersome and messy to be used in routine clinical practice. This study has shown that arm circumferential measurements in relation to anatomic landmarks are reliable and valid measurements of arm volumes, can be used with confidence, and appear to be much more efficient clinically. Future research could address the validity of data obtained with newer methods for measuring arm volumes, such as perometry.
| Footnotes |
|---|
Centre for Multilevel Modeling, Institute of Education, 20 Nedford Way, London, United Kingdom WC1H 0AL.
All authors provided concept/idea/research design. Dr Taylor, Dr Jayasinghe, and Ms Koelmeyer provided writing. Ms Koelmeyer provided data collection, and Dr Taylor and Dr Jayasinghe provided data analysis. Ms Koelmeyer and Dr Boyages provided subjects. Dr Boyages provided fund procurement and clerical support. Dr Jayasinghe, Ms Koelmeyer, and Dr Boyages provided consultation (including review of manuscript before submission). The authors acknowledge Gregory Heard, PhD, Research Manager, New South Wales Breast Cancer Institute, University of Sydney, Westmead Hospital, PO Box 143, Westmead, New South Wales, Australia, for contributing to project management and consultation. They also acknowledge the occupational therapists from the Occupational Therapy Department, Westmead Hospital, Westmead, New South Wales, Australia, who participated in the measurement study on which this article is based.
This study was approved by the Western Sydney Area Health Service Human Research Ethics Committee.
The New South Wales Breast Cancer Institute receives funding from the New South Wales Department of Health. Jobst Beiersdorf Australia Limited funded the participation of a staff occupational therapist in the measurement study on which this article is based.
This research, in part, was presented at the Australian Lymphoedema Association Conference, March 26–28, 2004, Brisbane, Queensland, Australia, and at the Leura V International Breast Cancer Conference, November 10–14, 2004, Sydney, New South Wales, Australia.
| References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussion and ConclusionsReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
