British Journal of Radiology (2003) 76, 832-834
© 2003 British Institute of Radiology
doi: 10.1259/bjr/31249864
Clinical agreement between automated and calculated ultrasound measurements of bladder volume
N J Dudley, BSc, MSc, FIPEM
1
M Kirkland, RGN, RSCN
2
J Lovett, RSCN
2 and
A R Watson, FRCR
2
Departments of 1 Medical Physics and 2 Paediatric Nephrology, Nottingham City Hospital, Hucknall Road, Nottingham NG5 1PB, UK
 |
Abstract
|
|---|
Non-invasive urine volume measurement is an important tool in the management of dysfunctional and neuropathic bladders in children. Ultrasound imaging devices have been used for many years for this purpose. An automated scanner (Bladderscan) is now available and has been recommended by a number of authors, but there is conflicting evidence in the literature regarding the accuracy and appropriate clinical application of the device. We aimed to assess the level of clinical agreement between the two methods. 36 urine volume measurements were made on 11 children using both instruments. Although there was a good correlation between the methods (r=0.97), the clinical agreement was poor (limits of agreement ±77 ml). 13 voided volumes were directly measured and compared with the difference between pre- and post-void ultrasound measurements. The systematic errors were small but the mean absolute errors were 54 ml and 23 ml, respectively, for the automated and ultrasound imaging methods. If used correctly, ultrasound imaging provides more accurate results and can compete with the cost, convenience and ease of use of the automated method. Low cost, highly portable ultrasound imaging devices are now available and should be used in preference to the Bladderscan.
|
Introduction
|
|---|
Measurement of residual urine volumes in children with dysfunctional and neuropathic bladders is important in assessing the need for catheterization. Until recently, we have measured urine volumes using a conventional ultrasound scanner (ATL UM4, ATL, Bothell, WA). Following the recommendation of Padmore et al [1] of the use of a convenient and easy to use automatic bladder scanner we purchased a Bladderscan BVI 2500 (Diagnostic Ultrasound Corporation, Redmond, WA).
There is conflicting evidence in the literature regarding the accuracy of the Bladderscan instrument for determining residual volumes. It has been shown to be a useful tool in determining the need for post-operative catheterization [2] but its value in accurately determining residual volumes remains questionable. An early device (Bladderscan BVI 2000) was shown to be sufficiently accurate only to differentiate between bladder volumes as small or large [3]. Coombes and Millard [4] recommended the use of the Bladderscan BVI 2500+ as an alternative to catheterization for the determination of residual bladder volume on the basis of a mean error of 24 ml (95% confidence interval 16 ml to 34 ml). Using the same instrument, Alnaif and Drutz [5] found that residual urine volumes were not accurate, 39% of readings below 50 ml and 90% of readings above 150 ml were in error by more than 25%. Salinas Casado et al [6] concluded that conventional ultrasound imaging is a more reliable method than use of the Bladderscan.
Ultrasound imaging and orthogonal linear measurements is a widely accepted method for bladder volume measurement [7–10]. Since this was an accepted and trusted method in our centre prior to introduction of the Bladderscan, we decided to focus our attention on the level of clinical agreement between the two techniques before changing practice.
|
Subjects and methods
|
|---|
The study subjects were children referred to the paediatric urology nurses by three nephrologists and a urologist for bladder assessment and residual urine measurement. Although informed consent was obtained for the procedure, no Research Ethics Committee approval was required as the two techniques were used in parallel for a period of validation. The children were given drinks in order to fill their bladder. Measurements of bladder volume were then made using both the Bladderscan and conventional ultrasound imaging with the patient supine and not changing position between measurements. The measurements were repeated post-voiding and in some cases, where incomplete filling or voiding of the bladder was suspected, further measurements were made after repeated voiding attempts. Voided urine volumes were directly measured and compared with the difference between the pre- and post-void measurements of both instruments.
Bladderscan measurements were made according to the manufacturer's instructions. With the probe held just above the symphysis pubis, the instrument automatically steers the mechanical transducer to collect data in 12 planes, providing a coronal bladder outline ("Aiming Icon") to aid positioning, then uses the images to automatically calculate the estimated bladder volume. Scans were performed until the on-screen cross hairs were centred in the bladder outline. The final value was recorded. The Bladderscan was a new instrument with a certificate of calibration from the supplier.
Ultrasound imaging was used to obtain sagittal and transverse images of the largest cross-sections of bladder visualized. The images were measured in three orthogonal directions: from the top to bottom of the bladder (y) and at 90° to this (z) in the sagittal plane and from left to right in the transverse plane (x). The volume was calculated using the formula Volume=x x y x z x k where k is 0.72 [11]. The ATL UM4 callipers were accurate to within 1 mm for linear measurements as confirmed by local quality control measurements.
Results from the two methods were compared by calculating the correlation coefficient and by using a Bland and Altman plot [12]. The latter method was used to demonstrate the difference within each pair of results, since a good correlation may sometimes be obtained despite significant differences between measurements. Mean and mean absolute differences between the methods were also calculated. The mean represents the systematic difference between the methods; since this can be small despite significant differences between measurements, the mean absolute difference was used as a measure of the magnitude of these differences.
Bih et al [11], in addition to providing a k-value for the above equation of 0.72, suggested that corrections for shape may improve the performance of the imaging method. They suggested correction coefficients of 0.89, 0.81 and 0.66 for cuboidal, ellipsoidal and triangular prism shaped bladders, respectively. Our pre- and post-void volume estimates, were recalculated using these coefficients to determine whether any improvement was possible.
|
Results
|
|---|
A total of 36 bladder volume measurements were made on 11 children with an age range of 5 years to 11 years. Pre-voiding measurements ranged from 36 ml to 640 ml (Bladderscan) and 22 ml to 477 ml (imaging). Figure 1
shows Bladderscan results versus ultrasound image measurements. There is a good correlation between the data sets (r=0.97). Figure 2, however, shows the poor clinical agreement between the two methods [12]; although the mean difference was 39 ml, the limits of agreement are ±77 ml, which is unacceptable when using a diagnostic threshold of less than 100 ml.
View larger version (14K):
[in this window]
[in a new window]
|
Figure 2. Bland-Altman plot showing the difference between measurements (Bladderscan minus imaging) against the mean of the two measurements.
|
|
13 voided volumes were measured, ranging from 20 ml to 425 ml. Figure 3 shows the difference between pre- and post-void measurements for the two instruments against true voided volume. The errors are shown in Table 1. The absolute error for the Bladderscan is more than twice the imaging error. The variation in errors (95% confidence interval) is also larger for the Bladderscan. The difference in errors is also clear from Figure 3.
View this table:
[in this window]
[in a new window]
|
Table 1. Errors (and 95% confidence intervals, CI) in the estimation of voided volume by the Bladderscan and by ultrasound imaging
|
|
In this study all bladders were considered to be cuboidal or ellipsoidal. Applying the appropriate correction coefficients (0.89 and 0.81, respectively [11]), rather than the value of 0.72, the systematic and absolute errors in voided volume from ultrasound imaging measurements are shown in Table 2. The systematic error is of opposite sign but similar magnitude, and the mean absolute error is increased, when using the shape dependent coefficients.
View this table:
[in this window]
[in a new window]
|
Table 2. Errors (and 95% confidence intervals, CI) in the estimation of voided volume by ultrasound imaging, using the single coefficient of 0.72 and the shape dependent coefficients suggested by Bih et al [11]
|
|
|
Discussion and conclusions
|
|---|
The results show that approximately half of the pre-voiding bladder volumes were 200 ml or more. This suggests the possibility of incomplete filling in some cases, which may result in residual urine following micturition. Although this is clinically important, it does not detract from the purpose of the study to assess the difference between the two measurement methods.
Other authors have recommended the Bladderscan for residual volume measurement based on the good correlation between measurements. A good correlation, however, does not necessarily indicate good clinical agreement [12]. The use of correlation coefficients and mean errors to justify a measurement method can be misleading; a high correlation coefficient does not necessarily indicate good agreement within individual pairs of measurements and many errors will be larger than the mean. The limits of agreement is a more useful measure for the assessment of the clinical significance of differences.
The Bladderscan relies heavily on its algorithms fitting the shape of the bladder. If the shape is very different from that expected, e.g. flat, the results could be very inaccurate. As the bladder shape changes between pre and post voiding measurements, the algorithms of either method may prove unreliable. Following the suggestion of Bih et al [11] to use different correction coefficients for cuboidal, ellipsoidal and triangular prism shaped bladders, respectively, did not reduce the errors in our volume estimates; in fact the addition of this subjective shape assumption increases the variation in the errors.
There is a major disadvantage to a technique that does not allow direct visualization of the measured structure. Patients are encouraged to drink during preparation for bladder voiding studies and this can produce fluid filled loops of bowel. We have seen these loops, confirmed by peristaltic motion, detected as residual bladder volumes by the Bladderscan when ultrasound imaging showed the bladder to be empty.
If used correctly, ultrasound imaging provides more accurate results and can compete with the cost, convenience and ease of use of the automated method. Low cost, highly portable ultrasound imaging devices are now available and should be used in preference to the Bladderscan.
Received for publication January 29, 2003.
Revision received June 25, 2003.
Accepted for publication July 24, 2003.
|
References
|
|---|
- Padmore DE, Anderson PAM, Tooth DS, Waye MA. Evaluation of a non-invasive method to determine bladder volume in children. Can J Urology 1997;4:305–8.
- Slappendel R, Weber EW. Non-invasive measurement of bladdervolume as an indication for bladder catheterisation after orthopaedic surgery and its effect on urinary tract infections. Eur J Anaesthesiol 1999;16:503–6.[CrossRef][Medline]
- Rohde T, Jensen KM, Colstrup H. Testing an ultrasonic scanner for determination of urinary bladder volume. Ugeskr Laeger 1992;154:3499–501.[Medline]
- Coombes GM, Millard RJ. The accuracy of portable ultrasound scanning in the measurement of residual urine volume. J Urol 1994;152:2083–5.[Medline]
- Alnaif B, Drutz HP. The accuracy of portable abdominal ultrasound equipment in measuring postvoid residual volume. Int Urogynecol J Pelvic Floor Dysfunct 1999;10:215–8.[Medline]
- Salinas Casado M, Sanchez Blasco E, Virseda Chamorro M, et al. Reliability of the estimate of post-voiding bladder volume with ultrasound. Comparison of 2 ultrasonography methods. Arch Esp Urol 1996;49:35–40.[Medline]
- Ravichandran G, Fellows GJ. The accuracy of a hand-held real time ultrasound scanner for estimating bladder volume. Br J Urol 1983;55:25–7.[Medline]
- Kjeldsen-Kragh J. Measurement of residual urine volume by means of ultrasonic scanning: a comparative study. Paraplegia 1988;26:192–9.[Medline]
- Hendrikx AJ, Doesburg WH, van den Stappen W, Hofmans PA, Debruyne FM. Ultrasonic determination of the residual bladder volume. Urol Int 1989;44:96–102.[Medline]
- Williot P, McLorie GA, Gilmour RF, Churchill BM. Accuracy of bladder volume determinations in children using a suprapubic ultrasonic bi-planar technique. J Urol 1989;141:900–2.[Medline]
- Bih LI, Ho CC, Tsai SJ, Lai YC, Chow W. Bladder shape impact on the accuracy of ultrasonic estimation of bladder volume. Arch Phys Med Rehabil 1998;79:1553–6.[Medline]
- Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;i:307–10.
Top
Abstract
Introduction
