- Email: [email protected]
Portable Ultrasonography and Bladder Volume Accuracy—A Comparative Study Using Three-Dimensional Ultrasonography
Ambulatory & Office Urology Portable Ultrasonography and Bladder Volume Accuracy—A Comparative Study Using Three-Dimensional Ultrasonography Khurshid R. Ghani, James Pilcher, David Rowland, Uday Patel, Daruish Nassiri, and Ken Anson OBJECTIVES
METHODS
RESULTS
CONCLUSIONS
To compare the ultrasound bladder volume accuracy and level of agreement between two portable bladder scanners (Bladderscan and Bardscan) and a three-dimensional ultrasound (3D-US) system. A total of 50 healthy volunteers were scanned using the Bladderscan BVI 3000, Bardscan, and 3D-US system (HDI 4000), in random sequence. The BVI3000 is a dedicated bladder volume calculator, and the Bardscan combines real-time ultrasonography with bladder volume calculation. The ultrasound bladder volumes were compared with the voided volume measurements. The volunteers underwent repeat scanning after voiding, and those with a measurable residual volume were excluded from the final analysis. A residual volume was detected in 16 subjects (32%). In the remaining 34 subjects, the mean voided volume ⫾ standard deviation was 252.9 ⫾ 167.4 mL (range 33 to 709). A significant correlation (P ⬍0.001) was found between the voided and ultrasound volumes with all three methods (Bardscan, r ⫽ 0.97; Bladderscan, r ⫽ 0.98; and 3D-US system, r ⫽ 0.99). No significant differences were found between the voided volumes and the Bladderscan or 3D-US volumes; however, the Bardscan significantly underestimated the voided volume by a mean of 21.4 mL (t ⫽ 2.84, P ⫽ 0.0076). The Bland-Altman 95% limit of agreement between the voided and calculated volumes was ⫺64.5 to 107.2 mL, ⫺73.7 to 88.4 mL, and ⫺28.9 to 40.0 mL for the Bardscan, Bladderscan, and 3D-US systems, respectively. The results of our study have shown that although the Bardscan has the advantages of real-time scanning with portability and instantaneous volume calculation, it is not as accurate as the Bladderscan. The accuracy and level of clinical agreement was greatest when using the 3D-US system to calculate the bladder volume. UROLOGY 72: 24 –28, 2008. © 2008 Elsevier Inc.
ladder and residual volume (RV) measurement remains an integral part of the evaluation of voiding dysfunction. Portable bladder scanners designed to address the drawbacks associated with the larger ultrasound machines are widely used in urologic practice. The Bladderscan, a one-purpose automatic device for measuring the bladder volume, is the most studied unit.1–12 Although the volumes are calculated using a three-dimensional (3D) volumetric probe, its performance compared with a fully configured 3D ultrasound (3D-US) system has not been studied. Despite its wide use, the Bladderscan has limitations. It is unreliable in pregnant patients,1 patients with cystic pelvic pathologic findings,2 those requiring peritoneal
B
dialysis,3 and after bladder augmentation.4 The measurements can be affected by hysterectomy, uterine prolapse,5 sex,6 and obesity.7 It is less reliable in young children8 and in neonates.9 Fluid-filled loops of bowel can also be mistaken for bladder volume.10 The Bardscan is a bladder scanner that combines conventional ultrasonography (CUS) with portability and ease of use. Unlike the Bladderscan, it provides a real-time image of the bladder, which provides advantages in assessing diverticula, calculi, tumor, and the position of catheters. Independent assessment of the Bardscan has been limited to only one study, with no comparison with other instruments.13 In the present study, we investigated the accuracy and level of agreement of the Bardscan, Bladderscan, and a 3D-US system in calculating the bladder volume.
From the Departments of Urology, Radiology, and Clinical Physics and Ultrasound, St. George’s Hospital Medical School, London, United Kingdom Reprint requests: Khurshid Ghani, B.Sc.(Hons.), M.B.Ch.B., M.R.C.S.(Ed.), Department of Urology, St. George’s Hospital, Blackshaw Road, London SW17 0JT United Kingdom. E-mail: [email protected] Submitted: November 25, 2007, accepted (with revisions): February 16, 2008
MATERIAL AND METHODS
24
© 2008 Elsevier Inc. All Rights Reserved
Subjects After institutional ethics review board approval, 50 healthy volunteers were recruited during a 2-week period and informed 0090-4295/08/$34.00 doi:10.1016/j.urology.2008.02.033
consent was obtained. The exclusion criteria included a history of bladder surgery or pathologic features, age younger than 18 years, weight greater than 420 lb, or pregnancy. The subjects underwent scanning with no instructions regarding hydration or voiding.
Equipment The Bladderscan BVI-3000 (Verathon, Bothell, Wash) uses a 2-MHz volumetric probe with a transducer that is steered to collect 12 cross-sectional images (15° apart) in one step and provide a coronal bladder outline on a monitor in the shape of a target with cross-hairs (BVI 3000 size 23 ⫻ 32 ⫻ 7 cm, weight 2.25 kg, cost $17,423). The Bardscan (Bard, Crawley, UK) uses a 3.5/5-MHz mechanical sector probe attached to an 8-in. color screen (640 ⫻ 480 pixels; Bardscan size 25 ⫻ 36 ⫻ 9 cm, weight 3 kg, cost $18,870). The default setting of 3.5 MHz was used in this study. The HDI 4000 (Phillips Medical Systems, Best, The Netherlands) is a 3D-US system incorporating a curved array, 3 to 5-MHz volumetric probe. With the probe held fixed, the array is automatically steered to acquire real-time 3D images on a color screen (1024 ⫻ 768 pixels).
Study Design The measurements were obtained according to the manufacturers’ instructions. For the Bladderscan, the probe was fixed superior to the symphysis pubis and directed toward the bladder. The scans were performed until the cross-hairs were iso-centered on the bladder outline on the monitor, at which point, three repeated measurements were recorded, and the greatest was used as the volume. Compared with the average or single reading, the greatest reading had the strongest correlation11 and least bias.6 With the Bardscan, the probe was adjusted to view the bladder in its largest cross-section. On pressing the probe, the picture freezes and an automatic volume tool outlines the bladder. If this does not correspond with the actual bladder outline, the measurement can be repeated. The manufacturer recommends scans in two planes (transverse and sagittal) for increased accuracy. Using this mode, the greatest of three repeated measurements was used as the volume. For the 3D-US, the probe was centered on the bladder to acquire the whole bladder within a single acquisition. The bladder volume was calculated at a later date by a single operator (K.R.G.), who was unaware of the results using the in-built virtual organ computer-aided analysis software. This automatically extracts contours to produce a 3D model and volume. A line appears along the bladder outline displayed in a multiplanar view. The line can be adjusted if the fit is suboptimal. The subjects were scanned in the supine position, without changing their position and with no delay between readings, by 1 of 2 physicians. The time taken to scan a subject with all three instruments was recorded. Both operators had previous experience with the Bladderscan and had undergone training for the Bardscan and 3D-US with the company representative. The instruments were used in a random sequence determined from random number tables. On completion of scanning, the subjects immediately voided and underwent repeat scanning with the 3D-US to determine the RV. The voided urine was measured using graduated measuring devices and compared with the ultrasound measurements. UROLOGY 72 (1), 2008
Figure 1. Scanned volume versus voided volume using all three ultrasound systems (line of equality shown).
Statistical Analysis Patients with a RV were excluded from the final analysis. The strength of the relationship between the voided and calculated volumes was assessed using Pearson’s correlation coefficient. The differences against the voided volume were assessed with the paired t test. Because the average of the readings between methods acts as the best estimate of the unknown true value, Bland-Altman plots and limits of agreement were calculated, including bias (mean difference), precision (1.96 ⫻ standard deviation of differences), and standard error.14 Bias and precision help to calculate the 95% range of readings for a measured value. For example, when a device with a bias of 20 mL measures the volume at 400 mL, the mean true urinary volume will be 420 mL, ranging from (420 mL ⫺ precision) to (420 mL ⫹ precision). All calculations were performed using Prism for Macintosh, version 4.0 (GraphPad Software, San Diego, Calif). P ⬍0.05 was considered statistically significant.
RESULTS In total, 27 male and 23 female subjects were scanned. The mean ⫾ SD age of subjects was 35.7 ⫾ 10.9 years (range 19 to 60). One operator (K.R.G.) scanned 35 subjects and the remainder were scanned by the second (J.P.). The mean ⫾ SD time taken to scan the subjects using all three instruments was 6.2 ⫾ 2.1 minutes (range 2.9 to 13.1). Sixteen patients (32%) had a RV (range 4 to 302 mL) and were excluded. In 1 subject, 3D-US could not calculate the volume because the bladder was too large for complete acquisition. This subject was excluded from the 3D-US group. In the remaining 34 patients, the mean ⫾ SD voided volume was 252.9 ⫾ 167.4 mL (range 33 to 709). Figure 1 shows the volumes obtained using all three instruments plotted against the voided volumes. An excellent correlation (P ⬍0.001) was found between the voided and calculated volume using the Bladderscan (r ⫽ 0.98), Bardscan (r ⫽ 0.97), or the 3D-US system (r ⫽ 0.99). The mean calculated volumes, bias, precision, and standard error for all three instruments are listed in Table 1. The volumes calculated using the Bardscan were significantly different from the voided volumes (t ⫽ 2.84, P ⫽ 0.0076). No significant differences were found between 25
Table 1. Mean difference and 95% confidence intervals and standard error in calculation of voided volume using Bladderscan, Bardscan, and 3D-US system Instrument Bladderscan Bardscan 3D-US
Mean ⫾ SD Volume (mL)
Bias (mL)
Precision (mL)
SEM Difference (mL)
245.5 ⫾ 189.7 (17–856) 231.5 ⫾ 152.8 (0–595) 233.5 ⫾ 149.7 (13.1–671.6)
7.4 21.4* 5.6
81.0 85.9 34.4
7.1 7.5 3.1
SD ⫽ standard deviation; SEM ⫽ standard error of mean; 3D-US ⫽ three-dimensional ultrasonography. * Compared with mean voided volume (P ⫽ 0.0076).
Figure 2. Bland-Altman analyses of voided and scanned volume (n ⫽ 34). Average of voided volume plus scanned volume (x axis) plotted against difference of voided minus scanned volume (y axis). Straight horizontal line indicates mean difference (bias) and dotted horizontal lines indicate 95% limits of agreement (LA). (A) Bardscan bias, 21.4 mL and 95% LA, ⫺64.5 to 107.2 mL; (B) Bladderscan bias, 7.4 mL and 95% LA ⫺73.7 to 88.4 mL; (C) 3D-US bias, 5.6 mL and 95% LA ⫺28.9 to 40.0 mL.
COMMENT
Figure 3. Difference between voided volume and scanned volume for entire range of voided volumes using all three ultrasound systems.
the voided and calculated volumes using the Bladderscan (t ⫽ 1.82, P ⫽ 0.08) or 3D-US system (t ⫽ 1.04, P ⫽ 0.31). The 3D-US system had the smallest mean difference (bias), precision, and standard error and was therefore the most accurate of all three instruments (Table 1). BlandAltman plots demonstrating the limits of agreement between the voided and scanned volumes are presented in Figure 2. The 3D-US system displayed the best level of clinical agreement. The superiority of the 3D-US system in calculating the bladder volumes across the entire range of voided volumes is demonstrated in Figure 3. The Bladderscan and Bardscan were inaccurate when faced with the large bladder volumes, with the Bardscan underestimating and the Bladderscan overestimating the measured volumes. 26
Earlier Bladderscan models (BVI 2000) used a two-plane technique with an algorithm based on the bladder as an ellipsoid.15,16 Later models (BVI 2500) using one-step volumetric scanning performed better.12 Subsequent studies found that the Bladderscan overestimated5 or underestimated the volumes compared with the catheterized volumes.7,17,18 We used the BVI 3000, which has a faster scan time and larger scan plane and has shown good correlation with catheterized volumes.3,19 –21 However, previous studies have incorrectly inferred the accuracy using correlation coefficients rather than the more robust bias, precision, and standard error we used. Very few studies have performed Bland-Altman analyses on measurements.6,10,22 Also, most studies have involved patients with abnormalities in bladder anatomy, which might have influenced the results. Our results have shown excellent correlation between the ultrasound-determined and voided volumes using all three instruments. Neither the Bladderscan nor the 3D-US system provided bladder volumes significantly different from the voided volume. In contrast, Bardscan volumes were significantly different from the voided volumes. The 3D-US system demonstrated the strongest agreement and was the most accurate of all three instruments. This should not be surprising, because we used a high performance 3D-US system. In contrast, the performance of the smaller, inexpensive, and easier to use Bladderscan was encouraging. The number of studies comparing the Bladderscan and CUS is limited.10,23–25 Bano et al.23 compared the BVI UROLOGY 72 (1), 2008
3000 bladder volumes with the suprapubic catheter and CUS-determined volumes in 26 postoperative women. Because of the strong positive linear association between all three methods (r ⫽ 0.97), the investigators thought the Bladderscan should be preferred over CUS despite no assessment of the bias and whether the volumes differed significantly. Repeated measurements were not taken, and the effect of the suprapubic catheter on the Bladderscan measurements was not considered. Dudley et al.10 compared the BVI 2500⫹ with CUS in 11 pediatric patients. They found good correlation (r ⫽ 0.97) between instruments. Compared with the voided volumes, the mean absolute error using the Bladderscan was more than twice that using CUS. In a larger study, Huang et al.24 also found that stationary CUS was more accurate than the Bladderscan BVI 3000. In contrast, Byun et al.25 using true bladder volumes determined by fluoroscopy and catheterization as the reference standard, found CUS significantly underestimated the bladder volume by 21.8% compared with 3.3% using the Bladderscan BVI 3000. In the only previous study of the Bardscan, the ultrasound bladder volumes were compared with the catheterization volumes in 54 female patients.13 Only single readings were taken, and the recommended two-plane method was not routinely used. The Bardscan had a correlation of r ⫽ 0.98 (P ⫽ 0.16), a mean difference of 17.8 mL, and a precision of 114.3 mL. Although the values for precision between the Bardscan and Bladderscan in our study were quite similar, this was offset by the significantly greater bias in the Bardscan, which made it less accurate in comparison. Despite the lower accuracy, clinically, the Bardscan remains a useful device because it can assess the bladder anatomy and rule out large tumors or stones during volume calculation. The CUS volume calculations are determined using measurements in two planes, with geometric assumptions that are less robust compared with those used with 3D-US methods.26 Also, a wide discrepancy exists in the formulas and correction factors used for volume calculation with CUS.27 A previous study of bladder volume calculation using a freehand 3D-US system incorporating an electromagnetic position sensor found that 3D volume measurements had a mean absolute error of 4.9% compared with 27.5% for CUS.28 We have shown that the volumetric 3D method should be considered the reference standard for ultrasound bladder volume calculation. However, it is more expensive, and the processing time is more demanding. Ours was a small study, and more participants would have made the data more statistically significant. The instruments in this study were tested in young healthy individuals, not older patients with prostate or bladder disorders. In this study of volunteers, we were unable to obtain ethical approval for invasive catheterization. By excluding participants with a RV, we tried to limit the error inherent in the voided measurements. Even studies UROLOGY 72 (1), 2008
using catheterized volumes do not obtain true bladder volumes, unless emptying is confirmed on fluoroscopy, because up to 30% of patients can have a RV after catheterization.20 Also, the use of three instruments in this study lengthened the scanning time, which could have affected the bladder volume because of ongoing diuresis. Furthermore, only one model of each scanner was tested. Thus, any variation between scanners of the same model was not assessed. Variations of this nature could account for some of the conflicting results in the published data, and this matter constitutes the basis for additional research in our department.
CONCLUSIONS The Bardscan has the advantages of real-time bladder scanning and instantaneous volume calculation incorporated into a portable system, but it is not as accurate as the Bladderscan. The Bladderscan BVI3000, although it has its limitations, uses a method that comes close to the reference standard 3D volume assessment using ultrasonography. Acknowledgment. To Bard UK Ltd., Verathon Medical UK Ltd., and Philips Medical Systems UK for supplying the equipment used in this study. References 1. Barrington JW, Arunkalaivanan AS, and Abdel Fattah M: The accuracy of Bladderscan in intrapartum care. Int Urogynecol J Pelvic Floor Dysfunct 14: 214-215, 2003. 2. Cooperberg MR, Chambers SK, Rutherford TJ, et al: Cystic pelvic pathology presenting as falsely elevated post-void residual urine measured by portable ultrasound bladder scanning: report of 3 cases and review of the literature. Urology 55: 590, 2000. 3. Yucel S, Kocak H, Sanli A, et al: How accurate is measuring postvoid residual volume by portable abdominal ultrasound equipment in peritoneal dialysis patient? Neurourol Urodyn 24: 358361, 2005. 4. Barrington JW, Jones A, Robinson J, et al: Estimation of bladder volume using portable ultrasound in clam enterocystoplasty patients. J Urol 155: 82-83, 1996. 5. Goode PS, Locher JL, Bryant RL, et al: Measurement of postvoid residual urine with portable transabdominal bladder ultrasound scanner and urethral catheterization. Int Urogynecol J Pelvic Floor Dysfunct 11: 296-300, 2000. 6. Brouwer TA, Eindhoven BG, Epema AH, et al: Validation of an ultrasound scanner for determining urinary volumes in surgical patients and volunteers. J Clin Monit Comput 15: 379-385, 1999. 7. Alnaif B, and Drutz HP: The accuracy of portable abdominal ultrasound equipment in measuring postvoid residual volume. Int Urogynecol J Pelvic Floor Dysfunct 10: 215-218, 1999. 8. Rosseland LA, Bentsen G, Hopp E, et al: Monitoring urinary bladder volume and detecting post-operative urinary retention in children with an ultrasound scanner. Acta Anaesthesiol Scand 49: 1456-1459, 2005. 9. Wyneski HK, McMahon DR, Androulakakis V, et al: Automated bladder scan urine volumes are not reliable in complex neonatal cases. J Urol 174: 1661-1662, 2005. 10. Dudley NJ, Kirkland M, Lovett J, et al: Clinical agreement between automated and calculated ultrasound measurements of bladder volume. Br J Radiol 76: 832-834, 2003. 11. Marks LS, Dorey FJ, Macairan ML, et al: Three-dimensional ultrasound device for rapid determination of bladder volume. Urology 50: 341-348, 1997.
27
12. Coombes GM, and Millard RJ: The accuracy of portable ultrasound scanning in the measurement of residual urine volume. J Urol 152: 2083-2085, 1994. 13. Abdel-Fattah M, and Barrington JW: The accuracy of Bardscan: a new tool for the measurement of the bladder volume. J Obstet Gynaecol 25: 186-188, 2005. 14. Bland JM, and Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307-310, 1986. 15. Massagli TL, Cardenas DD, and Kelly EW: Experience with portable ultrasound equipment and measurement of urine volumes: inter-user reliability and factors of patient position. J Urol 142: 969-971, 1989. 16. Fuse H, Yokoyama T, Muraishi Y, et al: Measurement of residual urine volume using a portable ultrasound instrument. Int Urol Nephrol 28: 633-637, 1996. 17. Borrie MJ, Campbell K, Arcese ZA, et al: Urinary retention in patients in a geriatric rehabilitation unit: prevalence, risk factors, and validity of bladder scan evaluation. Rehabil Nurs 26: 187-191, 2001. 18. Schott-Baer FD, and Reaume L: Accuracy of ultrasound estimates of urine volume. Urol Nurs 21: 193-195, 2001. 19. Demaria F, Amar N, Biau D, et al: Prospective 3D ultrasonographic evaluation of immediate postpartum urine retention volume in 100 women who delivered vaginally. Int Urogynecol J Pelvic Floor Dysfunct 15: 281-285, 2004. 20. Barrington JW, Edwards G, Ashcroft M, et al: Measurement of bladder volume following cesarean section using Bladderscan. Int Urogynecol J Pelvic Floor Dysfunct 12: 373-374, 2001.
28
21. Araki Y, Ishibashi N, Sasatomi T, et al: Effectiveness of the portable ultrasound bladder scanner in the measurement of residual urine volume after total mesorectal extirpation. Minim Invasive Ther Allied Technol 12: 245-248, 2003. 22. De Gennaro M, Capitanucci ML, Di Ciommo V, et al: Reliability of bladder volume measurement with BladderScan in paediatric patients. Scand J Urol Nephrol 40: 370-375, 2006. 23. Bano F, Arunkalaivanan AS, and Barrington JW: Comparison between Bladderscan, real-time ultrasound and suprapubic catheterisation in the measurement of female residual bladder volume. J Obstet Gynaecol 24: 694-695, 2004. 24. Huang YH, Bih LI, Chen SL, et al: The accuracy of ultrasonic estimation of bladder volume: a comparison of portable and stationary equipment. Arch Phys Med Rehabil 85: 138-141, 2004. 25. Byun SS, Kim HH, Lee E, et al: Accuracy of bladder volume determinations by ultrasonography: are they accurate over entire bladder volume range? Urology 62: 656-660, 2003. 26. Farrell T, Leslie JR, Chien PF, et al: The reliability and validity of three dimensional ultrasound volumetric measurements using an in vitro balloon and in vivo uterine model. Br J Obstet Gynecol 108: 573-582, 2001. 27. Hvarness H, Skjoldbye B, and Jakobsen H: Urinary bladder volume measurements: comparison of three ultrasound calculation methods. Scand J Urol Nephrol 36: 177-181, 2002. 28. Riccabona M, Nelson TR, Pretorius DH, et al: In vivo threedimensional sonographic measurement of organ volume: validation in the urinary bladder. J Ultrasound Med 15: 627-632, 1996.
UROLOGY 72 (1), 2008
METHODS
RESULTS
CONCLUSIONS
To compare the ultrasound bladder volume accuracy and level of agreement between two portable bladder scanners (Bladderscan and Bardscan) and a three-dimensional ultrasound (3D-US) system. A total of 50 healthy volunteers were scanned using the Bladderscan BVI 3000, Bardscan, and 3D-US system (HDI 4000), in random sequence. The BVI3000 is a dedicated bladder volume calculator, and the Bardscan combines real-time ultrasonography with bladder volume calculation. The ultrasound bladder volumes were compared with the voided volume measurements. The volunteers underwent repeat scanning after voiding, and those with a measurable residual volume were excluded from the final analysis. A residual volume was detected in 16 subjects (32%). In the remaining 34 subjects, the mean voided volume ⫾ standard deviation was 252.9 ⫾ 167.4 mL (range 33 to 709). A significant correlation (P ⬍0.001) was found between the voided and ultrasound volumes with all three methods (Bardscan, r ⫽ 0.97; Bladderscan, r ⫽ 0.98; and 3D-US system, r ⫽ 0.99). No significant differences were found between the voided volumes and the Bladderscan or 3D-US volumes; however, the Bardscan significantly underestimated the voided volume by a mean of 21.4 mL (t ⫽ 2.84, P ⫽ 0.0076). The Bland-Altman 95% limit of agreement between the voided and calculated volumes was ⫺64.5 to 107.2 mL, ⫺73.7 to 88.4 mL, and ⫺28.9 to 40.0 mL for the Bardscan, Bladderscan, and 3D-US systems, respectively. The results of our study have shown that although the Bardscan has the advantages of real-time scanning with portability and instantaneous volume calculation, it is not as accurate as the Bladderscan. The accuracy and level of clinical agreement was greatest when using the 3D-US system to calculate the bladder volume. UROLOGY 72: 24 –28, 2008. © 2008 Elsevier Inc.
ladder and residual volume (RV) measurement remains an integral part of the evaluation of voiding dysfunction. Portable bladder scanners designed to address the drawbacks associated with the larger ultrasound machines are widely used in urologic practice. The Bladderscan, a one-purpose automatic device for measuring the bladder volume, is the most studied unit.1–12 Although the volumes are calculated using a three-dimensional (3D) volumetric probe, its performance compared with a fully configured 3D ultrasound (3D-US) system has not been studied. Despite its wide use, the Bladderscan has limitations. It is unreliable in pregnant patients,1 patients with cystic pelvic pathologic findings,2 those requiring peritoneal
B
dialysis,3 and after bladder augmentation.4 The measurements can be affected by hysterectomy, uterine prolapse,5 sex,6 and obesity.7 It is less reliable in young children8 and in neonates.9 Fluid-filled loops of bowel can also be mistaken for bladder volume.10 The Bardscan is a bladder scanner that combines conventional ultrasonography (CUS) with portability and ease of use. Unlike the Bladderscan, it provides a real-time image of the bladder, which provides advantages in assessing diverticula, calculi, tumor, and the position of catheters. Independent assessment of the Bardscan has been limited to only one study, with no comparison with other instruments.13 In the present study, we investigated the accuracy and level of agreement of the Bardscan, Bladderscan, and a 3D-US system in calculating the bladder volume.
From the Departments of Urology, Radiology, and Clinical Physics and Ultrasound, St. George’s Hospital Medical School, London, United Kingdom Reprint requests: Khurshid Ghani, B.Sc.(Hons.), M.B.Ch.B., M.R.C.S.(Ed.), Department of Urology, St. George’s Hospital, Blackshaw Road, London SW17 0JT United Kingdom. E-mail: [email protected] Submitted: November 25, 2007, accepted (with revisions): February 16, 2008
MATERIAL AND METHODS
24
© 2008 Elsevier Inc. All Rights Reserved
Subjects After institutional ethics review board approval, 50 healthy volunteers were recruited during a 2-week period and informed 0090-4295/08/$34.00 doi:10.1016/j.urology.2008.02.033
consent was obtained. The exclusion criteria included a history of bladder surgery or pathologic features, age younger than 18 years, weight greater than 420 lb, or pregnancy. The subjects underwent scanning with no instructions regarding hydration or voiding.
Equipment The Bladderscan BVI-3000 (Verathon, Bothell, Wash) uses a 2-MHz volumetric probe with a transducer that is steered to collect 12 cross-sectional images (15° apart) in one step and provide a coronal bladder outline on a monitor in the shape of a target with cross-hairs (BVI 3000 size 23 ⫻ 32 ⫻ 7 cm, weight 2.25 kg, cost $17,423). The Bardscan (Bard, Crawley, UK) uses a 3.5/5-MHz mechanical sector probe attached to an 8-in. color screen (640 ⫻ 480 pixels; Bardscan size 25 ⫻ 36 ⫻ 9 cm, weight 3 kg, cost $18,870). The default setting of 3.5 MHz was used in this study. The HDI 4000 (Phillips Medical Systems, Best, The Netherlands) is a 3D-US system incorporating a curved array, 3 to 5-MHz volumetric probe. With the probe held fixed, the array is automatically steered to acquire real-time 3D images on a color screen (1024 ⫻ 768 pixels).
Study Design The measurements were obtained according to the manufacturers’ instructions. For the Bladderscan, the probe was fixed superior to the symphysis pubis and directed toward the bladder. The scans were performed until the cross-hairs were iso-centered on the bladder outline on the monitor, at which point, three repeated measurements were recorded, and the greatest was used as the volume. Compared with the average or single reading, the greatest reading had the strongest correlation11 and least bias.6 With the Bardscan, the probe was adjusted to view the bladder in its largest cross-section. On pressing the probe, the picture freezes and an automatic volume tool outlines the bladder. If this does not correspond with the actual bladder outline, the measurement can be repeated. The manufacturer recommends scans in two planes (transverse and sagittal) for increased accuracy. Using this mode, the greatest of three repeated measurements was used as the volume. For the 3D-US, the probe was centered on the bladder to acquire the whole bladder within a single acquisition. The bladder volume was calculated at a later date by a single operator (K.R.G.), who was unaware of the results using the in-built virtual organ computer-aided analysis software. This automatically extracts contours to produce a 3D model and volume. A line appears along the bladder outline displayed in a multiplanar view. The line can be adjusted if the fit is suboptimal. The subjects were scanned in the supine position, without changing their position and with no delay between readings, by 1 of 2 physicians. The time taken to scan a subject with all three instruments was recorded. Both operators had previous experience with the Bladderscan and had undergone training for the Bardscan and 3D-US with the company representative. The instruments were used in a random sequence determined from random number tables. On completion of scanning, the subjects immediately voided and underwent repeat scanning with the 3D-US to determine the RV. The voided urine was measured using graduated measuring devices and compared with the ultrasound measurements. UROLOGY 72 (1), 2008
Figure 1. Scanned volume versus voided volume using all three ultrasound systems (line of equality shown).
Statistical Analysis Patients with a RV were excluded from the final analysis. The strength of the relationship between the voided and calculated volumes was assessed using Pearson’s correlation coefficient. The differences against the voided volume were assessed with the paired t test. Because the average of the readings between methods acts as the best estimate of the unknown true value, Bland-Altman plots and limits of agreement were calculated, including bias (mean difference), precision (1.96 ⫻ standard deviation of differences), and standard error.14 Bias and precision help to calculate the 95% range of readings for a measured value. For example, when a device with a bias of 20 mL measures the volume at 400 mL, the mean true urinary volume will be 420 mL, ranging from (420 mL ⫺ precision) to (420 mL ⫹ precision). All calculations were performed using Prism for Macintosh, version 4.0 (GraphPad Software, San Diego, Calif). P ⬍0.05 was considered statistically significant.
RESULTS In total, 27 male and 23 female subjects were scanned. The mean ⫾ SD age of subjects was 35.7 ⫾ 10.9 years (range 19 to 60). One operator (K.R.G.) scanned 35 subjects and the remainder were scanned by the second (J.P.). The mean ⫾ SD time taken to scan the subjects using all three instruments was 6.2 ⫾ 2.1 minutes (range 2.9 to 13.1). Sixteen patients (32%) had a RV (range 4 to 302 mL) and were excluded. In 1 subject, 3D-US could not calculate the volume because the bladder was too large for complete acquisition. This subject was excluded from the 3D-US group. In the remaining 34 patients, the mean ⫾ SD voided volume was 252.9 ⫾ 167.4 mL (range 33 to 709). Figure 1 shows the volumes obtained using all three instruments plotted against the voided volumes. An excellent correlation (P ⬍0.001) was found between the voided and calculated volume using the Bladderscan (r ⫽ 0.98), Bardscan (r ⫽ 0.97), or the 3D-US system (r ⫽ 0.99). The mean calculated volumes, bias, precision, and standard error for all three instruments are listed in Table 1. The volumes calculated using the Bardscan were significantly different from the voided volumes (t ⫽ 2.84, P ⫽ 0.0076). No significant differences were found between 25
Table 1. Mean difference and 95% confidence intervals and standard error in calculation of voided volume using Bladderscan, Bardscan, and 3D-US system Instrument Bladderscan Bardscan 3D-US
Mean ⫾ SD Volume (mL)
Bias (mL)
Precision (mL)
SEM Difference (mL)
245.5 ⫾ 189.7 (17–856) 231.5 ⫾ 152.8 (0–595) 233.5 ⫾ 149.7 (13.1–671.6)
7.4 21.4* 5.6
81.0 85.9 34.4
7.1 7.5 3.1
SD ⫽ standard deviation; SEM ⫽ standard error of mean; 3D-US ⫽ three-dimensional ultrasonography. * Compared with mean voided volume (P ⫽ 0.0076).
Figure 2. Bland-Altman analyses of voided and scanned volume (n ⫽ 34). Average of voided volume plus scanned volume (x axis) plotted against difference of voided minus scanned volume (y axis). Straight horizontal line indicates mean difference (bias) and dotted horizontal lines indicate 95% limits of agreement (LA). (A) Bardscan bias, 21.4 mL and 95% LA, ⫺64.5 to 107.2 mL; (B) Bladderscan bias, 7.4 mL and 95% LA ⫺73.7 to 88.4 mL; (C) 3D-US bias, 5.6 mL and 95% LA ⫺28.9 to 40.0 mL.
COMMENT
Figure 3. Difference between voided volume and scanned volume for entire range of voided volumes using all three ultrasound systems.
the voided and calculated volumes using the Bladderscan (t ⫽ 1.82, P ⫽ 0.08) or 3D-US system (t ⫽ 1.04, P ⫽ 0.31). The 3D-US system had the smallest mean difference (bias), precision, and standard error and was therefore the most accurate of all three instruments (Table 1). BlandAltman plots demonstrating the limits of agreement between the voided and scanned volumes are presented in Figure 2. The 3D-US system displayed the best level of clinical agreement. The superiority of the 3D-US system in calculating the bladder volumes across the entire range of voided volumes is demonstrated in Figure 3. The Bladderscan and Bardscan were inaccurate when faced with the large bladder volumes, with the Bardscan underestimating and the Bladderscan overestimating the measured volumes. 26
Earlier Bladderscan models (BVI 2000) used a two-plane technique with an algorithm based on the bladder as an ellipsoid.15,16 Later models (BVI 2500) using one-step volumetric scanning performed better.12 Subsequent studies found that the Bladderscan overestimated5 or underestimated the volumes compared with the catheterized volumes.7,17,18 We used the BVI 3000, which has a faster scan time and larger scan plane and has shown good correlation with catheterized volumes.3,19 –21 However, previous studies have incorrectly inferred the accuracy using correlation coefficients rather than the more robust bias, precision, and standard error we used. Very few studies have performed Bland-Altman analyses on measurements.6,10,22 Also, most studies have involved patients with abnormalities in bladder anatomy, which might have influenced the results. Our results have shown excellent correlation between the ultrasound-determined and voided volumes using all three instruments. Neither the Bladderscan nor the 3D-US system provided bladder volumes significantly different from the voided volume. In contrast, Bardscan volumes were significantly different from the voided volumes. The 3D-US system demonstrated the strongest agreement and was the most accurate of all three instruments. This should not be surprising, because we used a high performance 3D-US system. In contrast, the performance of the smaller, inexpensive, and easier to use Bladderscan was encouraging. The number of studies comparing the Bladderscan and CUS is limited.10,23–25 Bano et al.23 compared the BVI UROLOGY 72 (1), 2008
3000 bladder volumes with the suprapubic catheter and CUS-determined volumes in 26 postoperative women. Because of the strong positive linear association between all three methods (r ⫽ 0.97), the investigators thought the Bladderscan should be preferred over CUS despite no assessment of the bias and whether the volumes differed significantly. Repeated measurements were not taken, and the effect of the suprapubic catheter on the Bladderscan measurements was not considered. Dudley et al.10 compared the BVI 2500⫹ with CUS in 11 pediatric patients. They found good correlation (r ⫽ 0.97) between instruments. Compared with the voided volumes, the mean absolute error using the Bladderscan was more than twice that using CUS. In a larger study, Huang et al.24 also found that stationary CUS was more accurate than the Bladderscan BVI 3000. In contrast, Byun et al.25 using true bladder volumes determined by fluoroscopy and catheterization as the reference standard, found CUS significantly underestimated the bladder volume by 21.8% compared with 3.3% using the Bladderscan BVI 3000. In the only previous study of the Bardscan, the ultrasound bladder volumes were compared with the catheterization volumes in 54 female patients.13 Only single readings were taken, and the recommended two-plane method was not routinely used. The Bardscan had a correlation of r ⫽ 0.98 (P ⫽ 0.16), a mean difference of 17.8 mL, and a precision of 114.3 mL. Although the values for precision between the Bardscan and Bladderscan in our study were quite similar, this was offset by the significantly greater bias in the Bardscan, which made it less accurate in comparison. Despite the lower accuracy, clinically, the Bardscan remains a useful device because it can assess the bladder anatomy and rule out large tumors or stones during volume calculation. The CUS volume calculations are determined using measurements in two planes, with geometric assumptions that are less robust compared with those used with 3D-US methods.26 Also, a wide discrepancy exists in the formulas and correction factors used for volume calculation with CUS.27 A previous study of bladder volume calculation using a freehand 3D-US system incorporating an electromagnetic position sensor found that 3D volume measurements had a mean absolute error of 4.9% compared with 27.5% for CUS.28 We have shown that the volumetric 3D method should be considered the reference standard for ultrasound bladder volume calculation. However, it is more expensive, and the processing time is more demanding. Ours was a small study, and more participants would have made the data more statistically significant. The instruments in this study were tested in young healthy individuals, not older patients with prostate or bladder disorders. In this study of volunteers, we were unable to obtain ethical approval for invasive catheterization. By excluding participants with a RV, we tried to limit the error inherent in the voided measurements. Even studies UROLOGY 72 (1), 2008
using catheterized volumes do not obtain true bladder volumes, unless emptying is confirmed on fluoroscopy, because up to 30% of patients can have a RV after catheterization.20 Also, the use of three instruments in this study lengthened the scanning time, which could have affected the bladder volume because of ongoing diuresis. Furthermore, only one model of each scanner was tested. Thus, any variation between scanners of the same model was not assessed. Variations of this nature could account for some of the conflicting results in the published data, and this matter constitutes the basis for additional research in our department.
CONCLUSIONS The Bardscan has the advantages of real-time bladder scanning and instantaneous volume calculation incorporated into a portable system, but it is not as accurate as the Bladderscan. The Bladderscan BVI3000, although it has its limitations, uses a method that comes close to the reference standard 3D volume assessment using ultrasonography. Acknowledgment. To Bard UK Ltd., Verathon Medical UK Ltd., and Philips Medical Systems UK for supplying the equipment used in this study. References 1. Barrington JW, Arunkalaivanan AS, and Abdel Fattah M: The accuracy of Bladderscan in intrapartum care. Int Urogynecol J Pelvic Floor Dysfunct 14: 214-215, 2003. 2. Cooperberg MR, Chambers SK, Rutherford TJ, et al: Cystic pelvic pathology presenting as falsely elevated post-void residual urine measured by portable ultrasound bladder scanning: report of 3 cases and review of the literature. Urology 55: 590, 2000. 3. Yucel S, Kocak H, Sanli A, et al: How accurate is measuring postvoid residual volume by portable abdominal ultrasound equipment in peritoneal dialysis patient? Neurourol Urodyn 24: 358361, 2005. 4. Barrington JW, Jones A, Robinson J, et al: Estimation of bladder volume using portable ultrasound in clam enterocystoplasty patients. J Urol 155: 82-83, 1996. 5. Goode PS, Locher JL, Bryant RL, et al: Measurement of postvoid residual urine with portable transabdominal bladder ultrasound scanner and urethral catheterization. Int Urogynecol J Pelvic Floor Dysfunct 11: 296-300, 2000. 6. Brouwer TA, Eindhoven BG, Epema AH, et al: Validation of an ultrasound scanner for determining urinary volumes in surgical patients and volunteers. J Clin Monit Comput 15: 379-385, 1999. 7. Alnaif B, and Drutz HP: The accuracy of portable abdominal ultrasound equipment in measuring postvoid residual volume. Int Urogynecol J Pelvic Floor Dysfunct 10: 215-218, 1999. 8. Rosseland LA, Bentsen G, Hopp E, et al: Monitoring urinary bladder volume and detecting post-operative urinary retention in children with an ultrasound scanner. Acta Anaesthesiol Scand 49: 1456-1459, 2005. 9. Wyneski HK, McMahon DR, Androulakakis V, et al: Automated bladder scan urine volumes are not reliable in complex neonatal cases. J Urol 174: 1661-1662, 2005. 10. Dudley NJ, Kirkland M, Lovett J, et al: Clinical agreement between automated and calculated ultrasound measurements of bladder volume. Br J Radiol 76: 832-834, 2003. 11. Marks LS, Dorey FJ, Macairan ML, et al: Three-dimensional ultrasound device for rapid determination of bladder volume. Urology 50: 341-348, 1997.
27
12. Coombes GM, and Millard RJ: The accuracy of portable ultrasound scanning in the measurement of residual urine volume. J Urol 152: 2083-2085, 1994. 13. Abdel-Fattah M, and Barrington JW: The accuracy of Bardscan: a new tool for the measurement of the bladder volume. J Obstet Gynaecol 25: 186-188, 2005. 14. Bland JM, and Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307-310, 1986. 15. Massagli TL, Cardenas DD, and Kelly EW: Experience with portable ultrasound equipment and measurement of urine volumes: inter-user reliability and factors of patient position. J Urol 142: 969-971, 1989. 16. Fuse H, Yokoyama T, Muraishi Y, et al: Measurement of residual urine volume using a portable ultrasound instrument. Int Urol Nephrol 28: 633-637, 1996. 17. Borrie MJ, Campbell K, Arcese ZA, et al: Urinary retention in patients in a geriatric rehabilitation unit: prevalence, risk factors, and validity of bladder scan evaluation. Rehabil Nurs 26: 187-191, 2001. 18. Schott-Baer FD, and Reaume L: Accuracy of ultrasound estimates of urine volume. Urol Nurs 21: 193-195, 2001. 19. Demaria F, Amar N, Biau D, et al: Prospective 3D ultrasonographic evaluation of immediate postpartum urine retention volume in 100 women who delivered vaginally. Int Urogynecol J Pelvic Floor Dysfunct 15: 281-285, 2004. 20. Barrington JW, Edwards G, Ashcroft M, et al: Measurement of bladder volume following cesarean section using Bladderscan. Int Urogynecol J Pelvic Floor Dysfunct 12: 373-374, 2001.
28
21. Araki Y, Ishibashi N, Sasatomi T, et al: Effectiveness of the portable ultrasound bladder scanner in the measurement of residual urine volume after total mesorectal extirpation. Minim Invasive Ther Allied Technol 12: 245-248, 2003. 22. De Gennaro M, Capitanucci ML, Di Ciommo V, et al: Reliability of bladder volume measurement with BladderScan in paediatric patients. Scand J Urol Nephrol 40: 370-375, 2006. 23. Bano F, Arunkalaivanan AS, and Barrington JW: Comparison between Bladderscan, real-time ultrasound and suprapubic catheterisation in the measurement of female residual bladder volume. J Obstet Gynaecol 24: 694-695, 2004. 24. Huang YH, Bih LI, Chen SL, et al: The accuracy of ultrasonic estimation of bladder volume: a comparison of portable and stationary equipment. Arch Phys Med Rehabil 85: 138-141, 2004. 25. Byun SS, Kim HH, Lee E, et al: Accuracy of bladder volume determinations by ultrasonography: are they accurate over entire bladder volume range? Urology 62: 656-660, 2003. 26. Farrell T, Leslie JR, Chien PF, et al: The reliability and validity of three dimensional ultrasound volumetric measurements using an in vitro balloon and in vivo uterine model. Br J Obstet Gynecol 108: 573-582, 2001. 27. Hvarness H, Skjoldbye B, and Jakobsen H: Urinary bladder volume measurements: comparison of three ultrasound calculation methods. Scand J Urol Nephrol 36: 177-181, 2002. 28. Riccabona M, Nelson TR, Pretorius DH, et al: In vivo threedimensional sonographic measurement of organ volume: validation in the urinary bladder. J Ultrasound Med 15: 627-632, 1996.
UROLOGY 72 (1), 2008