Developing a preoperative predictive model for ureteral length for ureteral stent insertion
Kawahara et al. BMC Urology
Developing a preoperative predictive model for ureteral length for ureteral stent insertion
Takashi Kawahara 0 1
Kentaro Sakamaki 2
Hiroki Ito 0 1
Shinnosuke Kuroda 0
Hideyuki Terao 0
Kazuhide Makiyama 1
Hiroji Uemura 1
Masahiro Yao 1
Hiroshi Miyamoto 3
Junichi Matsuzaki 0
0 Department of Urology, Ohguchi Higashi General Hospital , Yokohama , Japan
1 Department of Urology, Yokohama City University, Graduate School of Medicine , 3-9 Fukuura, Kanazawa-ku, Yokohama, Kanagawa 2360004 , Japan
2 Departments of Biostatistics, Yokohama City University Graduate School of Medicine , Yokohama , Japan
3 Departments of Pathology and Urology, Johns Hopkins University School of Medicine , Baltimore , USA
Background: Ureteral stenting has been a fundamental part of various urological procedures. Selecting a ureteral stent of optimal length is important for decreasing the incidence of stent migration and complications. The aim of the present study was to develop and internally validate a model for predicting the ureteral length for ureteral stent insertion. Methods: This study included a total of 127 patients whose ureters had previously been assessed by both intravenous urography (IVU) and CT scan. The actual ureteral length was determined by direct measurement using a 5-Fr ureteral catheter. Multiple linear regression analysis with backward selection was used to model the relationship between the factors analyzed and actual ureteral length. Bootstrapping was used to internally validate the predictive model. Results: Patients all of whom had stone disease included 76 men (59.8%) and 51 women (40.2%), with the median and mean (± SD) ages of 60 and 58.7 (±14.2) years. In these patients, 53 (41.7%) right and 74 (58.3%) left ureters were analyzed. The median and mean (± SD) actual ureteral lengths were 24.0 and 23.3 (±2.0) cm, respectively. Using the bootstrap methods for internal validation, the correlation coefficient (R2) was 0.57 ± 0.07. Conclusion: We have developed a predictive model, for the first time, which predicts ureteral length using the following five preoperative characteristics: age, side, sex, IVU measurement, and CT calculation. This predictive model can be used to reliably predict ureteral length based on clinical and radiological factors and may thus be a useful tool to help determining the optimal length of ureteral stent.
Ureteral length; Ureteral catheter; Predictive model; Ureter; Ureteral stent
Since the first description by Zimskind et al. in 1967 ,
ureteral stenting has been a fundamental part of various
urological procedures for the treatment of obstructing
ureteral calculi, ureteral stricture, ureteropelvic junction
obstruction, retroperitoneal tumor and fibrosis, and the
procedures that were developed after open or
endoscopic ureteral surgery . Placement of a stent that is
too long often causes complications, such as frequent or
urgent urination, incontinence, hematuria, and bladder
or flank pain, which have a negative impact on quality of
life of patients [3–9]. Conversely, a short ureteral stent
increases the risk of migration, resulting in
complications that require retraction and replacement [10, 11].
Thus, choosing a stent of optimal length is important
for reducing the incidence of stent migration and other
complications [10, 12–14]. In our previous study, 4% of
the ureteral stents were found to be too short and 19.5%
were too long .
Actual ureteral stent measurement is the most
accurate method to measure the ureteral length. However,
actual stent measurement requires additional radiation
exposure around 5.2 s and operation time of 0.2 min
© The Author(s). 2016 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
. Moreover, in some hospitals, assorted lengths of
ureteral stents are not stocked. Therefore, preoperative
prediction of ureteral length is often needed. We previously
reported the usefulness of direct measurement with a
ureteral catheter in predicting the actual ureteral length .
In another study, we showed the reliability of multiple
modalities for approximating the ureteral length in the
same group of patients . The length from renal vein to
ureteral orifice measured by CT scan (axial CT distance:
ACTD) showed a stronger association than any other
variables tested, including body height (BH) and intravenous
urography (IVU) measurement . Accurately predicting
the ureteral length is necessary for determining the
optimal stent length [10, 12].
The aim of the present study was to develop and
internally validate a preoperative predictive model for
predicting the actual ureteral length (AUL).
We measured the ureteral length in 362 patients by
direct measurement, as described below, at Ohguchi
Higashi General Hospital (Yokohama, Japan) from 2010
to 2014. In these patients, a total of 127 ureters had
previously been assessed by both IVU and CT scan,
preoperatively. This study was approved by the Review
Board of Kanagawa Prefecture Medical Association
(Approved number: H25-3169) and written informed
consent was obtained from the patients for their data to
be used for research purpose.
Measurement of actual ureteral length
The AUL, which was defined as the length between the
ureteropelvic junction (UPJ; detected by fluoroscopy
after retrograde pyelography) and the ureteral orifice
(detected by cystoscopy), was determined by direct
measurement using a 5-Fr ureteral catheter (TigerTail®,
BARD, Murray Hill, NJ, USA), as we previously
described . In order to avoid overly straightening of the
ureter, which might decrease its length, we measured the
AUL, using a ureteral catheter without the guidewire.
The correlations between the actual ureteral length and
the following variables were assessed: (1) BH, (2) age, (3)
sex, (4) side, (5) IVUa (the linear distance from the UPJ to
the uretero-vesicle junction [UVJ] by IVU), (6) IVUb (the
linear distance from the mid-kidney to the UVJ by IVU),
(7) IVUc (the ureteral trace was measured by tracing the
ureter from the ureteropelvic junction to the
ureterovesical junction and measuring the length of the trace with a
flexible ruler which was in the form of a string by IVU),
and (8) the distance from the level of the renal vein to the
ureteral trace by non-contrast ACTD. ACTD was the
distance from the renal vein level to the urinary orifice
level. It was defined as the number of CT slices multiplied
by the distance between each slice using axial CT. The
upper slice was defined as the level of the renal vein, and
the lower slice was defined as the level of UVJ.
Variables were selected by backward selection. Multiple
linear regression analysis with backward selection was
used to model the relationship between the factors
analyzed and actual ureteral length. The selection criterion of
p < 0.1 was used for elimination of a variable. Some
variables were initially excluded from the model because of
multicollinearity. Before multiple linear regression, we
exclude IVUc, because of the strong association between
IVUa and IVUc (r = 0.924). Decisions with respect to the
coding of the predictive model variables were made before
variable selection. A bootstrapping method was a
nonparametric data generating method in which new datasets
were repeatedly generated from an original dataset. Using
the bootstrap methods for internal validation, the
coefficient of determination (variance explained, R2) was
assessed. Statistical analyses were performed using SAS
9.3 (SAS Institute Inc., Cary, NC, USA).
A total of 127 patients all with stone disease, including
76 men (59.8%) and 51 women (40.2%), were enrolled in
this study. The median and mean (± SD) ages were 60
and 58.7 (±14.2) years (range: 28 - 86), respectively
(Table 1). Fifty-three (41.7%) were right ureters (41.7%)
and 74 (58.3%) were left ureters (58.3%). The median
and mean (± SD) AULs were 24.0 and 23.3 (±2.0) cm
(range: 16.0 – 27.5). Left side ureter are significantly
Table 1 Patients’ characteristics
IVU intravenous urography, ACTD axial CT distance
longer than right side (23.7 ± 1.69 versus 22.7 ± 2.2, P =
0.007) Ureteral length in male was not longer than in
female (23.4 ± 1.86 versus 23.1 ± 2.15, P = 0.350). On the
other hand, multiple linear regression showed that sex
was an independent factor (Table 2). The results of
multiple linear regression are presented in Table 2. This
predictive model contained five characteristics,
including age, side, sex, IVUa, and ACTD. The total score
of the predictive model was derived from the sum of
the individual scores of each predictive variable
(Fig. 1). The predicting ureteral length was written as
follow formula; Ureteral Length = 10.1 + 0.27 x ACTD
(cm) - 0.016 x Age + 0.302 x IVUa (cm) + 0.504 x side
(Right:0, Left:1) + 0.716 x Sex (Male:0, Female:1).
Using the bootstrap method, the explained variance
(R2) was 0.57 ± 0.07 (range; 0.291 to 0.805).
We previously reported the effectiveness of loop-type
ureteral stent that decreased the prevalence of stent-related
symptoms, compared with double pigtail ureteral stents.
Significantly lower scores for almost all of stent-related
symptoms, other than nocturia, were seen with loop-type
ureteral stents than in double pigtail stents . On the
other hand, even when using with loop-type ureteral stent,
the stent-related symptoms were not prevented with
overlong ureteral stent position (unpublished data). We
previously demonstrated the relationship between the AUL and
the ureteral stent position using direct ureteral length
measurements in 226 patients with loop-type ureteral
stents . The migration rate and overlong rate were
correlated with the ureteral length, when the proximal
end of the stent was in the renal pelvis. The appropriate
length of ureteral stent was the same or up to 1 cm
shorter than the measured ureteral length. In that study
89.0% of loop-type ureteral stents with the same length of
the ureter were appropriately positioned, whereas 11.0%
seemed to be overlong than AUL. When loop-type
ureteral stents were 1 cm shorter than AUL were placed,
88.6% were in appropriate positions, 8.6% were too long,
and 2.9% had migrated into the ureter. In addition,
looptype ureteral stents that were more than 2 cm shorter than
the measured ureteral length increased the incidence of
migration to more than 10% . Direct measurement
Table 2 Multiplelinear regression for ureteral length (cm)
Fig. 1 Predictive model predicting ureteral length
with a ureteral catheter is thought to be the optimal
method for selecting the appropriate length of ureteral
stent, but the patient and operator are exposed to
additional radiation and an extra procedure is required. Thus,
while the direct measurement of the ureteral length is
ideal, this predictive model may represent a new, more
accurate method for estimating the ureteral length.
It is easy to apply the BH in predicting the ureteral
length. Various studies have shown the association
between BH and the ureteral length or the ureteral stent
position [3, 16, 19, 20]. However, Shah et al.
demonstrated that they were not significantly associated .
Gregory et al. reported the association of some variables,
including body habitus, with ureteral lengths based on
Vitruvius’ and da Vinci’s theories of proportion . The
AUL was significantly correlated with the BH, followed
by the distances from the xyphoid process to the
umbilicus (X-P) and from the shoulder to wrist (S-W), but
they concluded that it was difficult to predict the
We previously investigated to define the best modality
for estimating ureteral length in patients undergoing
ureteral stent placement . ACTD showed the strongest
correlation (R2 =0.381) with AUL, compared with IVUa
(R2 = 0.274), IVUb (R2 = 0.230), IVUc (R2 = 0.206), BH
(R2 = 0.098), and body surface area (R2 = 0.095).
However, when inserting loop-type ureteral stent, more
accurate methods for predicting ureteral length are needed
because of easy to migrate to the ureter, because the
differences of 1 cm might change the risk of migration and
increase the ureteral stent related symptoms.
In the current study, we developed a predictive model
using the preoperative factors identified by multiple
regression analysis: age, side, sex, length measured by
IVU, and ACTD. Interestingly, IVUc which is a trace of
the whole ureteral length (according to the curve of
ureter) did not show a stronger correlation with the AUL.
We hypothesize that when we measured the AUL, the
ureteral catheter straightened the ureter, causing the AUL
to shorten. We chose IVUa because it was most highly
correlated with the AUL and because it was strongly
correlated with each of the IVU methods. To our
knowledge, this is the first report of a predictive model that
predicts ureteral length. This predictive model showed the
coefficient of determination (R2 = 0.566).
There are limitations in the present study. The endpoint
of this predictive model was ureteral length, but not stent
position or ureteral stent-related symptoms. We are now
conducting a study to assess the stent position, using this
predictive model. Furthermore, when a validated ureteral
stent symptom questionnaire (USSQ) becomes available
in Japan, we plan to conduct another study. The second
limitation is that this predictive model is for patients who
received both CT and IVU, not CT or IVU only. We
previously showed the formula for estimating the actual
ureteral length; however, this predictive model was found to
have greater accuracy . The third limitation is that we
did not assess the background characteristics of the
patients, including whether they were pregnant or had a
history of radiation treatment, which might have influenced
the length of the ureter. In this study, we only obtained
data from patients who were treated for stone disease. We
therefore speculate that these patients did not have a
significant effect on the predictive model. On the other hand,
this predictive model is not suitable for use with pregnant
patients or patients undergoing radiation therapy. The
forth one is that the same data was used for both variable
selection and parameter estimation there is a risk that
overfitting has occurred. Until an external validation is
conducted model fit estimates are like to be inflated.
We have developed a predictive model, for the first time,
which can precisely predict ureteral length, using the five
preoperative clinical and radiographical factors. This
predictive model may thus be a useful tool to help
determine the optimal length of a ureteral stent.
ACTD: Actual CT distance; AUL: Actual ureteral length; IVU: Intravenous
urography; UPJ: Uretero pelvic junction; UVJ: Uretero vesicle junction
Availability of data and materials
Due to ethical restrictions, the raw data underlying this paper is available
upon request from the corresponding author.
TK, HI, SK, HT, and JM performed the operation. TK conceived the study,
participated in its design and wrote the manuscript. KM, HU, HM, and MY
analyzed the data and corrected the draft. KS created and validated the
predictive model. All of the authors read and approved the final manuscript.
Ethics approval and consent to participate
This study was approved by the Review Board of Kanagawa Prefecture Medical
Association (Approved number: H25-3169) and written informed consent was
obtained from the patients for their data to be used for research purposes. A
copy of the written consent form is available for review from the Editor-in-Chief
of this journal.
1. Zimskind PD , Fetter TR , Wilkerson JL . Clinical use of long-term indwelling silicone rubber ureteral splints inserted cystoscopically . J Urol . 1967 ; 97 ( 5 ): 840 - 4 .
2. Borboroglu PG , Kane CJ . Current management of severely encrusted ureteral stents with a large associated stone burden . J Urol . 2000 ; 164 ( 3 Pt 1 ): 648 - 50 .
3. Ho CH , Huang KH , Chen SC , Pu YS , Liu SP , Yu HJ . Choosing the ideal length of a double-pigtail ureteral stent according to body height: study based on a Chinese population . Urol Int . 2009 ; 83 ( 1 ): 70 - 4 . doi:10.1159/000224872.
4. Joshi HB , Stainthorpe A , MacDonagh RP , Keeley Jr FX , Timoney AG , Barry MJ . Indwelling ureteral stents: evaluation of symptoms, quality of life and utility . J Urol . 2003 ; 169 ( 3 ): 1065 - 9 . doi:10.1097/01.ju. 0000048980 .33855.90. discussion 1069.
5. Pollard SG , Macfarlane R. Symptoms arising from Double-J ureteral stents . J Urol . 1988 ; 139 ( 1 ): 37 - 8 .
6. Kawahara T , Ito H , Terao H , Yoshida M , Matsuzaki J. Ureteral Stent Encrustation , Incrustation, and Coloring: Morbidity Related to Indwelling Times . J Endourol . 2011 . doi:10.1089/end.2011.0385.
7. Kawahara T , Ishida H , Kubota Y , Matsuzaki J. Ureteroscopic removal of forgotten ureteral stent . BMJ case reports . 2012 . doi:10.1136/bcr.02. 2012 .5736.
8. Kawahara T , Ito H , Terao H , Yamagishi T , Ogawa T , Uemura H , Kubota Y , Matsuzaki J. Ureteral stent retrieval using the crochet hook technique in females . PLoS One . 2012 ; 7 ( 1 ), e29292. doi:10.1371/journal.pone.0029292.
9. Kawahara T , Ito H , Terao H , Yamashita Y , Tanaka K , Ogawa T , Uemura H , Kubota Y , Matsuzaki J. Ureteral stent exchange under fluoroscopic guidance using the crochet hook technique in women . Urol Int . 2012 ; 88 ( 3 ): 322 - 5 . doi: 10.1159/000336870.
10. Slaton JW , Kropp KA . Proximal ureteral stent migration: an avoidable complication ? J Urol . 1996 ; 155 ( 1 ): 58 - 61 .
11. Breau RH , Norman RW . Optimal prevention and management of proximal ureteral stent migration and remigration . J Urol . 2001 ; 166 ( 3 ): 890 - 3 .
12. Wills MI , Gilbert HW , Chadwick DJ , Harrison SC. Which ureteric stent length? Br J Urol . 1991 ; 68 ( 4 ): 440 .
13. Pocock RD , Stower MJ , Ferro MA , Smith PJ , Gingell JC . Double J stents. A review of 100 patients. Br J Urol . 1986 ; 58 ( 6 ): 629 - 33 .
14. Chin JL , Denstedt JD . Retrieval of proximally migrated ureteral stents . J Urol . 1992 ; 148 ( 4 ): 1205 - 6 .
15. Kawahara T , Ito H , Terao H , Yoshida M , Ogawa T , Uemura H , Kubota Y , Matsuzaki J. Choosing an appropriate length of loop type ureteral stent using direct ureteral length measurement . Urol Int . 2012 ; 88 ( 1 ): 48 - 53 . doi:10. 1159/000332431.
16. Pilcher JM , Patel U. Choosing the correct length of ureteric stent: a formula based on the patient's height compared with direct ureteric measurement . Clin Radiol . 2002 ; 57 ( 1 ): 59 - 62 . doi:10.1053/crad.2001.0737.
17. Kawahara T , Ito H , Terao H , Yoshida M , Ogawa T , Uemura H , Kubota Y , Matsuzaki J. Which is the best method to estimate the actual ureteral length in patients undergoing ureteral stent placement? Int J Urol . 2012 ; 19 ( 7 ): 634 - 8 . doi:10.1111/j.1442- 2042 . 2012 .02998.x.
18. Kawahara T , Ito H , Terao H , Ogawa T , Uemura H , Kubota Y , Matsuzaki J. Changing to a loop-type ureteral stent decreases patients' stent-related symptoms . Urol Res . 2012 ; 40 ( 6 ): 763 - 7 . doi:10.1007/s00240- 012 - 0500 -4.
19. Paick SH , Park HK , Byun SS , Oh SJ , Kim HH . Direct ureteric length measurement from intravenous pyelography: does height represent ureteric length? Urol Res . 2005 ; 33 ( 3 ): 199 - 202 . doi:10.1007/s00240- 004 - 0461 -3.
20. Lee BK , Paick SH , Park HK , Kim HG , Lho YS . Is a 22 cm Ureteric Stent Appropriate for Korean Patients Smaller than 175 cm in Height? Korean J Urol . 2010 ; 51 ( 9 ): 642 - 6 . doi:10.4111/kju.2010. 51 .9. 642 .
21. Shah J , Kulkarni RP . Height does not predict ureteric length . Clin Radiol . 2005 ; 60 ( 7 ): 812 - 4 . doi:10.1016/j.crad. 2004 .08.018.
22. Hruby GW , Ames CD , Yan Y , Monga M , Landman J. Correlation of ureteric length with anthropometric variables of surface body habitus . BJU Int . 2007 ; 99 ( 5 ): 1119 - 22 . doi:10.1111/j.1464-410X.2007.06757.x.