Journal of Pediatric Surgery xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Journal of Pediatric Surgery

journal homepage: www.elsevier.com/locate/jpedsurg

Factors associated with peritoneal dialysis catheter complications

in children☆,☆☆
Camille L. Stewart a,⁎, Shannon N. Acker a, Laura L. Pyle b,c, Ann Kulungowski a,d, Melissa Cadnapaphornchai b,e,
Jennifer L. Bruny a,d, Frederick Karrer a,d
a
Department of Surgery, University of Colorado School of Medicine, 12631 E. 17th Ave., Aurora, CO 80045, USA
b
Department of Pediatrics, University of Colorado School of Medicine, Aurora, 12631 E. 17th Ave., CO 80045, USA
c
Department of Biostatistics and Informatics, University of Colorado School of Public Health, 13001 E 17th Pl., Aurora, CO 80045, USA
d
Division of Pediatric Surgery, Children's Hospital Colorado, 13123 East 16th Ave., Aurora, CO 80045, USA
e
Division of Pediatric Nephrology, Children's Hospital Colorado, 13123 East 16th Ave., Aurora, CO 80045, USA

a r t i c l e i n f o a b s t r a c t

Article history: Background/Purpose: Peritoneal dialysis (PD) is a common method of renal replacement therapy for children.
Received 3 October 2015 However, placement of PD catheters has risk, and some are never used.
Accepted 9 October 2015 Methods: We conducted a retrospective chart review of children with a PD catheter placed between 2000 and
Available online xxxx 2014. Logistic regression analyses were used to identify covariates associated with complications.
Results: We identified 175 children with PD catheters. 110 complications developed in 80 children (45.7%). Com-
Key words:
plications including unexpected return to the operating room and peritonitis increased as the length of time a
Pediatric peritoneal dialysis
Pediatric renal replacement therapy
catheter was in place increased. Children who weighed b 12.4 kg had 3.2 times greater odds of developing a
leak (95% CI 1.21–8.63, p = 0.02). Twelve children never used their PD catheters, 9 with acute kidney injury
(AKI) who recovered from their disease more quickly than expected. No covariate was associated with nonuse.
Conclusions: Complications with PD catheters are common and increase the longer catheters are in place. Lower
weight children are at greater risk of PD catheter leak. Decreased initial volumes of dialysate in smaller children
may mitigate this risk. Nonuse may be reduced if dialysis is permitted the day of placement for children with AKI.
© 2015 Elsevier Inc. All rights reserved.

Peritoneal dialysis (PD) is the preferred method of renal replace-
ment therapy for children with kidney failure at many centers. This is
true for a variety of reasons, including its simplicity, safety, and applica-
bility to children with a broad range of ages and sizes [1,2]. Vascular ac-
cess can be challenging in younger children and PD preserves vascular
access should it be needed later in life. Furthermore, growth rates and
residual renal function are improved with PD compared to hemodialysis
(HD) [3]. For children who need ongoing renal replacement therapy in
the outpatient setting, PD is also ideal because it can be performed at
night and at home, preventing disruption of school and other daytime
activities [1]. Despite these benefits, PD catheters have a high rate of
complications, many which must be addressed operatively. We also
☆ Authorship contributions: C.S.: study conception, study design, literature review, data
collection, data interpretation, writing the manuscript; S.A.: study design, writing the
manuscript; L.P.: data analysis, data interpretation, critical revisions to the manuscript;
A.K.: study conception, critical revisions to the manuscript; M.C.: study design, data inter-
pretation, critical revisions to the manuscript; J.B.: study conception, critical revisions to
the manuscript; F.K.: study conception, study design, critical revisions to the manuscript.
☆☆ Level of evidence: III.
⁎ Corresponding author at: Department of Surgery, University of Colorado School of
Medicine, 12631 E. 17th Ave., C302, Aurora, CO 80045, USA. Tel.: +1 303 724 2685;
fax: +1 303 724 2682.
E-mail address: Camille.Stewart@UCDenver.edu (C.L. Stewart).
noted in our center that some catheters we placed in children with
acute kidney injury (AKI) were ultimately never used. We sought to
identify features that were associated with PD catheter complications
and catheter nonuse to help us predict and possibly avoid these events.
Although previous studies have addressed similar issues [4–8], few have
used advanced statistics to identify predictors, and no previous studies
included nonuse as a variable of interest.
1. Materials and methods
After IRB approval, we performed a retrospective chart review of all
children who received a PD catheter at our hospital from January 2000
to January 2014. All children received a Tenckhoff curled double-cuffed
peritoneal dialysis catheter. PD catheter placement was requested by
the pediatric nephrology service for acute kidney injury (AKI) or chron-
ic kidney disease (CKD). Estimated creatinine clearance was calculated
using the bedside CKiD equation [9]. For children with AKI, two differ-
ent scoring systems were used to help quantify the degree of illness.
The pRIFLE score was used to quantify the severity of AKI [10]. It is cal-
culated based on estimated creatinine clearance and urine production,
and is an acronym for Risk of renal dysfunction, Injury to the kidney,
Failure of kidney function, Loss of kidney function, and End-stage kid-
ney disease [10]. The PRISM III-24 score was calculated to quantify the

http://dx.doi.org/10.1016/j.jpedsurg.2015.10.035
0022-3468/© 2015 Elsevier Inc. All rights reserved.

Please cite this article as: Stewart CL, et al, Factors associated with peritoneal dialysis catheter complications in children, J Pediatr Surg (2015),
http://dx.doi.org/10.1016/j.jpedsurg.2015.10.035
2 C.L. Stewart et al. / Journal of Pediatric Surgery xxx (2015) xxx–xxx
overall severity of disease [11]. This score is a value that correlates with
pediatric mortality and is calculated based on the following variables:
systolic blood pressure, temperature, mental status, heart rate, pupillary
reflexes, acidosis, pH, pCO2, total CO2, arterial PaO2, glucose, potassium,
creatinine, urea, white blood cell count, prothrombin time, partial
thromboplastin time, and platelets [11]. PRISM III-24 was calculated
based on the 24 hours preceding surgical consultation for PD catheter
placement. PD catheters were placed by 12 different fellowship-
trained pediatric surgeons. There was no standard operative technique
used for catheter placement but was instead based on the training and
discretion of the operating surgeon. It is, however, standard practice at
our institution to place a purse string suture around the posterior fascia
to secure the cuff in place in the rectus sheath, and to test drainage of
the catheter intraoperatively. Complications related to catheter place-
ment were reviewed. Peritonitis was defined as positive cultures from
peritoneal fluid; other infection was defined as an infection in the ab-
dominal wall around the catheter insertion site. Repair of inguinal her-
nias, umbilical hernias, and incisional hernias were considered
complications when they were directly related to the placement of a
PD catheter. Return to the operating room was defined as any operative
intervention that was related to the peritoneal dialysis catheter and
was unplanned.
2. Theory/calculation
Univariate and multivariate logistic regression models were used to
test whether covariates were associated with complications or catheter
nonuse. When significantly associated continuous covariates were iden-
tified, receiver operating characteristic area under the curve (ROC AUC)
analysis was used to identify a cutoff value that maximized the sensitiv-
ity and specificity for predicting the outcome of interest. All variables
were available for patients in the study except where noted otherwise.
The entire cohort was included in the analysis for complications, but
only children with a diagnosis of AKI were included in the analysis of
catheter nonuse. Covariates tested for association with complications
were selected to help identify predictive or modifiable features for chil-
dren with AKI or CKD, and included age, weight (n = 153), PRISM III-24
score (n = 76), laparoscopic or open procedure, number of ports used
(n = 173), simultaneous placement of a gastrostomy, number of days
from placement to first use (n = 151), continuous or intermittent use
(n = 135), number of days used (n = 167), the number of days the
catheter was in place (n = 155), and the surgeon name. Of note, models
for number of ports and surgeon name would not converge. Covariates
tested for association with catheter nonuse were selected to help iden-
tify features associated with renal recovery in children with AKI, and in-
cluded age, weight (n = 82), body mass index (n = 59), prior
hemodialysis, days from request to placement of the catheter (n =
82), urine production (n = 83, normal urine output, oliguric
(b 0.5 mL/kg/hr), or aneuric), serum potassium (n = 85), blood urea ni-
trogen (n = 85), serum creatinine (n = 87), pRIFLE score (n = 70),
PRISM III score (n = 68), days from placement to first use (n = 80),
complications (presence or absence), and number of days in the inten-
sive care unit (n = 82). The models for urine production and pRIFLE
score did not converge.
3. Results
3.1. Demographics
We identified 175 children who had PD catheters placed by 12 dif-
ferent pediatric surgeons during the study period. Each surgeon per-
formed between 3 and 61 PD catheter insertions. There were 32
children who had multiple catheters placed during the study period
(24 with CKD, 7 with AKI); only the first catheter insertion was included
in the study to facilitate statistical analysis, for a total of 175 PD cathe-
ters. There were 86 females and 89 males, with an average age of
Please cite this article as: Stewart CL, et al, Factors associated with peritoneal dialysis catheter complications in children, J Pediatr Surg (2015),
http://dx.doi.org/10.1016/j.jpedsurg.2015.10.035
6.0 ± 0.4 years (range 0.11–19.2 years), and average weight of
22.5 ± 1.4 kg (range 2.5–103.0 kg). Of these children 91 (52.0%) were
diagnosed with AKI, 84 (48.0%) with CKD, and 170 survived to discharge
(97.1%). The majority (138, 78.9%) of the catheters were placed
laparoscopically, using an average of 2 ports, and the majority also in-
cluded an omentectomy (135, 77.1%). Most catheters were placed with-
in 1 day of the primary team's request when the child had AKI (69/84,
82.1%). For children with AKI, the average number of days from place-
ment to first use was 1.0 ± 0.1 days (range 0–6 days); 33 children
with AKI used their PD catheter the day it was placed. Average follow-
up for each child was 3.6 ± 0.3 years (range 0.1–15.8 years).
3.2. Factors associated with complications
There were 80 children (45.7%) who experienced 110 complications,
36 (45.0%) with AKI and 44 (55.0%) with CKD. The most common com-
plication was poor drainage (Table 1). Examples of complications listed
under “other” include catheter cracking, cuff erosion, and a scar revision.
A total of 49 (28.0%) children required an unexpected return to the
operating room.
In univariate analysis (Table 2), we found that overall complications
were associated with longer presence of the PD catheter (p = 0.003).
This was true specifically for return to the operating room (p = 0.02)
and peritonitis (p = 0.01). Return to the operating room was also
more likely with increased time from placement to first use (p =
0.03), and peritonitis was significantly associated with higher PRISM
III-24 scores (p = 0.02). With multivariate analysis, the only clinically
meaningful association found was leak with lower weight (OR 0.728
95% CI 0.567–0.935, p = 0.01).
ROC AUC analysis was used to identify values that maximized the
sensitivity and specificity for predicting clinical outcomes of interest
(Table 3, Fig. 1). We found that children with their PD catheter in
place for N150 days had 5.22 times greater odds of returning to the op-
erating room for procedures related to their catheter (p b 0.001, Fig. 1a),
and that children who waited N3 days to use their PD catheter after
placement had 2.14 greater odds of returning to the operating room
for procedures related to their catheter (p = 0.04, Fig. 1b). Children
with PRISM III-24 scores N15 had 6.43 greater odds of developing peri-
tonitis (p = 0.02, Fig. 1c), where scores N 8 (range 0–74) generally sig-
nify critical illness [11]. Children who weighed below 12.4 kg had 3.23
times the odds of developing a leak (p = 0.02, Fig. 1d).
3.3. Factors associated with nonuse
There were 12 children who did not use their PD catheter, nine with
AKI and three with CKD. All nine of the children with AKI ultimately re-
covered from their acute illness and never used their catheters because
of more expeditious recovery than expected. PD was attempted in three
of these children but was unsuccessful because of poor drainage, and
only one child required HD prior to full renal recovery. Reasons for non-
use among children with CKD included kidney transplant in one,
Table 1
Complications associated with PD catheters.
Complication Number (% of all complications)
Poor drainage 38 (34.5)
Requiring return to OR 20 (18.2)
Leak 20 (18.2)
Requiring return to OR 15 (13.6)
Infection 32 (29.0)
Peritonitis 27 (24.5)
Other infection 5 (4.5)
Hernia 7 (6.4)
Umbilical 4 (3.6)
Inguinal 2 (1.8)
Incisional 1 (0.9)
Other (cracking, erosion, scar revision) 13 (11.8)
 C.L. Stewart et al. / Journal of Pediatric Surgery xxx (2015) xxx–xxx
Table 2
Significant results of univariate regression analysis.
Outcome Variable P-value OR (95% CI)
Any complication Days PD catheter in place 0.003 1.002 (1.001–1.002)
Peritonitis Days PD catheter in place 0.01 1.001 (1.000–1.002)
Peritonitis PRISM III-24 score 0.02 1.125 (1.020–1.241)
Return to OR Days PD catheter in place 0.02 1.001 (1.000–1.002)
Return to OR Days placement to 1st use 0.03 1.096 (1.008–1.190)
withdrawal of care in one, and parental preference after complications
in one. Logistic regression models were used in an attempt to identify
factors that were associated with catheter nonuse in children who
were diagnosed with AKI. No factors predicting nonuse were identified.
3.4. Discussion
Here we report our experience with PD catheters in 175 children
with AKI and CKD. The rates of complications and return to the operat-
ing room reported are comparable to those reported in similar studies
[5,6]. In our univariate analysis, we found that children with catheters
in for longer periods of time were more likely to have complications, in-
cluding peritonitis and return to the operating room. We also found that
children with higher PRISM III-24 scores, and therefore, who were more
severely ill, were more likely to develop peritonitis. Further, children
who had longer periods of time from placement to first use were
more likely to return to the operating room. In our multivariate analysis,
we found that lower weight children were more likely to develop leak,
and specifically that children weighing less than 12.4 kg were ~3 times
more likely to develop leaks.
While little can be changed about the duration a child requires PD, and
thus has a PD catheter, or how ill a child is when they require PD, the in-
formation presented here can be used to help prepare our patients and
their families for their future with PD. The association of patient weight
and leak rate may be more clinically applicable. Stone et al. [6] and
Phan et al. [14] both reported increased complications in infants, who
generally weigh less than 12.5 kg, and Cribbs et al. [5] reported increased
complications in children weighing less than 10 kg. Prior to our investiga-
tion, children with new PD catheters usually started at a dialysate volume
of 20 mL/kg. Based on our findings, we have decided to adjust the PD pre-
scription in lower weight children to provide lower dialysate volumes
with comparable clearance and ultrafiltration. The pass volume is often
decreased once PD catheter leak is evident, but we speculate that initiat-
ing lower pass volumes in this high-risk group may prevent catheter leak.
Further study will be performed to determine if this change in practice
decreases PD catheter leak rates in lower weight children.
Interestingly, we did not find an association between leak rate and
time from placement to first use in our analysis. In fact, we found that
children were twice as likely to return to the operating room if we
waited longer than 3 days to initiate PD. Previously, it had been our
practice to request delaying PD catheter use for 2–3 days after place-
ment for AKI to reduce the theoretical risk of leak. Several guidelines
support waiting some period of time from placement to first use, but
data supporting these guidelines are mostly derived from adults with
CKD. The IPSAD 2010 peritoneal access guidelines recommend not
disturbing the PD catheter for 5–10 days post-insertion [12], and the
SAGES guidelines also recommend “adequate time” after surgery for
Table 3
Significant results for receiver operating characteristic analysis.
Outcome Variable Cut-off
Return to OR Days PD catheter in place
Return to OR Days placement to 1st use
Peritonitis PRISM III-24 score
Leak Weight (kg)
Please cite this article as: Stewart CL, et al, Factors associated with peritoneal dialysis catheter complications in children, J Pediatr Surg (2015),
http://dx.doi.org/10.1016/j.jpedsurg.2015.10.035
3
healing, stating a standard of two weeks, but that “a more urgent start
should be considered when the benefits outweigh the risks” [13].
Thirty-three children in our study used their catheter the day of place-
ment because the risks of delayed dialysis outweighed the risks of
leak. Also of note, we found that nearly 10% of catheters placed for AKI
in our study were never used, most often because children recovered
their renal function and ultimately did not require dialysis. We expect
that permitting immediate catheter use would decrease the rate of non-
use, since our colleagues in nephrology and the pediatric intensive care
unit would no longer have to anticipate dialysis needs days in advance.
If the request for PD catheter placement was made the day of or day
prior to planned dialysis, this practice could ultimately result in fewer
unnecessary PD catheters being placed, may not substantially increase
the leak rate, and may decrease unplanned returns to the operating
room. Further larger scale studies are necessary to determine if the ben-
efits of allowing immediate use outweigh the risks.
We also did not find an association with complications and the pres-
ence or absence of omentectomy. Recent studies have reported that fail-
ure rates were higher in children who did not have an omentectomy at
the time of catheter placement using unadjusted analysis [4,5,7]. Phan
et al. [14] reported fewer complications in children who had an
omentectomy in a multivariate analysis, but interestingly did not find
fewer problems specifically with drainage. This study also reported
omental plugging in 3 patients who had an incomplete omentectomy.
In our series, one patient (b1%) required additional omentectomy dur-
ing a PD catheter revision for poor drainage, and two additional children
(1%), who did not have an omentectomy when the PD catheter was ini-
tially placed, had an omentectomy during an unplanned return to the
operating room for revision. The surgeons in our study frequently stated
their reasons for omitting an omentectomy in their operative report,
most commonly because the omentum was high enough in the abdom-
inal cavity that it was felt not to be a risk for catheter blockage. Thus, we
conclude that omentectomy may prevent catheter occlusion in some
patients, but that the surgeon's judgment is usually adequate to deter-
mine which patients will require it, making it acceptable to perform
omentectomy in select patients.
This study has several limitations, most inherent to its retrospective
nature and limited sample size. We also combined data from children
with AKI and CKD for our analysis of complications to increase our sam-
ple size. There may be differences that we did not account for between
children with AKI and CKD that contributed to complications.
3.5. Conclusions
PD catheter placement can be performed in children with relative
safety, but complications are common, particularly in lower weight chil-
dren, more severely ill children, and those with PD catheters for longer
periods of time. Leak is more frequently observed in lower weight chil-
dren. We suspect that decreasing the initial volumes of dialysate in
smaller children may mitigate this risk, but additional research is need-
ed for this to be proven. Nearly 10% of PD catheters placed for AKI are
never used because the children recover prior to use. Immediate use
of PD catheters after placement may not increase leak rate, may reduce
unplanned returns to the operating room, and may reduce frequency of
nonuse for children with AKI. Further study is needed to determine if
the benefits of permitting use on the day of placement outweigh
the risks.
AUC P-value OR (95% CI)
150 0.687 b0.001 5.22 (2.46–11.06)
3 0.600 0.04 2.14 (1.03–4.43)
15 0.731 0.02 6.43 (1.37–30.21)
12.4 0.623 0.02 3.23 (1.21–8.63)
4 C.L. Stewart et al. / Journal of Pediatric Surgery xxx (2015) xxx–xxx

A B

C D

Fig. 1. Receiver operating characteristic area under the curve analysis results.

References
[1] Warady BA. Peritoneal dialysis and the pediatric patient. Perit Dial Int 2012;32:393–4.
[2] Zaritsky J, Warady BA. Peritoneal dialysis in infants and young children. Semin
Nephrol 2011;31:213–24.
[3] Kaiser BA, Polinsky MS, Stover J, et al. Growth of children following the initiation of
dialysis: a comparison of three dialysis modalities. Pediatr Nephrol 1994;8:733–8.
[4] Macchini F, Valadè A, Ardissino G, et al. Chronic peritoneal dialysis in children: catheter
related complications. A single centre experience. Pediatr Surg Int 2006;22:524–8.
[5] Cribbs RK, Greenbaum LA, Heiss KF. Risk factors for early peritoneal dialysis catheter
failure in children. J Pediatr Surg 2010;45:585–9.
[6] Stone ML, LaPar DJ, Barcia JP, et al. Surgical outcomes analysis of pediatric peritoneal
dialysis catheter function in a rural region. J Pediatr Surg 2013;48:1520–7.
[7] Ladd AP, Breckler FD, Novotny NM. Impact of primary omentectomy on longevity of
peritoneal dialysis catheters in children. Am J Surg 2011;201:401–5.
[8] Copeland DR, Blaszak RT, Tolleson JS, et al. Laparoscopic Tenckhoff catheter place-
ment in children using a securing suture in the pelvis: comparison to the open ap-
proach. J Pediatr Surg 2008;43:2256–9.
[9] Schwartz GJ, Muñoz A, Schneider MF, et al. New equations to estimate GFR in chil-
dren with CKD. J Am Soc Nephrol 2009;20:629–37.
[10] Bresolin N, Bianchini AP, Hass CA. Pediatric acute kidney injury assessed by pRIFLE as
a prognostic factor in the intensive care unit. Pediatr Nephrol 2013;28:485–92.
[11] Tan GH, Tan TH, Goh DYT, et al. Risk factors for predicting mortality in a peadiatric
intensive care unit. Ann Acad Med Singapore 1998;27:813–8.
[12] Figueiredo A, Goh BL, Jenkins S, et al. Clinical practice guidelines for peritoneal ac-
cess. Perit Dial Int 2010;30:424–9.
[13] Haggerty S, Roth S, Walsh D, et al. Guidelines for laparoscopic peritoneal dialysis ac-
cess surgery. Surg Endosc 2014;28:3016–45.
[14] Phan J, Stanford S, Zaritsky JJ, et al. Risk factors for morbidity and mortality in pedi-
atric patients with peritoneal dialysis catheters. J Pediatr Surg 2013;48:197–202.

Please cite this article as: Stewart CL, et al, Factors associated with peritoneal dialysis catheter complications in children, J Pediatr Surg (2015),
http://dx.doi.org/10.1016/j.jpedsurg.2015.10.035