www.nature.

com/pr

CLINICAL RESEARCH ARTICLE OPEN

Pediatric central venous access devices: practice, performance,

and costs
✉
Amanda J. Ullman1,2,3 , Victoria Gibson2,3, Mari D. Takashima3, Tricia M. Kleidon1,2,3, Jessica Schults2,3, Masnoon Saiyed3,4,
Paula Cattanach , Rebecca Paterson3, Marie Cooke3, Claire M. Rickard1,3, Joshua Byrnes3,4 and Vineet Chopra3,5
2,3

© The Author(s) 2022

BACKGROUND: Healthcare delivery is reliant on a functional central venous access device (CVAD), but the knowledge surrounding
the burden of pediatric CVAD-associated harm is limited.
METHODS: A prospective cohort study at a tertiary-referral pediatric hospital in Australia. Children <18 years undergoing insertion
of a CVAD were screened from the operating theatre and intensive care unit records, then assessed bi-weekly for up to 3 months.
Outcomes were CVAD failure and complications, and associated healthcare costs (cost of complications).
RESULTS: 163 patients with 200 CVADs were recruited and followed for 6993 catheter days, with peripherally inserted central
catheters most common (n = 119; 60%). CVAD failure occurred in 20% of devices (n = 30; 95% CI: 15–26), at an incidence rate (IR) of
5.72 per 1000 catheter days (95% CI: 4.09–7.78). CVAD complications were evident in 43% of all CVADs (n = 86; 95% CI: 36–50), at a
1234567890();,:

rate of 12.29 per 1000 catheter days (95% CI: 9.84–15.16). CVAD failure costs were A$826 per episode, and A$165,372 per 1000
CVADs. Comparisons between current and recommended practice revealed inconsistent use of ultrasound guidance for insertion,
sub-optimal tip-positioning, and appropriate device selection.
CONCLUSIONS: CVAD complications and failures represent substantial burdens to children and healthcare. Future efforts need to
focus on the inconsistent use of best practices.
Pediatric Research (2022) 92:1381–1390; https://doi.org/10.1038/s41390-022-01977-1

IMPACT:
● Current surveillance of central venous access device (CVAD) performance is likely under-estimating actual burden on pediatric
patients and the healthcare system.
● CVAD failure due to complication was evident in 20% of CVADs. Costs associated with CVAD complications average at $2327
(AUD, 2020) per episode.
● Further investment in key diverse practice areas, including new CVAD types, CVAD pathology-based occlusion and dislodgment
strategies, the appropriate use of device types, and tip-positioning technologies, will likely lead to extensive benefit.

INTRODUCTION such as catheter breakage, dislodgement, and occlusion, are rarely

The insertion of a central venous access device (CVAD) signals the collected or compared. This potentially underestimates the true
commencement, or re-commencement, of life-changing treat- phenomenon of CVAD-associated harm, and the associated
ment for children and their families. Often a child’s first significant burden of CVAD-associated harm for children, their families, and
healthcare procedure, CVADs are a tool of the trade for most healthcare. It also impairs the ability for clinicians and researchers
pediatric health disciplines—used for treatments varying from the to effectively benchmark or target impactful and sustainable
administration of antibiotics for chronic osteomyelitis, to lifelong improvements. Our knowledge surrounding the burden of
parenteral nutrition for gut enteropathies. But a child’s healthcare pediatric CVAD-associated harm is incomplete.6
experience is often disrupted by complications caused by how Given the lack of these standardized metrics, the full economic
healthcare systems and clinicians select, insert, manage and costs of CVAD-associated harm in pediatrics are also unclear.
remove CVADs.1,2 Estimates of attributable costs, including length of hospital stay
CVAD performance can be a measure of hospital performance. and catheter-associated bloodstream infections (CABSI) are
However, conventionally only single outcomes or populations are significant, and have been described for children with cancer
benchmarked—most commonly infections or thromboses, in (additional 21.2 hospital days [95% confidence interval CI:
cancer or intensive care.3–5 Other forms of CVAD complications, 10.4–32.0]; $69,332 [2012 USD; 95% CI: 35,144–103,521]),4 in the

1
School of Nursing, Midwifery and Social Work, University of Queensland, Brisbane, QLD, Australia. 2Queensland Children’s Hospital, Brisbane, QLD, Australia. 3Menzies Health
Institute Queensland and School of Nursing and Midwifery, Griffith University, Brisbane, QLD, Australia. 4Centre for Applied Health Economics, Griffith University, Brisbane, QLD,
Australia. 5Department of Medicine, University of Colorado at Denver, Anschutz Medical Campus, Denver, CO, USA. ✉email: a.ullman@uq.edu.au

Received: 10 May 2021 Revised: 10 January 2022 Accepted: 23 January 2022

Published online: 8 February 2022
 A.J. Ullman et al.
1382

496 CVADs identified as potentially eligible

496 CVADs eligible

74 Inserted after hours

222 Maximum recruitment already reached

163 Patients recruited with 200 CVADs

0 Excluded from analysis due to missing data

200 Provided baseline and at least 1 follow-

up assessment

109 completed 14-day follow-up

67 completed 30-day follow-up

45 completed 90-day follow-up

Fig. 1 Flow diagram of study participation. CVADs Central venous access devices.

intensive care (additional 19 hospital days [95% CI: 14.3–23.8]; encompassing CVAD insertion and management procedures were
$55,646 [2011 USD; 95% CI: 38,785–72,507]),7 and in general prospectively collected, and participants were followed for up to three
pediatrics (additional 21 hospital days [95% CI: 7.3–34.8]; €13,727 months to report CVAD performance (including complications and
[2017 Euro; 95% CI: 5758–21,695]).8 But the economic costs of removals) and associated costs. Ethical approvals were obtained from
other forms of CVAD harm, that also result in additional the Children’s Health Queensland and Griffith University Human Research
Ethics Committee (HREC/18/QRCH/19; 2018/096). The study is reported in
procedures, hospital admissions, and treatment disruption are accordance with the Strengthening The Reporting of OBservational studies
relatively unknown but likely substantial. For example, we have in Epidemiology (STROBE) guidelines.14
previously conservatively costed repeated complications in one
child to have additional healthcare-costs of more than $10,000
(2016 AUD).1 Setting
The study included all clinical areas within the Queensland Children’s
The current high rate of complications and cost may stem from Hospital (QCH), Australia. QCH is Queensland’s tertiary-referral pediatric
non-adherence to best practice guidelines. However, children facility, with 359 beds it provides care to patients from maternity hospital
requiring CVADs are diverse and it is difficult to ensure guidelines discharge to 18 years of age, across all major disciplines (including
are pertinent across all pediatric cohorts. But there are recom- specialized cardiac services). The QCH does not admit neonates for
mendations for care that are supported by high-quality evidence, immediate post-birth management (e.g., prematurity, birth trauma).
that should be routinely implemented. For example, ultrasound
guidance for CVAD insertion;9–11 CVAD tip placement in the cavo- Participants and sample size
atrial junction;12 not using CVADs for a short duration, non- All children less than 18 years, undergoing insertion of a CVAD (including
vesicant infusates;13 and avoiding totally implanted devices for peripherally inserted central catheters [PICCs], non-tunneled CVADs,
neonates and infants.13 Examining whether these practices have tunneled (with or without a Dacron cuff) CVADs, hemodialysis catheters
been universally implemented within healthcare services serves [HDs], totally implanted venous devices [a.k.a. ports/TIVD]) at the QCH in
two complementary purposes: identifying the need for knowledge operating theatres and intensive care within the study period were eligible
translation, and identifying situations where non-routine practice for inclusion. Children having sub-specialty devices inserted, such as
is appropriate, and innovation is required. intrathoracic lines, ECMO cannula, and direct hemodialysis methods (i.e.,
fistulas), were not included.
To solve a complex problem, we first need to unravel its layers. Due to local resources and to ensure quality follow-up (minimizing
In this study, we aimed to describe CVAD insertion practices, missing data), a maximum of only 10 participants could be followed at a
performance, and healthcare-costs across a large pediatric health time. Whenever a patient finished follow-up (at study end), a new patient
service. We also aimed to identify risk for increased pediatric was consecutively commenced. To ensure the sample consecutively
CVAD-associated harm, and explore gaps between current recruited was representative of the pediatric CVAD population, a stratified
practice and best practice. This will provide a comprehensive sampling approach was incorporated to ensure representation across
explanation of the contemporary practice, performance, and CVAD types and primary diagnoses, based on a similar local historical
value, towards prioritized, tangible improvements. cohort15 (widened to include CVADs inserted in the intensive care).
Our 2015 meta-analysis established 25% of international pediatric
CVADs failed prior to completion of therapy (95% CI: 20.9–29.2) at a rate of
1.97 per 1000 catheter days (95% CI: 1.7–2.2).2 Accordingly, our targeted
METHODS sample size was 200 CVADs, to enable accurate benchmarking of
Study design complications (5% absolute precision, 90% confidence, to establish
A prospective cohort study was undertaken at a tertiary-referral pediatric predicted 25% failure),16 while facilitating exploratory model development
hospital in Australia, between September 2018 and March 2020. Data for CVAD failure risk.

Pediatric Research (2022) 92:1381 – 1390

 A.J. Ullman et al.
1383
Table 1. Participant and device characteristics at insertion (n = 200).
Variable PICC TIVD Non-tunnel Non- Tunnel Tunnel HD Tunnel cuff Total
tunnel HD non-cuff
CVAD 119 (60) 21 (10) 18 (9) 5 (3) 7 (4) 5 (3) 25 (13) 200
Catheter days 2680 1659 151 40 276 292 1895 6993
Patient characteristics
Age
Median [years] (IQR) 5 (2–10) 3 (3–10) 0.5 (0.1–5) 0.9 (0.5–5) 0.3 (0.2–3) 3 (3–16) 4 (1–8) 4 (0.1–10)
Neonates (0–30 days) 0 0 1 (6) 0 0 0 1 (4) 2 (1)
Infants (31 to <1 year) 15 (13) 1 (5) 10 (56) 3 (60) 5 (71) 0 5 (20) 39 (20)
Children (1 to <12 years) 80 (67) 18 (86) 7 (39) 1 (20) 2 (29) 3 (60) 15 (60) 126 (63)
Adolescent 24 (20) 2 (10) 0 1 (20) 0 2 (40) 4 (16) 33 (17)
(12–18 years)
Primary diagnosisa
Respiratory (non-CF) 52 (44) 3 (14) 0 0 1 (14) 0 0 56 (28)
Oncology and 17 (14) 17 (81) 1 (6) 3 (60) 0 1 (20) 16 (64) 55 (28)
hematology
Cystic fibrosis 30 (25) 3 (14) 0 0 0 0 0 33 (17)
General surgical 12 (10) 6 (29) 4 (22) 0 0 1 (20) 5 (20) 28 (14)
Gastroenterology 8 (7) 0 4 (22) 2 (40) 2 (29) 3 (60) 6 (24) 25 (13)
Coronary care/cardiac 8 (7) 0 4 (22) 0 1 (14) 1 (20) 1 (4) 15 (8)
Hepatic 3 (3) 0 5 (28) 2 (40) 2 (29) 0 1 (4) 13 (7)
Other 16 (13) 1 (5) 4 (22) 0 0 5 (100) 5 (20) 31 (16)
Previous CVADs (no.)
1 14 (27) 8 (53) 3 (33) 0 0 0 9 (60) 34 (32)
2–3 15 (29) 4 (27) 4 (44) 2 (50) 1 (20) 1 (20) 2 (14) 29 (27)
4–5 7 (14) 2 (13) 1 (11) 0 1 (20) 1 (20) 2 (14) 14 (14)
>5 16 (31) 1 (7) 1 (11) 2 (50) 3 (60) 3 (60) 2 (13) 28 (27)
Known occluded vesselsa
Basilic 15 (87) 2 (10) 2 (11) 0 3 (43) 0 3 (12) 25 (13)
Brachial 4 (4) 1 (5) 0 0 1 (14) 0 1 (4) 7 (4)
Cephalic 2 (2) 1 (5) 0 0 0 0 1 (4) 4 (2)
Axillary 2 (2) 1 (5) 1 (6) 0 2 (29) 0 2 (8) 8 (4)
Internal jugular 1 (1) 0 1 (6) 2 (40) 1 (14) 1 (20) 1 (4) 7 (4)
Other (e.g., subclavian) 4 (4) 0 2 (12) 2 (40) 1 (14) 1 (20) 1 (4) 11 (7)
Device characteristics
Vessel
Basilic 85 (71)b 0 0 0 0 0b 0 85 (43)
Brachial 18 (15) 0 0 0 0 0 0 18 (9)
Cephalic 5 (4) 0 0 0 0 0 0 5 (3)
Axillary 8 (7) 0 0 0 0 0 0 8 (4)
Internal jugular – 16 (76) 11 (61) 2 (40) 7 (100) 4 (80) 20 (80) 60 (30)
Subclavian – 5 (24) 2 (11) 0 0 0 5 (20) 12 (6)
Femoral 2 (2) – 5 (28) 3 (60) 0 0 0 10 (5)
Lumen numbers
One 103 (87) 21 (100) 1 (6) 0 (0) 2 (29) 0 (0) 4 (16) 131 (66)
Two 16 (13) 0 (0) 2 (11) 5 (100) 5 (71) 5 (100) 20 (80) 53 (27)
Three 0 (0) 0 (0) 15 (83) 0 (0) 0 (0) 0 (0) 1 (4) 16 (8)
Clinician characteristics
Multiple insertion
attempts
2 14 (12) 1 (5) 0c 0 0 0 4 (16)b 19 (10)
≥3 5 (4) 0 0 0 1 (14) 0 1 (4) 7 (4)
Guidance
Ultrasound 118 (99) 11 (52) 15 (83)b 4 (80) 7 (100) 2 (40) 11 (44) 166 (83)

Pediatric Research (2022) 92:1381 – 1390

 A.J. Ullman et al.
1384
Table 1. continued
Variable PICC TIVD Non-tunnel Non- Tunnel Tunnel HD Tunnel cuff Total
tunnel HD non-cuff
X-ray for confirmation 114 (96) 21 (100) 8 (44) 2 (40) 7 (100) 5 (100) 22 (88) 179 (90)
X-ray during placement 5 (4) 4 (19) 0 0 0 1 (20) 8 (32) 18 (9)
Cut down 0 0 1 (6) 0 0 0 4 (16) 6 (3)
Tip position
Cavo-atrial junction 104 (87) 5 (24)d 1 (6)c 1 (20)c 5 (71) 2 (40) 7 (28)b 125 (63)
Superior vena cava 12 (10) 8 (38) 5 (29) 0 2 (29) 1 (20) 13 (52) 41 (21)
Right atrium 0 6 (29) 5 (29) 1 (20) 0 2 (40) 4 (16) 18 (9)
Non-central 1 (1) 0 1 (6) 0 0 0 0 2 (1)
Inferior vena cava 2 (2) 0 2 (12) 0 0 0 0 4 (2)
Procedural time
Median [min] (IQR) 62 (46–82) 77 (58–89) 162 (78–298) 40 (40–57) 92 (74–119) 129 (71–149) 80 (66–137) 70 (53–98)
Dwell
Median [day] (IQR) 14 (11–23) >90 (>90) 7 (1–13) 3 (1–13) 26 (9–77) 90 (8–>90) 90 (71–>90) 16 (11–63)
In situ at study end 6 (5) 17 (81) 0 0 1 (14) 3 (60) 18 (72) 45 (23)
(3 months)
CVAD central venous access device, HD hemodialysis, IQR Interquartile range, PICC peripherally inserted central catheter, TIVD totally implanted venous device.
a
Multiple response options.
Missing data: b1 c3 d2.

Measures and covariates distribution was prospectively matched across device types and primary
Patient (e.g., age, primary diagnosis, comorbidities), CVAD (e.g., type, diagnosis with historical local CVAD databases (widened to include the
gauge, tip location, lumens), and clinician (e.g., specialty, number of intensive care setting).15 Information bias was decreased by having data
attempts) characteristics are descriptively reported, and were selected due collected by dedicated, experienced clinical research nurses (including
to previous association with reduced CVAD performance in pediatric or established inter-rater reliability of data collection processes),25 clear and
adult cohorts.15,17,18 rigorous outcome definitions (including CVAD performance outcome
CVAD performance was described by a report of CVAD failure (CVAD assignment by infectious disease physicians or radiologists, when
complications that result in permanent cessation of CVAD function, prior to appropriate), and prospective methods (eliminating recall bias).27
completion of therapy)2 and CVAD complications as a composite and
individual report of insertion and post-insertion complications including
those which are transient: infectious (CABSI,3 local site infection3) Healthcare cost estimation
thrombotic (venous thrombosis,19,20 breakage,21 occlusion,22) mechanical The cost for CVAD failure, CVAD complication, infectious complications,
thrombotic complications, dislodgement, insertion-related complication,
(dislodgement,20 migration20) and severe skin complications.23 All
and severe skin complications were estimated from a health care system
complications were defined in accordance with international contempor-
ary literature, with a full description available via online content perspective. The primary cost outcome is the expected cost per 1000
(Supplementary Material 1). CVADs calculated as the product of the incidence rate for 1000 CVADs and
Clinical practice variation was compared to practices with strong the estimated cost per episode (CVAD failure, CVAD complication,
evidence. Specifically positive practices were ultrasound guidance for infectious complications, thrombotic complications, dislodgement,
insertion-related complication, and severe skin complications). The cost
CVAD insertion9 (by proceduralist; documented in the medical record); tip
per episode includes the cost of devices (including replacement devices),
placement not outside of the cavo-atrial junction or right atrium (assessed
via imaging);12 and appropriate device selection13 (including no insertion dressings and securement materials, imaging guidance (X-ray or ultra-
of implanted devices during active infection [i.e., positive blood cultures],24 sound), inpatient hospital stay, operating room, pathological tests,
PICCs for short-term [<7 days] peripherally-compatible infusates,13 totally medications, and time of nursing and medical staff. Utilization of each
implanted devices for infants and neonates [<1 year].13) resource, including self-reported estimates of time taken to treat was
based on data collected within the study and entered using the REDCap
database. The cost per episode was estimated based on resource
Data sources utilization multiplied by unit prices (Supplementary Material 3). Prices for
Operating theatre lists and ICU admission records (i.e., where CVADs were salary and other items were based on prices faced by local public hospitals
inserted) were screened Monday-to-Friday by specialized research nurses collected as part of this study from the site hospital and from previously
to identify eligible patients. The patients were assessed bi-weekly until the published estimates of pediatric hospital-specific prices for materials
CVAD was removed or for 3 months. The assessments were carried out by associated with device insertion where such information was previously
the research nurses either in person (while still admitted to QCH) or over known.28,29 The cost of CABSI and venous thromboembolism (VTE)
the phone (while discharged or at an alternative site; a process we have complications were based on national average cost estimates of these
previously used reliably),25 with additional data sourced from the patient’s complications.30 All costs are reported in 2020 Australian dollars (A$).
electronic medical record. All data (including healthcare utilization) were
collected using a dedicated, secure, web-based REDCap (Research
Electronic Data CAPture, Vanderbilt) database.26 Statistical analysis
Data collected were thoroughly cleaned and checked for accuracy (10% by
second research nurse) prior to importing into Stata (version 13; StataCorp,
Bias College Station, TX) for statistical analysis. The CVAD was the unit of
Selection bias was minimized by participants being selected via an measurement, with some children having multiple CVADs within the
equitable inclusion criterion with sequential recruitment (based on cohort. Descriptive statistics for normally distributed (mean and standard
operating theatre lists and ICU admission records), however, some deviation) and non-normally distributed (median and interquartile range
participants were missed due to staff availability (i.e., weekends, after- [IQR]) are reported for clinical characteristics of included patients. The
hours) and maximum recruitment being reached, but participant sampling proportion and associated 95% confidences intervals (95% CI), cost per

Pediatric Research (2022) 92:1381 – 1390

 Table 2. CVAD performance (N = 200; 6993 catheter days) and projected economic cost.
Proportion of 95% CI Episodes (N) Incidence rate per 95% CI Cost per 95% CI Cost per
CVADs (%) 1000 catheter days episode 1000 CVADs
CVAD failure 20 15–26 40 5.72 4.09–7.78 A$826 A$762–A$890 A$165,372
Tunneled, non-cuffed (n = 7) 43 9–82 3 10.87 2.25–31.44 A$824 A$591–A$1060 A$354,320
Tunneled HD (n = 5) 40 5–85 2 6.85 0.83–24.52 A$889 $581–A$1200 A$355,600
Non-tunneled (n = 18) 39 17–64 7 46.35 18.83–93.18 A$569 A$464–A$674 A$221,910
Non-tunneled HD (n = 5) 20 0–72 1 24.68 0.63–131.58 A$569 A$291–A$847 A$113,800
Totally implanted (n = 21) 19 5–42 4 2.41 0.66–6.16 A$1464 A$1110–A$1820 A$278,160
Tunneled, cuffed (n = 25) 16 5–36 4 2.11 0.58–5.40 A$824 A$622–A$1030 A$131,840

Pediatric Research (2022) 92:1381 – 1390

PICC (n = 119) 16 10–24 19 7.09 4.27–11.05 A$649 A$576–A$722 A$103,840
CVAD complicationa 43 36–50 86 12.29 9.84–15.16 A$2327 A$2200–A$2450 A$1,000,610
Infectious complications
CABSI 5 2–9 10 1.43 0.69–2.63 A$14,943b A$12,600–A$17,300 $749,150
Local infection 0 0 0 0 0 $0 $0 $0
Thrombotic complications
Occlusion 25 19–31 49 7.00 5.19–9.25 A$239 A$223–A$256 A$59,750
Breakage 4 2–8 8 1.14 0.49–2.25 A$381 A$315–A$447 A$15,240
Venous thrombosis 3 1–6 5 0.72 0.23–1.67 A$7045b A$5510–A$8590 A$211, 350
Dislodgement
Partial dislodgement 16 11–21 31 4.43 3.01–6.29 A$43 A$39–A$47 A$30,400
Complete dislodgement 2 1–5 4 0.57 0.16–1.46 A$1469 A$1110–A$1830 A$29,380
Insertion-related complication 1 0–3 2 0.20 0.00–1.03 A$09 A$5–$13 A$90
Severe skin complication 8 4–12 15 2.15 1.20–3.53 A$1533 A$1340–A$1730 A$122,640
a
One device can have more than one complication type; expected cost per episode adjusted to include only one replacement device in instances of multiple complications.
b
A.J. Ullman et al.

Source: National Hospital Cost Data Collection. Public Hospitals Cost Report, Cost Weights, for Ar-Drg Version 8.0 Round 22 Sydney: IHPA (2020).
1385
 A.J. Ullman et al.
1386
1.00 CVADs, non-tunneled CVADs had the highest incidence rate per
1000 catheter days (46.35; 95% CI: 18.83–93.18). As displayed in
Fig. 2, failure was common earlier in the device dwell for non-
0.75
tunneled and permanent HD (i.e., <20 catheter days), while failure
of other devices was frequently later (i.e., ≥20 catheter days).
0.50 CVAD complications were evident in 43% of CVADs (n = 86;
95% CI: 36–50), at a rate of 12.29 per 1000 catheter days (95% CI:
0.25
9.84–15.16). The most common complications during CVAD dwell
were occlusion (25%; n = 49 [95% CI: 19–31]; IR 7.00 [95% CI:
5.19–9.25]) and partial dislodgement (16%; n = 31 [95% CI: 11–21];
0.00 IR 4.43 [95% CI: 3.01–6.29]). While insertion-related complications
0 20 40 60 80 were uncommon, several significant skin complications (e.g.,
Dwell time (in days) surgical wound dehiscence, allergic dermatitis) were evident
PICC (n = 119) Totally implanted (n = 21)
(8%; n = 15 [95% CI: 4–12]; IR 2.15 [95% CI: 1.20–3.53])
Non-tunneled (n = 18) Non-tunneled HD (n = 5)
The expected costs associated with CVAD failure were A$826
Tunneled, non-cuffed (N = 7) Tunneled HD (n = 5) per episode, and A$ 165,372 per 1000 CVADs. The highest CVAD
Tunneled, cuffed (n = 25) failure costs per episode was A$1464 for totally implanted devices,
however, per 1000 CVADs costs were highest with tunneled HD (A
Fig. 2 Kaplan–Meier survival estimates of CVAD failure per device $354,320) and non-tunneled CVADs (A$355,600). This reflects the
type. HD Hemodialysis, PICC Peripherally inserted central catheter. higher incidence of failure associated with these procedures. Total
CVAD complication costs per 1000 CVADs was A$1,000,610. Per
episode, costs were greatest for CABSI at A$14,943, and per 1000
episode, and cost per 1000 CVADs, are reported for CVAD-associated CVADs (A$749,150). Further significant episode level costs were
complications and serious adverse events, and variations in care. Missing associated with venous thrombosis (A$7045), complete dislodge-
data were not imputed. Kaplan–Meier estimates were used to estimate the ment (A$1469), and severe skin complications (A$1533).
probabilities of CVAD failure and complication. Univariable and multi- The results of the univariable and multivariable regression for
variable analyses of the association between CVAD device types and time risk of CVAD failure are available in Supplementary Material 4.
to first CVAD failure were performed with Cox proportional hazards model Univariable cox-regression results showed a strong association
with shared frailty term set at participant’s level to account for intra-subject (p < 0.05) of CVAD failure with age (HR 0.92; 95% CI: 0.86–0.99) and
correlation (hazard ratios (HR) reported). Covariates were selected based
non-tunneled CVADs (HR 6.82; 95% CI: 2.70–17.20). However, only
on relationship laid out on Directed Acyclic Graphs (DAG) using clinical
knowledge and knowledge acquired from previous studies and consisted non-tunneled CVADs (HR 4.27; 95% CI: 1.49–12.18; referent PICC)
of age, previous device, previous vessel occlusion, and primary diagnosis had a statistically significant association in CVAD failure in the final
(Supplementary Material 2).15,17,18,31 Clinical practice variation compared to multivariable analysis.
practices with strong evidence (was explored descriptively across clinical
characteristics (e.g., primary diagnosis, catheter types)). Variations in care
Variations between recommended and actual care are described
in Table 3. Ultrasound guidance for insertion was not used for 16%
RESULTS (n = 32) CVADs, mainly for tunneled, cuffed (n = 14; 56% of
Participants tunneled, cuffed CVADs inserted) and totally implanted devices
As described in Fig. 1, 496 CVADs were inserted over the study (n = 10; 48% of totally implanted devices inserted). Sub-optimal
period. From this sample, 296 were missed due to research nurse tip-positioning (outside of the cavo-atrial junction or right atrium)
availability. Finally, 163 patients with 200 CVADs were recruited was evident in 22% (n = 43) of the cohort, primarily for PICCs (n =
and followed for 6993 catheter days. 13; 11% of PICC inserted) and tunneled, cuffed (n = 13; 52% of
The most commonly inserted CVADs were PICCs (n = 119; 60%; tunneled, cuffed inserted). Over 10% of PICCs (10.5%; 21% of
2612 catheter days) (Table 1). However, totally implanted venous PICCs inserted) had fewer than seven days of peripherally-
devices and tunneled cuffed CVAD were often in situ at 3-month compatible IV therapy, for children with orthopedic (n = 7),
study end, and thereby contributed considerable catheter days neurological (n = 5), infectious disease (n = 3), ear, nose and
(TIVD: n = 21; 10%; 1659 catheter days; tunneled cuffed: n = 25; throat (n = 3) and burns (n = 2).Only one totally implanted device
13%; 1895 catheter days). Short-term CVADs (e.g., PICCs, non- was inserted for an infant receiving treatment for oncology/
tunneled CVADs) were mainly inserted for children with respira- hematology.
tory conditions and surgical (including cardiac) procedures, while
long-term CVADs (e.g., TIVD, tunneled cuffed) were predominantly
used for children with oncological, hematological, and gastro- DISCUSSION
enterological conditions. Over a quarter of the cohort had CVADs are a vital component of modern healthcare provision. This
received >5 previous CVADs (n = 28; 27%), and many had known is the first study to comprehensively and prospectively describe
occluded vessels limiting CVAD insertion locations. Multiple CVAD insertion practices, performances, and healthcare-costs
insertion attempts were necessary for 14% of insertions (n = 26; across a single large pediatric health service. From these data,
14%) mainly for PICCs and tunneled cuffed CVADs, however, we have a better understanding of the current insertion,
image guidance was not consistently used for the insertion of performance, and cost burden of CVADs in pediatric healthcare.
TIVD, tunneled cuffed and HD catheters, and catheter tip We also identified key opportunities to improve CVAD perfor-
placement outside of the cavo-atrial junction was evident in 43 mance, both locally and internationally.
CVADs (22%). Overall, CVAD failure prior to completion of therapy in this
cohort is similar to previous international estimates,2 however this
CVAD performance early evaluation of new CVAD types (i.e., tunneled, non-cuffed
As reported in Table 2, CVAD failure was 20% (n = 30; 95% CI: CVADs) demonstrate challenges (n = 7; 43% failure; 11.1 per 1000
15–26), at an IR of 5.72 per 1000 catheter days (95% CI: 4.09–7.78). catheter days). This may be because these new devices are being
While failure proportion was highest in tunneled, non-cuffed (43%; implemented as a “rescue” device, when other, traditional CVAD
n = 4), tunneled HD (40%; n = 2), and non-tunneled (39%; n = 7) routes are no longer available or additional lumens are necessary

Pediatric Research (2022) 92:1381 – 1390

 A.J. Ullman et al.
1387
Table 3. Variations in care (n = 200).
Variation Device characteristics N (%)a Patient characteristics N (%)b
No ultrasound guidance for insertion 32 (16)
Device Tunneled, cuffed 14 (56) Oncology/hematology 9 (64.3)
General surgical 4 (28.6)
Gastroenterology 4 (28.6)
Orthopedic 1 (7.1)
Other 4 (28.6)
Totally implanted 10 (48) Oncology/hematology 9 (64.3)
General surgical 3 (30.0)
Cystic fibrosis 1 (10.0)
Respiratory (non-CF) 1 (10.0)
Non-tunneled 3 (17)
Permanent HD 3 (60)
PICC 1 (<1)
Temporary HD 1 (20)
Sub-optimal tip-positioning 43 (22)
Device PICC 13 (30) Other respiratory 6 (46)
Oncology/hematology 2 (15)
Other 2 (15)
Gastroenterology 2 (15)
Cystic fibrosis 1 (6)
Hepatic 1 (6)
General surgical 1 (6)
Tunneled cuff 13 (30) Oncology/hematology 10 (77)
General surgical 3 (23)
Gastroenterology 2 (15)
Other 2 (15)
Totally implanted 8 (19)
Non-tunneled 6 (14)
Tunneled, non-cuffed 2 (4)
Permanent HD 1 (2)
Devices inserted with current infection 27 (14)
Device PICC 83 (70)
Non-tunneled 9 (50)
Totally implanted 3 (14)
Permanent HD 2 (40)
Tunneled, non-cuffed 1 (14)
Tunneled, cuffed 4 (16)
Temporary HD 0 (0)
PICCs < 7 day therapy with peripherally-compatible 21 (11) Orthopedic 7 (33)
infusates (n = 119) Neuro 5 (24)
Respiratory (non-CF) 3 (14)
Other 17 (81)
Neonates and infants with totally implanted devices 1 (<1) Oncology/hematology 1 (100)
CF cystic fibrosis, HD hemodialysis, PICC peripherally inserted central catheter.
a
Denominator of proportion (%) is the device type sample.
b
Denominator of proportion (%) is the case device type sample.

for small vasculature, and further data regarding their perfor- 1000 catheter days; 27.0 per 1000 catheter days; respectively). This
mance are necessary. Non-tunneled, short-term devices, such as should not be an acceptable outcome of clinical practice.
non-tunneled CVADs (inserted in the jugular, subclavian or As evident in other recent descriptions,15,31–33 CABSI rates have
femoral vein) and non-tunneled HD catheters were associated reduced in recent decades, being 5% in our cohort, however other
with the highest rate of failure, per 1000 catheter days (49.3 per CVAD complications, especially occlusion and partial dislodgment,

Pediatric Research (2022) 92:1381 – 1390

 A.J. Ullman et al.
1388
are high (25%, 7.2 per 1000 catheter days; 16%, 4.6 per 1000 to other hospitals, settings, and diagnostic groups. Also, not all
catheter days; respectively), and should become the focal point of children with CVADs were able to be recruited, which may have
innovation and improvement. The prevention of occlusion needs introduced sampling bias related to lack of after-hours recruitment.
to encompass the range of occlusion pathologies; including Additionally, we were not able to recruit neonates immediately after
thrombotic (i.e., insertion technique, tip position, catheter birth (admitted for birth-associated injuries or conditions), since these
materials), infusate precipitation (i.e., flushing, compatibility neonates are cared for at a maternity hospital adjacent to the
mapping) and mechanical (i.e., tip position, securement, quality pediatric hospital. However, the prospective design over multiple
external equipment) causes. Similarly, partial dislodgement data points (including 6787 catheter days), overall transparency, and
innovations need to center on effective securement and tip- rigorous, and internationally benchmarked outcome measures
positioning.31,34,35 All causes of CVAD failure result in treatment strengthen the study’s reliability and validity. Lastly, the cost
disruption—and this has been consistently demonstrated to be estimates were potentially skewed (i.e., low frequency and very high
associated with reduced survival for children with cancer.36–38 cost) in some cases given the lack of sufficient sample sizes.
The financial costs of CVADs in healthcare have previously
centered on CABSI4,7,8 and, less-often, thrombosis.39 As demon-
strated in our study, the financial benefits of reducing the CONCLUSION
incidence of CABSI remain considerable. However, all aspects of This study has provided a comprehensive explanation of CVAD
CVAD failure (including the removal and insertion procedure) and contemporary practice, performance, and value in a tertiary-
complications (primarily treatment; but also need for replacement referral pediatric hospital. We have highlighted several tangible
devices for some complications) are associated with immediate practice areas that should be targeted towards improved clinical
hefty costs. Our results are based on a combination of and health services outcomes, which are likely to have relevance
prospectively collected data and established estimates, including to many pediatric hospitals. In particular, we recommend further
the Australian National Hospital Costing Data Collection.40 These research and clinical practice improvement surrounding the
are likely to be an under-estimation of costs as they fail to integration of new CVAD types (e.g., tunneled, non-cuffed CVADs),
incorporate long-term health consequences, including increased CVAD pathology-based occlusion and dislodgment strategies, the
morbidity and mortality associated with complications. However, appropriate use of PICCs, and the potential for tip-positioning
conservatively for 1000 CVAD procedures, the CVAD complications technologies. Further investment in these key diverse practices
cost is more than A$1,000,000 annually. These costs are not only will likely lead to extensive benefit for health services, both
associated with CABSI and venous thrombosis, and should help financially and clinically.
direct policy makers in the future. Consequently, the prevention of
CVAD failure and complications is a ripe area for investment with
significant potential for high returns in terms of both improved REFERENCES
quality of treatment and reduction in healthcare resources. 1. Ullman, A. J., Kleidon, T., Cooke, M. & Rickard, C. M. Substantial harm associated
Exploring gaps between current practice and best practice, with failure of chronic paediatric central venous access devices. BMJ Case Rep.
highlight many opportunities for correction and invention. The sub- 2017, bcr2016218757 (2017).
2. Ullman, A. J., Marsh, N., Mihala, G., Cooke, M. & Rickard, C. M. Complications of central
optimal catheter tip placement described in the study (22%; n = 43)
venous access devices: a systematic review. Pediatrics 136, e1331–e1344 (2015).
was occurred more frequently than desired, and is also likely to be an 3. US Government. Centers for Disease Control and Prevention 1–9 (US Government,
underestimate of true events, as it relied on accurate interpretation of 2020).
radiological images (i.e., chest X-ray, fluoroscopy). These are relatively 4. Wilson, M. Z., Rafferty, C., Deeter, D., Comito, M. A. & Hollenbeak, C. S. Attributable
subjective, and easily influenced by image acquisition and respiratory costs of central line-associated bloodstream infections in a pediatric hematology/
movement.41,42 Sub-optimal tip placement is a key risk factor in the oncology population. Am. J. Infect. Control 42, 1157–1160 (2014).
development of many catheter complications; including thrombosis, 5. Noailly Charny, P. A. et al. Increased risk of thrombosis associated with periph-
non-thrombotic occlusion, arrythmia, and cardiac erosions. Technol- erally inserted central catheters compared with conventional central venous
ogies, such as intracavitary electrocardiography-based PICC tip catheters in children with leukemia. J. Pediatr. 198, 46–52 (2018).
6. Schults, J. et al. International recommendations for a vascular access minimum
confirmation, show potential to reduce insertion, and post-insertion-
data set: a Delphi Consensus Building Study. BMJ Qual Saf. 30, 722–730 (2020).
related complications, in addition to other procedural improvements 7. Goudie, A., Dynan, L., Brady, P. W. & Rettiganti, M. Attributable cost and length of
(e.g., faster placement, reduced cost).42,43 stay for central line-associated bloodstream infections. Pediatrics 133,
PICCs have been highlighted as a device at risk of over- or e1525–e1532 (2014).
inappropriate use, resulting in preventable harm.44,45 Within this 8. Karagiannidou, S., Zaoutis, T., Maniadakis, N., Papaevangelou, V. & Kourlaba, G.
cohort we identified PICCs as having the lowest proportion of Attributable length of stay and cost for pediatric and neonatal central line-
device failure (16%; 95% CI: 10–24), but a moderate incidence rate associated bloodstream infections in Greece. J. Infect. Public Health 12, 372–379
of failure (12.7 per 1000 catheter days). Inappropriate use was (2019).
evident, with 21 PICCs (10.5% of PICC cohort) inserted for a 9. de Souza, T. H., Brandão, M. B., Nadal, J. A. H. & Nogueira, R. J. N. Ultrasound
guidance for pediatric central venous catheterization: a meta-analysis. Pediatrics
peripherally-compatible therapy administration of fewer than
142, e20181719 (2018).
7 days, potentially placing these patients at preventable and 10. Gorski, L. A. et al. Infusion therapy standards of practice, 8th edition. J. Infus. Nurs.
increased risk for thrombosis and infection.13 The indications for 44, S1–S224 (2021).
safe and appropriate use of PICCs and other CVADs were recently 11. Bodenham Chair, A. et al. Association of anaesthetists of great Britain and Ireland:
defined in the Michigan Appropriateness Guideline for Intrave- safe vascular access 2016. Anaesthesia 71, 573–585 (2016).
nous Catheters in pediatrics (miniMAGIC), and should be used as a 12. Fletcher, S. J. & Bodenham, A. R. Safe placement of central venous catheters:
baseline standard of care.13 Totally implanted devices inserted in where should the tip of the catheter lie? Br. J. Anaesth. 85, 188–191 (2000).
neonates and infants are generally inappropriate, due to the risk 13. Ullman, A. et al. The Michigan appropriateness guide for intravenous catheters in
of wound dehiscence due to lack of adipose tissue.13,46 The pediatrics: minimagic. Pediatrics 145, S269–S284 (2020).
14. von Elm, E. et al. The strengthening the reporting of observational studies in
appropriate use of all vascular access devices can play a vital role
epidemiology (strobe) statement: guidelines for reporting observational studies.
in vessel health preservation, and complication prevention, for Ann. Intern Med. 147, 573–577 (2007).
chronic, acute, and critically ill children.13 15. Kleidon, T. M. et al. Development of a paediatric central venous access device
database: a retrospective cohort study of practice evolution and risk factors for
Limitations device failure. J. Paediatr. Child Health 56, 289–297 (2020).
This study has limitations, primarily related to it being based in a 16. Dhand, N. K. & Khatkar, M. S. Statulator: An Online Statistical Calculator. Sample
single, metropolitan, tertiary pediatric facility, limiting generalizability Size Calculator for Estimating a Single Proportion (2014).

Pediatric Research (2022) 92:1381 – 1390

 A.J. Ullman et al.
1389
17. Chopra, V. et al. Picc-associated bloodstream infections: prevalence, patterns, and 44. Chopra, V. et al. The Michigan appropriateness guide for intravenous catheters
predictors. Am. J. Med 127, 319–328 (2014). (magic): results from a multispecialty panel using the rand/ucla appropriateness
18. Chopra, V. et al. Peripherally inserted central catheter-related deep vein thrombosis: method. Ann. Intern. Med. 163, S1–S40 (2015).
contemporary patterns and predictors. J. Thromb. Haemost. 12, 847–854 (2014). 45. Paje, D. et al. Patterns and predictors of short-term peripherally inserted central
19. Lobo, B. L., Vaidean, G., Broyles, J., Reaves, A. B. & Shorr, R. I. Risk of venous catheter use: a multicenter prospective cohort study. J. Hosp. Med. 13, 76–82 (2018).
thromboembolism in hospitalized patients with peripherally inserted central 46. Paterson, R. et al. Pediatric vascular access appropriateness: a systematic review.
catheters. J. Hosp. Med. 4, 417–422 (2009). Pediatrics 145, s243–s268 (2020).
20. Yamamoto, A. et al. Sutureless securement device reduces complications of periph-
erally inserted central venous catheters. J. Vasc. Inter. Radio. 13, 77–81 (2002).
21. Matsuzaki, A. et al. Long-term use of peripherally inserted central venous catheters for
cancer chemotherapy in children. Support Care Cancer 14, 153–160 (2006). ACKNOWLEDGEMENTS
22. Goossens, G. A. et al. Diagnostic accuracy of the catheter injection and aspiration We thank the children, families, and clinicians who participated in this project.
(Cinas) classification for Assessing the function of totally implantable venous
access devices. Support Care Cancer 24, 755–761 (2016).
23. Broadhurst, D., Moureau, N. & Ullman, A. J. Management of central venous access
device-associated skin impairment: an evidence-based algorithm. J. Wound Ost- AUTHOR CONTRIBUTIONS
A.J.U. conceptualized and designed the study, drafted the initial manuscript, and
omy Cont. Nurs. 44, 211–220 (2017).
reviewed and revised the manuscript. V.G., J.S., R.P., T.M.K., and P.C. designed the data
24. Baskin, K. M. et al. Evidence-based strategies and recommendations for preservation
collection instruments, collected data, and reviewed and revised the manuscript. M.D.
of central venous access in children. J. Parenter. Enter. Nutr. 43, 591–614 (2019).
T. and M.S. carried out the initial analyses and reviewed and revised the manuscript.
25. Webster, J. et al. Insertion site assessment of peripherally inserted central
C.M.R., M.C., J.B., and V.C. designed the study, and critically reviewed the manuscript
catheters: Inter-observer agreement between nurses and inpatients. J. Vasc.
Access 19, 370–374 (2018). for important intellectual content. All authors approved the final manuscript as
26. Harris, P. A. et al. Research electronic data capture (redcap)−a metadata-driven submitted and agree to be accountable for all aspects of the work.
methodology and workflow process for providing translational research infor-
matics support. J. Biomed. Inf. 42, 377–381 (2009).
27. Hammer, G. P., du Prel, J.-B. & Blettner, M. Avoiding bias in observational studies:
FUNDING
part 8 in a series of articles on evaluation of scientific publications. Dtsch. Arztebl.
This project was independently funded by the Australian Government’s National
Int. 106, 664–668 (2009).
Health and Medical Research Council. Open Access funding enabled and organized
28. Kleidon, T. M. et al. Comparison of midline catheters and peripherally inserted
by CAUL and its Member Institutions.
central catheters to reduce the need for general anesthesia in children with
respiratory disease: a feasibility randomized controlled trial. Paediatr. Anaesth. 31,
985–995 (2021).
29. Tuffaha, H. W. et al. Cost of vascular access devices in public hospitals in COMPETING INTERESTS
Queensland. Aust. Health Rev. 43, 511–515 (2018). Amanda Ullman reports investigator-initiated research grants and speaker fees
30. National Hospital Cost Data Collection. Public Hospitals Cost Report, Cost Weights, provided to Griffith University (her previous employer) from vascular access product
for Ar-Drg Version 8.0 Round 22 (IHPA, Sydney, 2020). manufacturers (3M, Cardinal Health, and Becton Dickinson), unrelated to this project.
31. Ullman, A. J. et al. Innovation in central venous access device security: a pilot Tricia Kleidon declares investigator-initiated research grants and speaker fees
randomized controlled trial in pediatric critical care. Pediatr. Crit. Care Med. 20, provided to Griffith University from vascular access product manufacturers (3M,
e480–e488 (2019). BD-Bard, Baxter, Medical Specialties Australia, Smiths Medical, Vygon), unrelated to
32. Quirt, J. et al. Reduction of central line associated bloodstream infections and line this project. Jessica Schults declares investigator-initiated research and educational
occlusions in pediatric intestinal failure patients on long-term parenteral nutrition grants provided to Griffith University by vascular access product manufacturers
using an alternative locking solution, 4% tetrasodium ethylenediamine tetra- (Baxter, Becton Dickinson, Entrotech Life Sciences), unrelated to this project. Marie
acetic acid (Edta). J. Parenter. Enteral. Nutr. 45, 1286–1292 (2020). Cooke declares investigator-initiated research and educational grants and speaker
33. Baier, C. et al. Incidence, risk factors and healthcare costs of central line- fees provided to Griffith University (her employer) by vascular access product
associated nosocomial bloodstream infections in hematologic and oncologic manufacturers (Baxter, Becton Dickinson, Entrotech Life Sciences), unrelated to this
patients. PLoS ONE 15, e0227772 (2020). project. Claire Rickard reports investigator-initiated research grants and speaker fees
34. Kleidon, T. M. et al. A pilot randomized controlled trial of novel dressing and provided to Griffith University from vascular access product manufacturers (3M,
securement techniques in 101 pediatric patients. J. Vasc. Inter. Radio. 28, Angiodynamics; Baxter; BBraun; BD-Bard; Medtronic; ResQDevices; Smiths Medical),
1548–1556.e1541 (2017). unrelated to this project. The remaining authors declare no competing interests.
35. Ullman, A. et al. Innovative dressing and securement of tunneled central venous
access devices in pediatrics: a pilot randomized controlled trial. BMC Cancer 17,
595 (2017).
36. Forbrigger, Z. & Kulkarni, K. Use of tissue plasminogen activator as a surrogate CONSENT TO PARTICIPATE
measure for central venous catheter dysfunction and survival outcome in chil- Informed consent was obtained from the health service.
dren with cancer: a population-based retrospective cohort study. J. Pediatr.
Hematol. Oncol. 37, 554–560 (2020).
37. Athale, U. H., Siciliano, S., Cheng, J., Thabane, L. & Chan, A. K. Central venous line
dysfunction is an independent predictor of poor survival in children with cancer. CONSENT FOR PUBLICATION
J. Pediatr. Hematol. Oncol. 34, 188–193 (2012). Informed consent was obtained from the health service.
38. Deitcher, S. R., Gajjar, A., Kun, L. & Heideman, R. L. Clinically evident venous throm-
boembolic events in children with brain tumors. J. Pediatr. 145, 848–850 (2004).
39. Goudie, A. et al. Costs of venous thromboembolism, catheter-associated urinary ADDITIONAL INFORMATION
tract infection, and pressure ulcer. Pediatrics 136, 432–439 (2015).
Supplementary information The online version contains supplementary material
40. Independent Hospital Pricing Authority. National Hospital Cost Data Collection.
available at https://doi.org/10.1038/s41390-022-01977-1.
https://www.ihpa.gov.au/what-we-do/nhcdc (2014).
41. Gnannt, R. et al. Variables decreasing tip movement of peripherally inserted
Correspondence and requests for materials should be addressed to Amanda J.
central catheters in pediatric patients. Pediatr. Radio. 46, 1532–1538 (2016).
Ullman.
42. Kleidon, T. M. et al. Peripherally inserted central catheter thrombosis after
placement via electrocardiography vs traditional methods. Am. J. Med. 134,
Reprints and permission information is available at http://www.nature.com/
e79–e88 (2020).
reprints
43. Yuan, L. et al. Superior success rate of intracavitary electrocardiogram guidance
for peripherally inserted central catheter placement in patients with cancer: a
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims
randomized open-label controlled multicenter study. PLoS ONE 12, e0171630
in published maps and institutional affiliations.
(2017).

Pediatric Research (2022) 92:1381 – 1390

 A.J. Ullman et al.
1390
Open Access This article is licensed under a Creative Commons article’s Creative Commons license and your intended use is not permitted by statutory
Attribution 4.0 International License, which permits use, sharing, regulation or exceeds the permitted use, you will need to obtain permission directly
adaptation, distribution and reproduction in any medium or format, as long as you give from the copyright holder. To view a copy of this license, visit http://creativecommons.
appropriate credit to the original author(s) and the source, provide a link to the Creative org/licenses/by/4.0/.
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the © The Author(s) 2022

Pediatric Research (2022) 92:1381 – 1390