Selection and Insertion of Vascular Access Devices in Pediatrics: A Systematic Review

A systematic review was undertaken to synthesize existing evidence on selection and insertion of pediatric VADs following the RAND-UCLA Appropriateness Method.⁹  The systematic review protocol was registered and published with the International Prospective Register of Systematic Reviews (PROSPERO; CRD201994286)¹⁰  and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standards.¹¹ 

We conducted searches of the US National Library of Medicine (Medline), Cumulative Index to Nursing and Allied Health, Cochrane Library databases, Embase, and international clinical trial databases for all studies in which authors reported VAD use (success and complications) in a pediatric population from 2008 to May 16, 2018. Search terms were developed in collaboration with an experienced medical librarian. We used exploded Medical Subject Headings (MeSH) terms (eg, catheters) and relevant keywords and their variants (eg, child, pediatrics). Table 1 reveals the electronic database search strategy.

Following the RAND-UCLA Appropriateness Method, our goal was to provide a critical review of the literature summarizing the scientific evidence available surrounding the appropriateness of pediatric VAD selection, insertion, and characteristics.⁹  This meant a range of study designs was eligible for inclusion, including existing CPGs, systematic reviews, randomized control trials (RCTs), quasi RCTs, before-after trials, cohorts, or case-control studies. Additionally, the guidelines and studies must have been published in a peer-reviewed journal and authors must have reported on complications and/or risk and reliability of VADs in patients aged term to 18 years in a pediatric hospital. We defined VADs to include intraosseous devices; midline catheters; peripherally inserted central catheters (PICCs); short and long peripheral intravenous catheters (PIVCs); tunneled, tunneled-cuffed, and nontunneled central venous access devices (CVADs); totally implantable venous devices; and umbilical catheters. We excluded studies that were not published in English and conference abstracts, animal studies, NICU studies, n = 1 studies, case reports, case-series reports, and qualitative reports. Although eligibility criteria were focused on pediatric studies (ie, term to 18 years), we determined that including preterm neonate and adult studies was preferential to no evidence.

The primary outcomes were defined a priori as (1) device and insertion characteristics that impact the success of VAD insertion and (2) device and insertion characteristics associated with VAD failure, due to complications before the completion of therapy, or successful VAD insertion. Device characteristics included VAD type, device catheter-to-vein ratio, and device lumens. Insertion characteristics included insertion site and location and the use of vessel visualization technology. Complications included but were not limited to central line–associated bloodstream infection (CLABSI), VAD-associated thrombosis, occlusion, catheter dislodgement, catheter-tip migration, catheter breakage or rupture, local infection, and phlebitis.

Title and abstract screening was performed independently by 2 review authors (E.B. and A.J.U.), excluding studies that did not meet eligibility criteria when this could be determined by the abstract alone. Full-text articles included for screening were reviewed by 2 review authors (E.B. and A.J.U.) and independently assessed against the inclusion or exclusion criteria. Duplicate publications were excluded. When individual studies that had been evaluated in a systematic review also returned in the search, the primary study was excluded to avoid repetition, and the systematic review was referred to. Any discrepancies between review authors were resolved through mutual discussion and, when required, a third, independent review author (M.C.) was consulted.

All full-text articles that met inclusion criteria were independently reviewed by 3 review authors (R.S.P., E.B., and A.J.U.) to extract details on the number of patients and catheters, population, age of participants, VAD type, study method, indication, comparators, and the frequency of VAD failure or complications. These details were summarized in a data extraction sheet and were cross-checked for accuracy and agreement. Additional relevant references were identified by examining reference lists of included studies and guidelines. Hand-searched references were evaluated to ensure that they met inclusion criteria. After screening, pragmatic inclusion of wider studies was employed. That is, if no studies identified meeting the preferential inclusion or exclusion criteria, we included wider studies (eg, a priori area of deficit included neonates outside of the NICU). The extracted data were then combined by using narrative (descriptive) synthesis by categories (ie, outcome, vascular device, indication).

The methodologic quality, transparency, and relevance of all individual included studies were independently assessed by 2 review authors (E.B. and A.J.U.) by using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline¹²  and Critical Appraisal Skills Program (CASP) Cohort Study checklist.¹³  RCTs and systematic reviews were preferentially included as the gold standard level of evidence for evaluating VADs. However, when this level of evidence did not exist, a pragmatic approach was taken so that studies outside of the scope of the review (eg, laboratory, premature neonate, or adult studies) and CPGs (which may be limited by the number or quality of included studies) were incorporated into the review. To provide a synthesis of the available literature for the purpose of the RAND-UCLA Appropriateness Method,⁹  studies were categorized according to their methodology: (1) CPG, (2) systematic review or other review, (3) RCT, (4) observational study with comparator, and (5) other (clinical review, pilot study, laboratory study).

The results of the search strategy and study selection are summarized in Fig 1. Electronic database searches identified 7430 articles, and hand searches of the bibliography of included studies and clinical guidelines identified 30 additional articles for potential inclusion. After removal of duplicates and screening for eligibility, a total of 133 studies and CPGs met eligibility criteria and were included in data extraction.

The review includes 27 CPGs (20.4%), 11 systematic reviews (8.3%), 10 RCTs (7.6%), 79 observational studies (59.8%), 3 pilot studies (2.3%), 2 clinical reviews (1.5%), and 1 laboratory study (0.8%). These were sourced from research teams based in Africa (1.6%), Asia (9.3%), Europe (31.0%), the Middle East 2.3%), North America (40.3%), Oceania (11.6%), and South America (3.9%), with the majority sourced from the United States (n = 34; 37.8%). Subjects within the included studies ranged in age, from premature neonates up to 66 years, and required treatment of oncologic or hematologic conditions, support during critical care admission, or vascular access for hemodialysis, postsurgical, or general infusion therapy or parenteral nutrition (PN). Due to the heterogeneity across age and conditions in pediatric patients, included articles were divided into specific clinical subspecialties (eg, hospitalized pediatric patients, hematology or oncology; see Fig 2). Results from individual studies based on specific hospitalized populations are presented in Table 2 and include key characteristics for each study.

A total of 3 CPGs and 26 studies were included for general hospitalized pediatric patients, including 1 systematic review, 1 RCT, and 23 observational studies. These studies compared all VADs in hospitalized adult (n = 3); child (n = 13), infant (n = 2), and neonatal (n = 3); and surgical (n = 1) populations. With the exception of 1 CPG, no studies reported on midlines in the pediatric population, so adult studies were also included.^14,43,94 

Overall, umbilical catheters were associated with high rates of complications, including catheter-related bloodstream infection, occlusion, dislodgement, thrombosis, and local infection or phlebitis, with device failure being common.^{6,15,30,54,57,125,127}  Umbilical catheter malposition and catheter-tip migration during treatment also frequently occurred,^6,142  with concerns that malpositioned or dislodged umbilical catheters may lead to severe hemorrhage and even death if not detected or rectified in a timely manner.⁵⁸  Despite this, the literature reviewed indicated that umbilical catheter placement is common practice in neonates up to the first 7 days of life.²³  Evidence for umbilical catheter dwell time was scarce and supported placement for short durations only, due to of increased risk of device failure.^24,57  Consequently, guidance from available CPGs was limited but recommended placement only for as long as clinically necessary, or ≤14 days, if managed aseptically.¹³⁶  Beyond 15 days, ambiguity in the literature regarding the patency of the umbilical vascular system and catheter for longer durations was evident.^136,142  One CPG did support replacing any umbilical catheter with a PICC for central access >7 days to reduce risk of infection.¹³⁶  Alternative VADs for neonates include PICCs and CVADs^6,34,66,136 ; however, it was recognized that neonates frequently had poor or difficult-to-access venous assets,^45,55  were more likely to have higher risk of insertion-related complications,⁵⁵  and were at high risk of blockage across all devices.⁶ 

PICCs are commonly the first VAD choice for infants.¹³⁶  Numerous studies found that PICCs in infants were associated with a lower risk of complications, particularly thrombosis,^28,122  leading current CPGs to recommend their use in children <1 year and for therapies for longer durations (>7 days).^56,122,136  Conversely, PIVCs had significantly higher rates of dysfunction in infants (50%) compared with children >1 year and for longer dwell times,¹²⁵  with one CPG only recommending PIVCs for therapies <6 days.¹³⁶  None of the included evidence evaluated the complications associated with midline catheters in infants; however, the National Association of Neonatal Nurses recommended them as an appropriate alternative for peripherally compatible intravenous (IV) therapy for <6 to 10 days in infants.¹³⁶ 

Compared to PICCs, tunneled-cuffed CVADs were associated with higher rates of thrombosis (although thrombosis tended to occur after longer durations in situ), with 60% of tunneled-cuffed CVADs developing deep vein thrombosis (DVT).⁶⁷  Tunneled-cuffed CVADs were also associated with higher rates of insertion failure in infants and young children.⁵⁵  Risk of infection in CVADs placed in infants was high, especially compared with toddlers, and was linked to the use of totally implantable venous devices.⁵⁹  As tunneled CVADs and totally implantable venous devices restrict the future use and availability of accessed veins and insertion was related to a higher risk of complication, studies and CPGs indicated that their use in infants should be limited.^97,136 

For hospitalized children and adolescents, PIVCs and midline catheters were reported as appropriate for short-term peripherally compatible therapies in multiple studies because of their low risk of catheter-related infections and thrombosis in comparison to PICCs.^{14,43,94,100,127,143}  However, in one study, the authors suggested that these devices are associated with increased risk of occlusion with extended dwell times.¹²⁷  PICCs were reported to have high insertion success rates⁸⁴  and low failure rates in children and adolescents,^6,59  although one study reported severe complications in 40.1% of failed PICCs, associated with increasing patient age.²⁸  PICC insertion was commonly associated with short-term complications such as occlusion, CLABSI, and thrombosis,^6,30  with rates of these events increasing with longer durations of therapy.³⁰ 

In comparison, tunneled-cuffed CVADs had high rates of infection (4.8%–19.9%)⁶  and were associated with short-term complications, such as infection, malfunction, leak, and malposition,^59,134  and high rates of occlusion (12.1%).⁶  Although totally implantable venous devices had low rates of infection (0.01–0.28 per 1000 catheter days),⁶  they were associated with early complications, such as bleeding, pneumothorax, nerve lesions, catheter misplacement, occlusion, and skin damage,^97,103  and long-term complications, such as infection, thrombosis, catheter fracture or disconnection, secondary dislocation, and skin breakdown over port septum.^6,56,103  Overall, tunneled-cuffed CVADs and totally implantable venous devices had high insertion success rates^55,59,87  and low failure rates,^1,6  and can be used as alternatives to PICCs in patients requiring frequent vascular access.⁵⁶  Given their high insertion success rate,⁸⁷  nontunneled CVADs can be placed after the failed placement of other CVADs.⁵⁶ 

For patients undergoing treatment of malignant hematologic and oncological conditions, infection and thrombotic complications, as well as disruption to treatment, are important considerations for VAD selection.^25,119  A total of 17 studies, including 1 systematic review and 16 observational studies, and 5 CPGs were included. Most studies were focused on tunneled-cuffed CVADs and totally implantable venous devices in mixed hematology and oncology (n = 5), PN (n = 1), and leukemia (n = 3) populations. Other studies compared PICCs, CVADs, and totally implantable venous devices in mixed hematology and oncology populations (n = 5).

Overall, tunneled-cuffed CVADs and totally implantable venous devices had low rates of insertion-related complications, device failure, and malfunction^{40,65,79,80,119,128}  and were associated with low rates of thrombosis.^2,112,132  Reporting on rates of infection between devices was variable. Two studies found that rates of infection were not statistically different between tunneled-cuffed CVADs and totally implantable venous devices.^79,93  In contrast, other studies found higher rates of CLABSI in totally implantable venous devices compared to CVADs¹¹⁹  and higher risk of overall infection in totally implantable venous devices compared to tunneled-cuffed CVADs.¹²⁸  Additional studies found that overall infection rates were higher in tunneled-cuffed CVADs compared to totally implantable venous devices.^20,40,72,106  Although occlusion was common in this population,^{46,65,112,119}  rates were not different between tunneled-cuffed CVAD and totally implantable venous devices.^65,133  Compared to totally implantable venous devices and tunneled-cuffed CVADs, PICCs had higher rates of device-related complications as well as overall infection and CLABSI, thrombosis, and occlusion.^2,40,112,138 

International guidelines recommend the use of tunneled-cuffed CVADs for pediatric patients undergoing hematologic and oncological treatment requiring frequent and continuous vascular access, particularly for frequent blood sampling, PN, and complex IV therapies.^50,56  Totally implantable venous devices may also be appropriate across these indications,⁵¹  especially in patients ≥10 kg.⁴⁵  These devices were only recommended for intermittent use, however, due to the increased risk of infection and thrombosis.^45,50  One CPG endorsed the use of PICCs for short- to medium-term treatments,⁵⁰  although there was insufficient evidence to support their routine use,^117,120  and chemotherapy treatment via peripheral venous access was not recommended.^50,120  No included studies reported on nontunneled CVADs; however, one international guideline supported the placement for intrahospital use or short durations only.⁵⁰ 

Unlike other patient groups, for critically ill infants, children, and adolescents, choice of VAD may be prioritized by whether the patient is stable or unstable.^21,22  Seven observational studies, 1 systematic review, and 1 RCT were included for critical care patients and included the emergency department (ED) (n = 2), emergency surgery (n = 1), moderate-severe burns (n = 1), out-of-hospital critical care (n = 1), and PICU (n = 4) populations. No studies explored midline devices in the critical care population, although one study compared short and long PIVCs.¹⁰¹  Overall, the focus of the studies in this population was on infection and thrombosis and not on outcomes such as dwell time, occlusion, infiltration, extravasation, or bleeding complications.

Regarding stable, critically ill patients requiring short-term acute therapy, 2 CPGs supported the use of nontunneled CVADs for any type of infusion therapy up to 7 to 10 days despite risk of infection and thrombosis.^34,56  Evidence to support the use of PICCs in this population was mixed. Whereas the Infusion Therapy Standards of Practice cautioned against the use of PICCs in critically ill patients (adults and pediatrics) because of the risk of infection or thrombosis,⁶⁶  2 other clinical guidelines recommended PICCs for both short-^34,76  and long-term³⁴  durations in critically ill patients. This variability may be due to the scarcity of high-quality evidence directly comparing these devices. PIVCs were frequently reported within this population^101,137  and were recommended over intraosseous devices in nonemergent situations.^21,22  However, compared to PICCs, PIVCs were commonly associated with obstruction, leakage, and dislodgment,¹³⁷  with short PIVCs in particular linked to higher rates of local infections, dislocation, infiltration, occlusion, and thrombosis compared to long PIVCs.¹⁰¹  There were limited studies reporting on nontunneled CVAD use in critically ill children; however, authors of one study reported children >13 years old with a nontunneled CVAD were at higher risk of DVT compared with children <1 year.⁶⁰  Tunneled-cuffed CVADs and totally implantable venous devices were associated with high rates of infection and thrombosis but were reported as suitable for long-term therapies in this population.^{34,73,111,115}  No studies or CPGs reported on the suitability of midline devices in this population specifically, although they were more broadly considered appropriate for therapies lasting 1 to 4 weeks.⁷⁶ 

In unstable, critically ill patients, speed of access was prioritized. International guidelines recommend venous access via PIVC placement unless attempts to cannulate take >60 seconds or ≥2 attempts.^21,22  When IV access was difficult, intraosseous devices were consistently reported as fast alternatives for children and adolescents,^{21,44,66,70,76,131}  and umbilical catheter was recommended for neonates in the first week of age.²² 

For patients with congenital cardiac conditions, VAD selection prioritizes vessel preservation to ensure key vasculature can be used for future life-saving procedures.⁵⁶  Overall, 3 CPGs and 4 observational studies of pediatric and young adult patients were included in the systematic review and encompassed patients with univentricular and biventricular physiology and patients in cardiac surgery subgroups. Comparisons between umbilical catheters, PICCs, tunneled CVADs, and nontunneled CVADs based on dwell time, occlusion, thrombosis, and infection risk were the primary focus of the included studies.^15,31,35,88 

Generally, short-term peripheral and nonperipheral compatible therapy delivered via umbilical catheters in infants can be used.⁶⁶  PICCs were recommended for patients with congenital cardiac conditions requiring ≥7 days of IV therapy.¹³⁶  because of lower complication rates, PICCs inserted in lower extremity vessels were recommended.¹³⁶  CVAD (tunneled and nontunneled) use was reported in this population; however, placement ≥7 days was associated with an increased risk of CLABSI.^31,56  None of the included studies or guidelines included evidence to support the use of totally implantable venous devices in this population.

Regarding specific use in neonates with univentricular physiology, umbilical catheters had low rates of CLABSI, thrombosis, and occlusion when compared to other devices,¹⁵  and they can be used when preventing significant vessel loss for future procedures is a prioirity. Although PICCs were generally recommended in congenital cardiac populations, clinical guidelines emphasized the need for specialist consultation for patients with cardiac malformations.¹³⁶  Femoral nontunneled CVADs were reported as appropriate for therapies lasting <14 days in single-ventricle populations, with ≥14-day durations associated with a higher risk of thrombosis but not occlusion.¹⁵  Comparatively, other femoral devices (tunneled, uncuffed) were not related to an increased risk of thrombosis or occlusion for therapies lasting ≥14 days.¹⁵ 

Long-term vascular access dependency encompassed nonmalignant hematologic, respiratory, gastrointestinal, metabolic, and immunologic conditions requiring long-term (>2 months) and very long-term (>1 year) VAD placement.²⁹  In total, 4 studies and 9 CPGs described this diverse patient population across PN and non-PN therapies and continuous or intermittent therapies. Despite the heterogeneity of reported populations, vessel preservation and complication prevention were the common themes when comparing PICC, short- and long-catheter PIVC, midline, tunneled-cuffed CVAD, and totally implantable venous device indications.^16,32,90,109 

CLABSI, occlusion, and thrombosis were common complications of long-term PN therapy across all VADs.^16,32  Peripheral devices (PIVCs and midlines) are unsafe for delivering PN because of the risk of venous damage^50,62,66 ; however, they may be indicated for limited time periods in hospitalized patients with restricted dextrose and protein concentrations (<10% and/or 5%, respectively).^66,76,107  One single-center observational study supported PICCs for long-term PN,³²  and another CPG recommended PICCs for home PN¹⁰⁷ ; although totally implantable venous devices were generally preferred when PN was required for long durations, despite being associated with an increased risk of infection.^{16,29,32,50,56,92} 

Evidence for the use of VADs in patients requiring long-term non-PN infusates was similar. Specifically, except for in one CPG, all peripheral devices (PIVCs and midlines) were regarded as inappropriate for long-term–dependent non-PN therapies, especially in patients with chronic renal failure.^{56,62,66,76,109}  Tunneled CVADs and totally implantable venous devices were instead preferred for non-PN therapies required for continuous or intermittent and extended durations, respectively.^{56,62,66,76,103,109} 

For pediatric patients with difficult venous access, insertion success rates and the number of attempts are priority considerations for VAD selection. Factors that affect these include patient physiology, pathology, damage caused by VAD, and the procedural skill of the clinician.¹⁴⁴  Five observational studies were included for patients with difficult venous access in ED (n = 2), elective surgery (n = 1), PICU (n = 1), and hospitalized children (n = 1) populations. The majority of studies compared nontunneled CVADs (n = 3) and evaluated the insertion success rates and complications associated with the number of insertion attempts. Two studies evaluated PIVCs with and without ultrasound-guided (USG) techniques in pediatric populations.^123,130 

In patients with difficult vascular access, USG PIVC insertion was associated with higher overall and first- and second-attempt success rates after failed insertion via the landmark method.¹³⁰  Up to 17.2% of CVADs in an ED setting experienced a complication (arterial puncture, hematoma, pneumothorax, and arrhythmia), and ≥3 attempts were significantly related to complication development.¹⁴⁰  In a study comparing insertion success in neonates and nonneonates, nontunneled CVAD failure was common (63.1%–76.3%), and the success rate after >1 attempts was significantly lower with decreasing patient age and associated with greater insertion-related complications.⁵⁵  Similarly, nontunneled CVAD placement in PICU patients led to high early mechanical complications (17.5%) and was associated with more insertion attempts and insertion using the subclavian or jugular approach.¹¹³ 

Determining the appropriate catheter-to-vein ratio is a difficult balancing act of optimizing patient and therapy needs while ensuring risk of catheter-related complications, such as occlusion, is minimized. Although catheter-to-vein ratio was recognized as important for preventing phlebitis, occlusion, and thrombosis,⁵⁶  few studies reported on catheter-to-vein ratios in neonate and pediatric patients. Only one observational study reported on PICCs in hospitalized children⁹¹ ; therefore the included studies were broadened to incorporate an adult and laboratory study.^95,118  An additional 6 CPGs were referred to for guidance.^{34,50,56,66,82,136} 

Clinical guidelines emphasized selecting the smallest practical sized gauge or French (F) that met treatment and patient need.^{34,56,66,95,136}  Inadequate catheter-to-vein ratio was linked to an increased risk of thrombosis in 2 studies, especially in infants with congenital heart disease, and smaller catheter size was associated with higher rates of occlusion.^50,56  In contrast, larger catheters had a corresponding increase in rates of phlebitis.⁶⁶  For any VAD (PICC, PIVC, midline, CVAD, or totally implantable venous device) a catheter-to-vein ratio of <50% in pediatric patients and <33% in neonates was generally recommended.^56,66,136  Specific to PICCs, thrombosis risk increased with a catheter-to-vein ratio ≥0.33⁹¹  and ≥0.45,¹¹⁸  in hospitalized children and adults, respectively. Depending on the patient vessel size, 22- to 24-gauge peripheral catheters were considered appropriate for both pediatric patients and neonates.^66,76  According to the Peripherally Inserted Central Catheters: Guideline for Practice from the National Association of Neonatal Nurses, 1.1F to 3F catheters (20–28 gauge) were commonly used catheter sizes for neonates.¹³⁶  There were no specific recommendations or empirical evidence for the size of CVADs (tunneled or nontunneled), midline catheters, or totally implantable venous devices.

Choosing the optimum number of device lumens to deliver planned therapy while reducing the risk of catheter-associated complications, particularly infection, occlusion, and thrombosis, is both complex and important.³⁴  Nine studies described lumen number outcomes in hospitalized adult (n = 2), children (n = 2) and infant (n = 1), immune-competent (n = 1), and hematology and oncology (n = 3) populations. An additional 11 CPG recommendations were included in the review. All compared single- versus double- or multilumen catheters in CVADs (n = 3), tunneled-cuffed CVADs (n = 2), and PICCs (n = 4).

Universally, CPGs recommended the minimum number of lumens necessary for therapy provision. Because of increased risk of infection, occlusion, and thrombosis, multilumen VADs are only appropriate when indicated (ie, for hematopoietic stem cell transplant, critically ill patients, and patients requiring concurrent infusion of noncompatible infusates such as blood and blood product, PN, or chemotherapy).^* A dedicated lumen for PN was also frequently endorsed,^34,50,86,107  but authors of 2 CPGs stated there was insufficient evidence to support this recommendation.^56,96  Similarly, a dedicated lumen for blood sampling, through the largest lumen, was recommended by some, but with limited evidence.^66,76 

Evidence evaluating multilumen PICC outcomes was mixed. One study found DVT risk increased with ≥2 lumens,⁶⁷  another study reported a nonsignificant reduction in CLABSI risk with single-lumen use,³⁹  while a number of other studies reported that PICC lumen number was not significantly associated with the risk of thrombosis^81,122  or infection.⁸¹  Overall, most studies found occlusion was the most common complication associated with catheter lumen number, and ≥2 lumen PICCs were associated with the highest risk of occlusion.^28,39,81  Similarly, studies reported mixed outcomes on the basis of CVAD lumen number. Two studies reported no association between single- and double-lumen catheters and rates of infection⁶⁵  and major CVAD-related complications,²⁶  whereas others reported higher rates of CLABSI,^46,106  exit-site or tunnel infection,⁴⁶  malfunction or occlusion,⁴⁶  and complications requiring repositioning²⁶  for double-lumen CVADs.

Optimal catheter site selection in pediatric patients is more complex than in adults as pediatric patients typically have fewer accessible veins due to their smaller size.^56,124  Determining the appropriate insertion site is important for minimizing risk of insertion and preventing post-insertion related complications^{19,56,75,86,124} ; however, only partial evidence for appropriate vessels and insertion locations for VADs was available. A total of 21 studies and 12 CPGs were included describing insertion location outcomes. Overall, most studies (n = 15) compared CVADs in cardiac, neurology, and general surgical (n = 5), PICU (n = 4), hospitalized infant (n = 1), oncology (n = 1), and stem cell collection (n = 1) populations. Five studies compared PIVC device location in surgical (n = 3), hospitalized children (n = 1), and neonate and infant (n = 1) populations, and 3 studies evaluated PICC device location in oncology (n = 1), hospitalized children (n = 1), and immunocompetent (n = 1) populations.

According to the guidelines reviewed, common insertion sites varied by device type. PIVC insertion was frequently recommended in the hand and upper extremities,^66,76  with the scalp and foot suggested as alternative insertion sites for infants and toddlers.⁶⁶  The Infusion Therapy Standards of Practice⁶⁶  and Provisional Infusion Therapy Standards of Practice⁷⁶  guidelines recommended avoidance of areas of flexion including the wrist. Insertion of PICCs via the basilic vein was the preferred insertion site, although the brachial, cephalic, axillary, temporal, and posterior auricular veins were acceptable alternatives.^{56,62,66,76,107}  For neonates, the best available vein was recommended without specific guidance as to what constituted the best vein.⁶⁶  Infants could also have PICCs inserted at saphenous and popliteal veins.^56,66,76  Similarly, for midline catheter insertion, the basilic, brachial, and cephalic veins were suggested for neonates, infants, and pediatric patients,^76,96  as well as alternative insertion sites such as the scalp and leg.^66,76  For PIVCs, the external jugular vein was recommended only in emergency settings or if no other vein was available.⁶⁶  Overall, there were no preferred insertion sites for tunneled and nontunneled CVADs or totally implantable vascular devices in neonates, infants, and pediatric patients.^{66,76,82,117,120}  No studies evaluated the insertion locations for tunneled and nontunneled CVADs in children; however, in adults, tunneled and nontunneled CVADs were commonly inserted into the internal and external jugular, subclavian, or femoral vein, although the subclavian vein was the recommended insertion site.^56,62,76,107 

PIVCs had higher success rates when inserted in the cephalic vein in the proximal forearm under USG techniques or the antecubital fossa^109,123  and a longer life span when inserted into the arm compared to scalp, hand, or leg insertion sites.³³  CVAD insertion via the axillary vein using USG techniques resulted in fewer insertion attempts and significantly shorter time to guide-wire insertion and time to cannulation.⁷⁸  Similarly, CVAD insertion in the subclavian vein was associated with shorter median puncture time, less insertion attempts, and significantly less guide-wire misplacement compared to insertion via the infraclavicular approach.⁴¹  In one study, there was higher overall insertion success for insertion via the subclavian compared to the internal jugular vein when using the landmark technique.⁴²  In a sample of critically ill newborns and children (0–14 years), USG brachiocephalic insertion had significantly higher first-attempt success, fewer insertion attempts, and a shorter procedure time compared to the internal jugular vein.⁹⁸  In critically ill neonates and infants, image-guided placement of tunneled CVADs via saphenous or femoral veins using a surgical cutdown was associated with high placement success.⁶³ 

In pediatric patients, PIVCs inserted at the bend of the arm or lower extremities were associated with increased risk of infiltration, erythema, pain, inability to administer medications, no or poor flow due to gravity, and kinked catheter,^125,127  whereas insertion at the foot, ankle, or scalp was significantly associated with increased risk of occlusion.^125,127  In neonates and infants, PICC insertion at foot or ankle sites was significantly associated with an increased risk of phlebitis, thrombosis, and dysfunction.^47,125,136  Overall, there was no direct association between PICC insertion at brachial, cephalic, or saphenous insertion sites and thrombosis in pediatric patients¹²² ; however, left-sided PICC insertion was associated with higher rates of PICC-related complications.³⁷  The majority of studies, with 2 exceptions,^18,78  found that rates of infection and complication varied significantly on the basis of CVAD insertion sites.^† Generally, complications were similar between internal jugular and subclavian veins.⁴²  One study, however, reported higher rates of early mechanical complications in nontunneled CVADs inserted through the subclavian compared to the jugular or femoral vein.¹¹³ 

Across CVAD types, insertion via the internal jugular in infants and pediatric patients was associated with increased risk of high arterial puncture,^42,113  postoperative chylothorax,³⁶  thrombosis,^67,69,89  and infection.^42,69  Insertion through the subclavian vein was correlated with an increased risk of high arterial puncture (left-side approaches),^42,113  arrhythmias and misplacement (right-side approaches),¹¹³  malposition and occlusion,⁴²  postoperative chylothorax,³⁶  and sepsis.³¹  In pediatric oncology patients, subclavian insertion sites had higher rates of thrombosis compared to external jugular and cephalic sites, but there was no difference in rates between right- or left-side insertion sites.¹³²  Comparatively, brachiocephalic insertion sites were associated with significantly lower CLABSI and thrombosis compared to jugular and subclavian approaches.⁶⁹  Overall, CVAD insertion through the femoral vein was linked to higher risk of thrombosis^51,67,89  but lower complication and infection rates in infants <5 kg.⁸⁵ 

Vessel and catheter-tip visualization technologies, including ultrasound guidance, transillumination, near-infrared (NIR) light device guidance, fluoroscopy, and electrocardiogram (ECG), are commonly used in pediatric clinical practice.^34,82  Despite the variety of available technologies, all vessel visualization technologies aim to minimize complications and increase success rates during cannulation. Overall, 31 studies describing vessel visualization outcomes for PIVC, midline, PICC, CVAD, and totally implantable venous devices and 15 CPGs were included. Among those, 17 studies were focused on VADs in critically ill children, neonate, and infant (n = 4); hospitalized children (n = 3); cardiac, neurologic, and general surgical (n = 3); mixed cardiac surgical, congenital heart disease, and PICU (n = 1); ED (n = 1); and neonate (n = 1) populations. Across these heterogeneous populations, the focus was on increasing first-attempt success rates and overall successful insertion and correct catheter-tip positioning.

Across all CPGs, USG insertion by trained clinicians was recommended for all pediatric populations, device types, and insertion sites.^‡ It was indicated in most of the evidence reviewed that USG insertions were associated with high insertion success, first-attempt success rates, lower procedure time, fewer attempts, and fewer complications in PIVCs and CVADs.^§ Only 2 systematic reviews reported no difference in CVAD insertion success rates between ultrasound guidance and landmark techniques.^121,135  PIVC insertion was frequently improved by use of vessel visualization devices,^101,123,130  with fewer complications noted.¹⁰¹  Few studies compared visualization techniques for totally implantable venous device placement. One study reported USG percutaneous puncture of totally implantable venous devices had similar success rates, procedure times, and complication rates when compared to surgical cutdown methods.⁴⁸  Another study found that insertion of totally implantable venous devices by using ultrasound guidance was significantly more effective in reducing complication rates, had shorter procedure times, and was more cost-efficient compared to open surgical cutdown techniques.⁷¹ 

However, recommendations for other visualization techniques in pediatric populations, including NIR light (vessel visualization) and ECG (catheter-tip confirmation) device use, were scant. Use of NIR light devices may be efficacious in selected high-risk subpopulations¹⁰²  and may modestly improve first-attempt success rates^77,104,110 ; however, the current evidence does not support an overall benefit⁷⁴  or a benefit that is better than non–image-guided methods.¹⁰⁴  Similarly, low levels of evidence prevent the recommendation of ECG assistance for PICC placement¹⁰⁵ ; however, one study also found higher first-attempt success and correct tip position success by using intracavitary ECG PICC insertion compared to landmark techniques.¹⁴¹  Other techniques, such as ECG techniques, for CVAD placement were reported as successful and accurate.^27,116  In particular, the ECG technique was significantly more accurate and was associated with fewer complications when compared to landmark techniques²⁷  and when intracavitary ECG was undertaken by using a dedicated ECG monitor.¹¹⁶  An article discussed vessel visualization techniques for umbilical catheters and recommended plain radiographs for confirming catheter course and location.¹⁴² 

VAD selection and insertion grounded in evidence-based, standardized decision-making can reduce risk of complications, pain, length of hospital stay, and costs and can improve overall safety and treatment efficacy.^1–6  Despite many individual studies, systematic reviews, and targeted CPGs, there is no evidence-based guide to assist clinical decision-making in common pediatric indications. This systematic review is the first of its kind to evaluate the evidence that informs VAD selection and insertion for common pediatric indications by using rigorous methodology and a wide breadth of scope. Overall, our review synthesized the available high-quality evidence to inform clinical decision-making, while also highlighting practices in need of further inquiry.

This systematic review revealed evidence- and guideline-based recommendations for VAD selection and insertion in numerous pediatric populations.^56,66,76,136  There was a large quantity of evidence to support VAD selection and insertion decision-making in general hospitalized pediatric patients and specialized populations, such as malignant hematologic and oncological and critically ill pediatric patients. Use of most VADs and their associated complications, especially for PICCs, was well evidenced within these population groups. Evidence to support the use of single-lumen devices unless otherwise indicated (eg, for PN) was broadly supported within the literature and in CPGs.^‖ Similarly, evidence to suggest that VAD insertion assisted by USG techniques reduced complications and improved insertion success was repeatedly reported by high-quality studies and recommended across multiple CPGs.^¶ The strength of the evidence for device selection and insertion in these populations therefore facilitates the implementation of quality clinical decision-making.

On the other hand, our review highlighted gaps in evidence, especially in the form of RCTs, for some pediatric populations, devices, and indications. In some pediatric populations, evidence was so sparse that the scope of the review had to be broadened to include adult and laboratory studies. Specific populations, such as neonates, cardiac patients, and patients with difficult venous access, relied on a few observational studies, making recommendations on device selection challenging. Similarly, although there was a significant proportion of quality evidence on VADs in patients dependent on long-term PN, there was limited evidence for other long-term VAD-dependent populations (eg, cystic fibrosis). Universally, there was a dearth of evidence evaluating midline catheters for the majority of pediatric indications. The literature for recommendations regarding optimal catheter-to-vein ratio was also limited; therefore, this review had to be broadened to include adult and laboratory studies. Finally, although there was abundant evidence to support use of USG techniques for the majority of device insertions, there was a shortage of evidence for the use of other technologies, particularly NIR light devices. As such, additional high-quality research evaluating these populations, device types and characteristics, and insertion procedures is warranted.

Although this study undertook a systematic and rigorous review of the available literature, the results should be interpreted with caution and in the context of its limitations. First, we did not undertake formal assessment of the quality of evidence (eg, using the Grading of Recommendations Assessment, Development and Evaluation [GRADE] approach); however, all individual included studies were independently assessed by 2 review authors, and their quality indicated by their study methodology, in accordance with the RAND-UCLA methodology.⁹  Additionally, although this study prioritized the review of RCTs and systematic reviews as the gold standard for evaluating VADs, this level of evidence was rarely available. In accordance with the RAND-UCLA Appropriateness Method, the purpose of this review was to include the best available evidence to provide a synthesis of information to guide panel decision-making.⁹  To avoid indications for which there was no evidence, we included evidence from lower-quality studies (eg, cross-sectional studies, surveillance studies, consecutive cases) and studies outside the initial scope of this review (NICU, adult, and laboratory studies). The inclusion of these studies for some indications means that future clinical decision-making in certain pediatric populations is not guided by strong, evidence-based recommendations. However, most of the evidence that was broadened beyond the original scope of the systematic review was well supported by CPGs, suggesting that these findings can be implemented into clinical practice with confidence.

In this systematic review, we provide the first synthesis of the breadth of evidence available for the selection and insertion of VADs in pediatric patients to guide clinical decision-making. There was strong evidence to support and facilitate appropriate clinical decision-making in some pediatric indications. However, certain populations, device types and characteristics, and insertion procedures were poorly evidenced, necessitating the application of clinical judgment for some indications. Overall, the findings of this review will be vital to inform criteria using the RAND-UCLA Appropriateness Method to determine the appropriateness of VADs in pediatric patients.

CPG

clinical practice guideline

CVAD

central venous access device

DVT

deep vein thrombosis

ECG

electrocardiogram

ED

emergency department

IV

intravenous

MeSH

Medical Subject Headings

NIR

near-infrared

PICC

peripherally inserted central catheter

PIVC

peripheral intravenous catheter

PN

parenteral nutrition

RAND-UCLA

RAND Corporation–University of California, Los Angeles

RCT

randomized control trial

USG

ultrasound-guided

VAD

vascular access device