Ultrasound for Pediatric Peripheral Intravenous Catheter Insertion: A Systematic Review

This systematic review with meta-analysis was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. The study was prospectively registered in the PROSPERO database of systematic review protocols (CRD42020213387).

Medline, Embase and CENTRAL databases were systematically searched on November 27th, 2020, based on predefined variables (Supplemental Information). This involved using controlled vocabularies where possible (MeSH and EMTREE headings) with wider free text and synonyms using nesting and use of “OR” Boolean operator. No limits were applied to the variables of “age” or “language,” to not limit studies. Publications were searched from January 1, 2000 to November 27, 2020. Publication type was restricted to randomized controlled trials (RCTs) through The Cochrane Highly Sensitive Search Strategies for identifying randomized trials.

This search was supplemented by reviewing the reference lists of all full text individual papers, including preprints, identified as eligible studies. ClinicalTrials.gov and World Health Organization International Clinical Trials Registry Platform were also concurrently searched for unpublished trials with preliminary data (search limited to pediatric studies with results available). To augment this, a SCOPUS (Elsevier B.V.) database search was performed on final included studies.

Search results were imported to EndNote (X9, Clarivate Analytics, PA). Two reviewers (E.M. and P.S.) independently identified potentially eligible studies for inclusion using a stepwise approach, including removal of duplicates, manual screening by title, and abstract removing papers that clearly did not meet the eligibility criteria (Table 1), and then full text review of the remaining studies. RCTs describing the success rate (first attempt or overall) of ultrasound-guided PIVC insertion in children (<18 years) were included. Studies were excluded if they did not have an adequate reference standard (landmark technique) or studied central venous (eg, peripherally inserted central catheter, or central venous line) or arterial access (Table 1).

The primary outcome measure was first attempt success of ultrasound-guided PIVC insertion, compared with the first attempt success of traditional landmark-based techniques. Secondary outcome measures included overall success of PIVC insertion and the time to catheterization success in the ultrasound-guided and landmark groups.

Initially, 2 reviewers (E.M. and P.S.) independently screened eligible studies gathered according to the above-described approach on title and abstract, and classified them as being relevant, potentially relevant, or not relevant. Next, the full text of the articles that were classified as being relevant were analyzed by both reviewers independently, deciding individually whether they were eligible, based on the inclusion and exclusion criteria (Table 1). Any discrepancy between reviewers was resolved with a final decision from a third independent investigator (P.J.). Eligibility of studies initially classified as potentially relevant was also decided by the third investigator, after which those studies with a positive final decision were included. Data tables were developed, modeled off Cochrane data extraction forms for RCTs.¹³  Data were extracted under subgroups of study details, eligibility, demographics, specific patient cohort details, study design, catheterization details, ultrasound details, and outcomes, using an Excel spreadsheet (Microsoft Corp, Redmond, WA). These data were independently collected and verified by both reviewers (E.M. and P.S.).

Methodological quality of each study was assessed at the study inclusion level using the Revised Cochrane risk-of-bias tool for randomized trials (RoB 2).¹⁴  This tool provides a standardized, table-based approach to assessment for selection, detection, attrition, and reporting bias.¹⁵  Individual templates were independently completed by 2 reviewers (E.M. and P.S.) with overall risk of bias judgement reported before any disagreements resolved by discussion.

Extracted data were analyzed using Stata (16.1, StataCorp LLC, TX) software. Pooled odds ratio (OR) and percentage success rate were derived for dichotomous variables (first attempt and overall success) and weighted mean difference (WMD) for continuous variables (time to catheterization) for the comparison of ultrasound to landmark approaches. DerSimonian-Laird random-effects model was used for all outcome measures. I² was calculated as a quantifying measure of heterogeneity, with I² > 60% defined as significant heterogeneity. Forest plots were used to present results for each outcome, with meta-analysis results included when significant heterogeneity was not present. Funnel plots were used to look for evidence of publication bias. A sensitivity analysis was performed by excluding studies with alternate study designs.

Post hoc analysis by subgroup was performed for identified important variables with data available from at least 3 studies (Supplemental Table 6). These categories included age ≤3 years, physician operator, clinical setting (“emergency department,” “operating room”), ultrasound technique (“dynamic,” “single operator,” “short axis”), site (“lower limb,” “other”), sedation, and “difficult intravenous access.”

Initial search of CENTRAL, Embase, and Medline identified 1033 total articles, and after removal of duplicates, 759 remained (Fig 1). Title and abstract screening reduced this to 41 papers for full text evaluation. Of these, 12 studies were excluded without a pediatric cohort of patients, 3 without landmark catheterization as a control, 1 without an inclusion outcome, 2 containing the same data set of an included study, 1 was a comparative trial with nonrandom consecutive allocation of patients, and 13 listed trials without published works or results available. The remaining 9 studies underwent a SCOPUS database search, as well as manual review of their reference list, without any further eligible studies identified. Trials databases found 1 eligible study with published results but it had already been included.¹⁶  Manual review of related published systematic reviews failed to identify missing studies.^9–12 

A total of 1350 patients across the 9 eligible studies were included for analysis (Table 2).^16–24  One study only included patients less than 5 years of age,²⁰  while 4 studies were limited to patients ≤3 years of age.^{16,19–21,23}  The male proportion was 58.4%, with 1 study not reporting sex.²⁰  Distinct intervention and control groups were present throughout, other than Gopalasingam et al²³  being a crossover study. Bian et al¹⁹  was the only paper not yet published in a peer reviewed journal at the time of initial search.

Six studies included only patients with DIVA, defined as either catheterization attempt(s) failure, age ≤3 years, history of DIVA or use of validated tool,^{16,19–22,24}  with Bian et al¹⁹  also limited to infants or toddlers with congenital heart disease (Supplemental Table 7). Four studies provided sedation during catheterization attempt, all conducted in the operating room.^16,19,21,23  Two studies used solely the static (vein marked) ultrasound technique,^19,20  and the rest performed dynamic (real-time needle guidance) ultrasound^16,21–24  or a mixture of the 2 techniques.^17,18  Catheterization sites varied, with 3 studies predominantly²¹  or only^16,19  using lower limbs.

There was significant methodological variability between studies for total number of catheterization attempts, ranging from 2 to 4 total attempts, and with and without total time limits.^{16,19,21–24}  This subsequently affected overall success and number of attempt rates. Vinograd et al²⁴  measured time to catheterization from randomization while all other studies reporting time from procedure commencement (ultrasound placement versus tourniquet placement) to PIVC flush without evidence of extravasation.

None of the included studies were assessed to be at high risk of bias, allocated either low risk or as having some concerns regards bias (Table 3). Of concern there was an alternative patient allocation strategy in Gopalasingam et al,²³  with a crossover study design. Given its structured allocation and general inability to blind intervention, it was included in the analysis. The funnel plots showed no obvious asymmetry to indicate publication bias (Supplemental Figures 6–8).

First attempt success was reported in 8 studies, with a total of 623 patients in the ultrasound group and 637 in the control group (Table 4). Statistically significant improvement in first attempt success with ultrasound was detected in 5 of 8 studies (Fig 2). Overall, first attempt success rate was 78% in the ultrasound group compared with 66% in the control group (Table 4). Meta-analysis suggested improvement in likelihood of catheterization success on the first attempt with ultrasound compared with the landmark technique, with pooled OR 2.61 (95% CI 1.18 to 5.76, P = .018). However, this analysis was limited by significant heterogeneity, with I² = 86.4%. Sensitivity analysis with Gopalasingam et al²³  excluded showed similar results, including excessive heterogeneity (pooled OR 2.52, 95% CI 1.04 to 6.09, P = .041, I² = 87.9%). A funnel plot of individual effect estimates showed significant heterogeneity but no asymmetry to suggest reporting bias (Supplemental Fig 6).

Six studies reported overall success rates, with a total of 300 patients in the ultrasound group and 301 patients in the control group (Table 4). Of those 6 studies, 4 showed statistically significant improvement in overall success with ultrasound, with a combined success rate of 93% in the ultrasound group and 78% in the control group. Meta-analysis showed that ultrasound improved overall success rates compared with the landmark technique, with pooled OR 3.57 (95% CI 2.05 to 6.21, P < .001) (Fig 3). Heterogeneity of this secondary outcome measure was low, with I² = 0.0%. Sensitivity analysis with removal of Gopalasingam et al²³  showed similar results (pooled OR 3.33, 95% CI 1.89 to 5.87, P < .001, I² = 0.0%).

Time to success was reported in 7 studies. Statistically significant improvement in time to success was seen in 4 out of 7 studies (Table 4). The weighted mean difference was 131 seconds (95% CI 22 seconds to 240 seconds, P = .019) improvement in time to catheterization in the ultrasound group (Fig 4). However, this analysis was limited by marked heterogeneity (I² = 93.6%). Vinograd et al²⁴  was identified as a statistical outlier with markedly different time to catheterization outcome data, related to an alternate measure of time compared with other studies (from randomization). Sensitivity analysis after removal of Vinograd et al²⁴  and Gopalasingam et al²³  showed no difference in time to catheterization, with heterogeneity remaining very high (WMD = 87 seconds, 95% CI −24 to 199 seconds, P = .125, I² = 92.2%)

Meta-analyses of data by subgroup showed statistically significant improvements in likelihood of first attempt success rates with ultrasound-guided PIVC insertion compared with the landmark technique for the subgroups of: physician operator, operating room setting, dynamic ultrasound technique, single operator ultrasound technique, lower limb site, sedation, and DIVA (Table 5, Fig 5). All of these subgroups also had acceptable heterogeneity, with I² < 60%. Four studies (Bian et al,¹⁹  Benkhadra et al,²¹  Gopalasingam et al,²³  and Hanada et al¹⁶ ) were conducted in the operating room with patients receiving sedation, with improvement in first attempt success seen in this subgroup (n = 386, pooled OR 4.58, 95% CI 2.45 to 8.56, P < .001, I² = 37.6%). Ultrasound also improved first attempt success in patients with difficult IV access (n = 495, pooled OR 4.60, 95% CI 2.34 to 9.07, P < .001, I² = 57.4%). Improvement was not demonstrated in the emergency department (ED) subgroup, which was also limited by high heterogeneity (n = 492, OR 1.98, 95% CI 0.45 to 8.69, P = .368, I² = 90.4%). The combination of single operator, dynamic, and short-axis technique was used in Gopalasingam et al,²³  Hanada et al,¹⁶  and Vinograd et al,²⁴  and was associated with marked improvement in first attempt success (n = 367, pooled OR 5.97, 95% CI 3.57 to 10.0, P < .001, I² = 0.0%). The subgroup analysis of overall success showed similar results, with improved success with ultrasound seen for the subgroups of age 3 or less, physician operator, operating room setting, dynamic, single operator and short-axis ultrasound technique, both lower limb and other catheterization sites, sedation, and DIVA (Table 5, Fig 5).

This systematic review suggested that the use of ultrasound increased first attempt and overall success PIVC insertion rates in pediatric patients compared with a standard (landmark) approach. Previous systematic reviews on the topic of ultrasound for pediatric PIVC insertion were limited by a paucity of studies to enable a dedicated meta-analysis.^9–12,25  A systematic review on pediatric catheterization strategies has been published with only 3 clinical papers identified and did not support the use of ultrasound, necessitating the need for further RCTs.²⁵  The uptake of ultrasound-guided PIVC insertion has since burgeoned over the past decade, along with numerous studies supporting its benefit in the pediatric population.^{16,19,21,23,24}  This systematic review showed statistically significant improvement in first attempt and overall success with ultrasound in the majority of included studies. Meta-analysis demonstrated improved rates of overall success with ultrasound. However, meta-analysis of the primary outcome of first attempt success was limited by high heterogeneity.

The predominant application for ultrasound-guided PIVC insertion is in patients considered to have DIVA.⁸  The definition for this in children is broad and includes factors such as having no visible and/or palpable veins, younger age, previous history of DIVA, history of prematurity, increased adiposity or obesity, dehydration, frequent phlebotomy or PIVC insertion or comorbidities.^26–28  The use of ultrasound in patients with DIVA was associated with an overall benefit in first attempt and overall success rates, which is inherently the main utility of ultrasound-guided insertion. This was further supported by subgroup analysis of overall success in patients at or under 3 years of age, who are also considered to have DIVA. However, the use of ultrasound was reported to be beneficial for PIVC insertion in children with obesity in the study by Hanada et al,¹⁶  but had no demonstrable difference in the study by Curtis et al.¹⁷ 

Although several studies support the use of ultrasound for pediatric PIVC insertion in the ED,^20,22,24  the subgroup analysis of this setting overall was not statistically significant and had high heterogeneity. This could partly be attributed to the confounding issue of children not being still enough for appropriate use of ultrasound in some studies,¹⁷  and no children were sedated. It could also reflect the often-chaotic environment or a higher proportion of unwell children compared with other settings, although many studies excluded children from their trial if they were critically unwell.^17,22,24  The ultrasound technique and lack of adequate training on actual patients may have also been contributing factors. More studies are required in this setting for conclusive evidence.

Unsurprisingly, the use of ultrasound for pediatric PIVC insertion in the operating room setting was highly effective. Highly skilled operators (anesthesiologists) combined with patients who were anesthetized and still, provided conditions highly conducive to successful ultrasound-guided PIVC insertions.^16,19,21,23  Furthermore, many of the sedative agents used, such as Sevofluorane, are known to be potent vasodilators.²⁹  Additionally, the majority of patients were undergoing elective procedures or imaging.^16,21,23  All studies involved children ≤3 years and, apart from 1 study,²³  all primarily used the saphenous vein,^16,19,21  which may otherwise have been challenging in an awake infant or young child.

The technique for ultrasound-guided PIVC insertion varied considerably between studies. The dynamic ultrasound method, which involved visualizing the needle tip in real-time, in either the long or short axis, was associated with a higher insertion success rate than the use of a static ultrasound method, which involved marking-up the vein, before “blind” insertion. The single-operator technique and short-axis technique subgroups had improved overall success over the landmark technique but had excessive heterogeneity for meta-analysis of first attempt success. The single-operator technique has the additional advantage of being less resource intensive. The short-axis was the most used axis and has been demonstrated to be the most effective axis for PIVC insertion in a dedicated systematic review, although this only included RCTs on adult patients or phantom models.³⁰  However, the long-axis was demonstrated to be useful for large, straight veins (eg, saphenous) in a sedated child²¹  or can be used to confirm placement after short-axis insertion.^22,31 

Combining all these elements, the standard technique for ultrasound-guided PIVC insertion in children is arguably a dynamic, single-operator, short-axis approach using a high-frequency linear probe.^31,32  This combination was used by ultrasound operators in 3 studies, 2 in the operating room^16,23  and the other in the ED,²⁴  and was associated with high first attempt and overall success rates. Routine tourniquet use with this technique may also assist with avoiding vein compression from the ultrasound probe.^16,23  Moreover, topical anesthetic medication may mitigate pain and thereby increase cooperation in an awake child.³³  Appropriate training and experience of the operators is crucial to the successful implementation of ultrasound, with 1 study reporting higher PIVC success rates after at least 15 ultrasound-guided insertions in adult patients,³⁴  which may imply more training is required in children, given the finer psychomotor skills required.

Besides improving PIVC insertion success rates, ultrasound has numerous other potential advantages over a landmark technique insertion. Use of ultrasound has been associated with fewer attempts (skin punctures)^{16,19–22,24}  and redirections (partial withdrawal with advancement in new direction),^19,20,22,23  which are important factors for long-term vessel health preservation.³⁵  It also provides the ability to grade the size and quality of veins before selection for insertion,^16,19,23,24  it enables insertion away from flexural regions for joint mobility and comfort compared with the landmark technique,²³  and can allow for placement of longer or larger peripheral devices for increased longevity.²⁴  Furthermore, ultrasound has been associated with a higher level of parental satisfaction.²⁴  Most studies reported no significant increase in complications compared with the landmark technique, apart from an inconsequential arterial puncture in 1 study.²² 

Ultrasound use was associated with a reduction in time to catheterization in 4 of 7 studies. This was confounded by significant heterogeneity of studies, with Vinograd et al²⁴  attempting to address the total procedure time for ultrasound (locating the machine, cleaning, and operating it) by defining it from time of randomization. This high heterogeneity precluded meta-analysis and subgroup analysis.

The main limitation from this systematic review and meta-analysis was the significant heterogeneity of studies. A random-effects model, using skewed weighting of individual data sets, was used to help compensate for this. Where heterogeneity was significant (I² > 60%), meta-analysis was treated as hypothesis-generating only for main outcome measures and not performed for subgroups. The control method varied with 3 studies allowing adjunct methods, such as transillumination,^22,24  infrared,¹⁷  or heat packs,²⁴  although, this may have only overstated the outcomes in the landmark technique group. Due to the open-label nature of the intervention, operators were unable to be double-blinded. Gopalasingam et al²³  performed a cross-over trial with subsequent concern for a degree of carry-over effect.

In terms of strengths, several features support the validity of this systematic review. Publication bias was minimized by searching the literature as broadly as possibly, including an unpublished study (preprint) in the final analysis. A strict, stepwise search and data extraction was performed by independent reviewers following a prospectively published protocol. Of the included studies, none were identified as highly biased. A sensitivity analysis was performed throughout, excluding the Gopalasingam et al²³  study with an alternate design.

Future RCTs should be consistent with clear reporting on the ultrasound technique, standardization of training, and outcome measures. They could also potentially incorporate the neonatal age group (<3 months), which is currently lacking in the literature. Furthermore, in environments outside of the operating room, methods of restraint, presence of parent or guardian, distraction tools (eg, virtual reality), topical anesthesia, and methods of appropriate sedation should be explored, as this could improve the effectiveness of ultrasound for patients with DIVA by facilitating a still target vessel.

This systematic review suggested that ultrasound improved the first attempt success rate of pediatric PIVC insertion, although meta-analysis of this outcome measure was limited by high study heterogeneity. ultrasound-guided PIVC insertion improved overall success rates as well as both first attempt and overall success rates in patients with DIVA. The standard ultrasound technique should be a single-operator, dynamic, short-axis approach. This study provides a robust evidence base to support the routine use of ultrasound for PIVC insertions in children with DIVA and should be implemented as standard of care across clinical settings.

We thank Ms Sarah Thorning (Research Librarian, Gold Coast University Hospital) for guidance on the search strategy; Dr Ian Hughes (Biostatistician, Gold Coast University Hospital) for assistance with the statistical analysis; Professor Robert Ware (Biostatistician, Griffith University); and Professor Gerben Keijzers (Emergency Physician, Gold Coast University Hospital) for their expert advice and support.

CI

confidence interval

DIVA

difficult intravenous access

LM

landmark

NR

not reported

OpR

operating room

OR

odds ratio

PIVC

peripheral intravenous catheter

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-analyses

RCT

randomized controlled trial

RoB 2

revised Cochrane risk-of-bias tool for randomized trials (RoB 2)

WMD

weighted mean difference