Ultrasound-guided vascular access in the neonatal intensive care unit: a nationwide survey

Ultrasound-guided vascular access (USG-VA) is recommended by international practice guidelines but information regarding its use in the neonatal intensive care unit (NICU) is lacking. Our objective was to assess neonatologist’s perceptions and current implementation of USG-VA in Spain. This was a nationwide online survey. The survey was composed of 37 questions divided in 4 domains: (1) neonatologist’s background, (2) NICU characteristics, (3) personal perspectives about USG-VA, and (4) clinical experience in USG-VA. One-hundred and eighty survey responses from 59 NICUs (62% of Spanish NICUs) were analyzed. Most neonatologists (81%) perceive that competence in USG-VA is indispensable or very useful in clinical practice. However, 64 (35.5%) have never used USG-VA in real patients. Among neonatologists with some experience in USG-VA most perform less than 5 procedures per year (59% in venous access and 80% in arterial access) and a 38% and 60% have never used USG for venous and arterial access, respectively, in very low birth weight infants (VLBWI). More than a half of neonatologists (55.5%) use US to check catheter tip location but a 46.6% always perform a radiography for confirmation. Spanish neonatologists report that resident/fellow training in USG-VA is absent (52.2%) or unstructured (32%) in their units. The lack of adequate training is identified by a 60% of neonatologists as the most important barrier for implementation of USG-VA and 87% would recommend that future neonatologists receive formal training. Conclusion: Spanish neonatologists perceive that USG-VA is important in clinical practice but currently, these techniques are largely underused. Our results indicate that specific training in USG-VA should be implemented in the NICU. What is Known: • Ultrasound-guided vascular access is recommended as the preferred method for central venous access and arterial line placement in children and adults. • The degree of current implementation of ultrasound for vascular access in the NICU and the perceptions of neonatologist about its use are largely unknown. What is New: • Most neonatologists consider that competence in ultrasound-guided vascular access is an indispensable aid for clinical practice. • However, most neonatologists are not adequately trained in ultrasound-guided vascular access and the technique is largely underused. What is Known: • Ultrasound-guided vascular access is recommended as the preferred method for central venous access and arterial line placement in children and adults. • The degree of current implementation of ultrasound for vascular access in the NICU and the perceptions of neonatologist about its use are largely unknown. What is New: • Most neonatologists consider that competence in ultrasound-guided vascular access is an indispensable aid for clinical practice. • However, most neonatologists are not adequately trained in ultrasound-guided vascular access and the technique is largely underused.


Introduction
Percutaneous vascular access (PVA) in children is more challenging than in the adult patient due to several factors such as vessel size, anatomical variations, lack of collaboration or need for sedation. [1,2] The neonatal population represents a very particular subset where vascular access is highly problematic, especially in the preterm infant and there is a lack of high level evidence to guide vascular access practices [3][4][5]. Most preterm and sick neonates require a central venous access. Umbilical vessels are the rst choice for central access during the rst days of life.
[6] Thereafter, most patients receive a peripherally inserted central catheter (PICC) like the epicutaneous-cava catheter (ECC). [7] While such devices su ce for routine care in most instances, the sickest patients who need high intensity care bene t from large-bore central venous catheters (CVC) and arterial lines. Placement of a CVC in neonates has been traditionally considered a high-risk procedure. However, there are several studies that con rm that USG-CVC is feasible and safe in neonates and even preterm infants. [8][9][10][11] Recent clinical practice guidelines on point of care ultrasound (POCUS) have been released by the European Society of Pediatric and Neonatal Intensive Care (ESPNIC) [12]. In these guidelines the use of USG is recommended for placement of internal jugular vein (IJV) catheters (strong agreement, quality of evidence A) and it is considered helpful for insertion of other CVCs, peripherally inserted central catheters (PICC), arterial lines as well as for the assessment of central venous catheter (CVC) tip location (strong agreement, quality of evidence B). Despite of the paramount clinical relevance of vascular access in neonates, there is a complete lack of data regarding the current implementation of USG-PVA in the NICU.
With the hypothesis that USG-PVA is underused, the objective of the present study was to assess current practice regarding USG-PVA in the Neonatal Unit as well as to analyze the perceptions of Spanish neonatologist regarding the clinical utility, risks and barriers for implementation of these techniques in their current practice.

Materials And Methods
This was an online-survey based study promoted by the Working Group on Ultrasound (WG-US) of the Spanish Society of Neonatology (SENeo). The WG designed the study according to the CHERRIES and CROSS guidelines for survey studies. [13,14] The target population were active Spanish neonatologists who were identi ed from the a liate registry of the SENeo. One expert in USG-PVA from the WG and one additional expert not involved in the WG-US activities developed the rst draft of the survey. The initial proposal included 46 questions divided in 4 domains: 1) neonatologist's data and background, 2) NICU characteristics, 3) clinical experience in USG-PVA and 4) perceptions about USG-PVA). This initial proposal was reviewed by a panel of 6 neonatologist from the WG-US with interest and expertise in USG-PVA. The WG members discussed the appropriateness of every question. Additionally, each member of the panel was allowed to add and remove up to three questions. Each question was assigned a priority (high, medium, low) based on rating of the content validity and reliability using a prede ned color-coded Likert scale. [15] Four rounds of corrections were needed until a consensus was reached. The nal survey consisted in 37 questions (Appendix 1 in supplementary content) including an adaptative question to reduce complexity (only those responders who reported some experience in USG-PVA were allowed to answer questions in the domain 3). No personal information from participants apart from the email address was recorded. Only one response per participant was allowed by requesting and checking email address with each response. Incomplete surveys were discarded from the analysis.
The survey was uploaded to Google Forms®. The survey was headed by a brief text with a description of the research group and the study objectives. The questionnaire was arranged in four screen pages and estimated time for completion was 15 minutes. Responders were allowed to review and change responses ("back-button") before nal submission.

Results
One hundred and ninety-three survey responses were received (response rate 25%). Thirteen surveys were discarded because incomplete data (n=8) or duplicates (n=5). Finally, a total of 180 surveys coming from 59 NICUs (62% of Spanish NICUs) were included in the analysis. Data related to neonatologist's background and NICU characteristics are summarized in Table 1. The neonatologist's mean (SD) age and years of clinical practice were 41 (7.6) and 12.6 (7.4) years respectively. Most neonatologist worked in teaching units (66.7%) but reported that training of residents/fellows in USG-PVA was absent (52.2%) or occasional (32.2%) in their units. Sixty-four (35.6%) of the responders had no experience placing USG-PVA in real patients. These practitioners were signi cantly older (p<0.001), have received previous training in POCUS (p<0.001) or USG-PVA (p<0.001) less frequently and worked in lower-complexity (p<0.001) and smaller NICUs (p<0.001) compared to those who have performed USG-PVA in clinical practice. Nearly a half of responders (51.1%) reported that less than 25% of the attending neonatologists perform USG-PVA in their units and only a 13.3% reported that more than>75% of the staff perform these techniques regularly. Among those neonatologists (n=116) with some clinical experience in USG-PVA (Table 2) a 62% perform less than 5 USG-CVC insertions per year and a 38% have never used this technique in VLBWI ( Figure 1A). Most neonatologist report that the main indication for USG-CVC insertion is as a rescue procedure after exhausted peripheral vein access (53.4%) followed by acute hemodynamic instability (27.5%).
Only a minority perform this technique electively (18.9%) for administration of multiple drugs, blood draws or in high-risk surgical patients. The femoral vein (FV) is the preferred insertion site for most neonatologist (62%) followed by the internal jugular vein (IJV) in 32.7%. The use of supraclavicular brachiocephalic (BCV)/subclavian vein cannulation was infrequent (5.1%). Regarding arterial lines most neonatologists (79%) report that they perform less than 5 procedures per year and a 60% have not inserted any USG arterial line in VLBWI ( Figure 1B). The main indication for USG arterial line placement is invasive blood pressure monitoring (70.7%). The preferred insertion site for most neonatologist is the femoral artery (62%) followed by the radial artery (31%). Only responses from neonatologists with some experience using USG-PVA are included in this description. *These data are referred to the whole sample (n=180). High complexity NICU is de ned as that providing cardiac surgery and ECMO. CVC: central venous catheter; ECHO: echocardiography; NICU: neonatal intensive care unit; PVA: percutaneous vascular access; USG: ultrasound guided; US: ultrasound, UVC: umbilical vein catheter; VLBWI: very low birth weight infants.
Regarding personal perceptions about USG-PVA (Table 3), most neonatologists believe that attaining competence in these techniques is indispensable (30%) or very useful (56.7%) for their practice and would recommend that current Neonatology trainees be taught USG-PVA as an indispensable (40.6%) or very useful skill (46.1%). A 47.8% of the neonatologist believe that reduced vessel size is the main factor that makes PVA di cult. A 60% of the responders identify the lack of training as the main factor limiting the implantation of USG-PVA over other factors such as technical di culty (18.9%) or lack of adapted equipment for neonates (16.7%). When asked about different locations for insertion of USG central lines, only a 33% identi ed the BCV as the largest central vein amenable for percutaneous puncture and only a 15.6% would consider supraclavicular cannulation of the BCV as the optimal USG technique for central venous access in neonates. A signi cant proportion (32.2%) believe that FV is larger than IJV or BCV. Most neonatologists consider that the risk/bene t balance is favorable to conventional ECCs compared to PVA of central veins, especially in preterm infants. In multivariate logistic regression lower operator age and previous training in the technique were associated with current use of USG-PVA in clinical practice (supplementary table 1). However, when factors associated with a higher number of USG central venous access performed per year were assessed, we found that the main predictors were longer clinical experience using USG-PVA and higher operator's age (supplementary table 2). The relationship between operator's age and use of USG-PVA is shown in gure 2.

Discussion
Effective and safe vascular access is a critical component of care of sick neonates. USG-PVA is recommended for insertion of central venous catheters and arterial lines. [12,16,17] There have been several studies assessing the implementation of different POCUS applications, including USG-PVA, in the pediatric intensive care unit[18-21] but a similar information in the NICUs is lacking. In particular, the degree of uptake USG-PVA in the NICU is largely unknown. Our nationwide survey is the rst study assessing neonatologist's personal perspectives about USG-PVA and current implementation in the NICU. Our results indicate that neonatologists perceive USG-PVA as an indispensable or very useful aid in daily clinical practice and recognize the need for training during Pediatrics residency and Neonatology fellowship. However, our results indicate that USG-PVA is used infrequently in Spanish NICUs. As many as one third of the neonatologists in our survey have never used USG-PVA in their practice. Among those who use USG-PVA in their practice the number of procedures per year is quite low specially regarding placement of arterial lines and the use is even lower in the case of preterm infants.
Our results provide some insight into possible explanations behind the underuse of USG-PVA in the NICU. The rst relates to training in vascular access. Most of the neonatologist who responded to our survey had not received formal training in POCUS or USG-PVA. In addition, an important proportion of neonatologists reported that residents and fellows are not trained in USG-PVA in their units. In this line, most neonatologists identify the lack of training as the most important limitation for implementation of USG-PVA above others such as lack of equipment, di culty of the procedure or inherent risks. Those neonatologists working in high-complexity and large NICU used more frequently USG-PVA. However, in multivariate analysis only operator's previous training and age were independent predictors of the use of USG-PVA. This nding suggests that structured training is essential to acquire USG-PVA competence in neonates as these procedures are performed infrequently in clinical practice. For most neonatologists 10 to 50 procedures would be necessary to attain competence in USG-PVA. This number is similar to that recommended by some expert authorities. [22] According to our results, this may take years of clinical practice. Therefore, there is a need to implement training strategies to acquiring competence in USG-PVA more rapidly without compromising patient's safety. There are several recommendations on how USG-PVA training should be accomplished [22,[22][23][24]. Simulation is a powerful aid for training US guided procedures including vascular access. [25][26][27] Several vascular access models can be customized to recreate tiny vascular structures that are easy and cheap to build [18,28]. In these models, the trainee gets familiarized with US vascular exploration and can repeatedly train dynamic needle tip guidance which is the essential skill for effective vascular access.
A second possible explanation for the scarce uptake of USG-PVA relates to the perception of risk and bene ts of inserting CVCs and arterial lines in neonates. Most neonatologists considered that CVCs have a less favorable risk/bene t ratio compared to standard ECCs, specially in preterm infants. [29] In fact, the main indication for USG-CVC placement in our study was rescue from exhausted peripheral vein access. It seems that, USG-PVA is used mainly as a last resource for vascular access and perhaps performed late during evolution of NICU stay. In our opinion, this "high-risk" perception regarding CVC insertion in neonates may be an inertia from the past when CVCs were inserted by surgical cut-down in often extreme situations but current data in the US era do not support this conception. Recent research indicates that USG-PVA is safe in neonates. In particular, a growing number of reports have found that supraclavicular access to the BCV is feasible even in preterm infants with high insertion success rate and few mechanical complications and has been proposed as the optimal technique for CVC insertion in small infants and neonates [9,10,30,31]. Furthermore, some data suggest that this technique may reduce catheter related blood stream infection compared to PICCs and other CVCs [32,33]. However, the reported use of the BCV access in our survey was anecdotical. The most common location for central venous access was the FV. This is surprising as the FV is much smaller compared for instance to the IJV or BCV. [4,31,34] In addition, the FV may increase the risk of CRBSI or thrombosis. When asked to identify the largest vein amenable for PVA in neonates only one third of neonatologists mentioned the BCV and a similar number believed that the FV is the largest central vein for PVA. This indicate that most basic anatomical features of central veins might be ignored.
Beyond perceived risks, it is also possible that practitioners do not appreciate much bene t from the use of CVCs in neonates and this may hold true in most instances. ECCs are the most common devices for mid-term central venous access in neonates and they are excellent for routine care such as administration of parenteral nutrition or antibiotics. However, there are important limitations in the performance of these catheters such as low infusion rates, inability to infuse simultaneous medications, impossibility to draw blood or obtain hemodynamic monitoring as well as frequent catheter block and tip malposition. The sickest neonates and preterm infants who need high intensity care bene t from a large-bore safe and stable central catheter and competence in USG-CVC insertion is essential in these cases.
In our survey US was frequently used to check CVC tip position but still a majority of neonatologist perform a radiography for con rmation and expose the infant to ionized radiation. There have been several studies demonstrating accuracy of US in detecting central line tip and some have suggested that US should be the rst-line modality to check central line tip location. [35,36] Regarding USG for arterial line placement, current use is even less frequent with most responders performing less than 5 procedures per year. In addition, most of the responders have never placed an USG arterial line in VLBWI. As expected, the most common reported indication for arterial cannulation was invasive arterial pressure monitoring. However, it seems that arterial lines are being placed infrequently and that most neonatologist rely on non-invasive blood pressure. This may not be surprising as technical di culty for placing arterial lines in neonates is high and the risks derived from catheter related arterial ischemia are substantial. However, it has been demonstrated that non-invasive blood pressure is not reliable, especially in hypotensive preterm infants. [37] Hypotension is a strong predictor of mortality and adverse neurological outcome[38] and therefore, accurate and continuous measurement of arterial blood pressure is essential in the sickest infants with hemodynamically instability. USG facilitates not only cannulation, but also catheter selection based on actual arterial size a should be considered in hemodynamic compromised neonates when umbilical artery catheterization is not available. The reported preferred site for arterial line placement in our survey was the femoral artery. However, the use of USG radial artery catheterization may offer some advantages such a less risk of infection and occlusion-related ischemia given the presence of collateral circulation though the cubital artery. [39] One interesting result in our study was the effect of operator's age in the reported use of USG-PVA. Age was an independent predictor of USG-PVA use. As expected, those practitioners with some experience in USG-PVA were younger and had less years of clinical practice compared to those who had never used this technique. Several studies have shown that senior practitioners may perform worse when switching to US and therefore may be reluctant to use USG-PVA in their practice ("old dog and new tricks" phenomenon) [20]. However, when considering only responses from neonatologist who currently use USG-PVA, we observed that older practitioners place more USG-CVCs per year compared to the younger ones. This may suggest that acquiring con dence in these techniques in neonates is a slower process compared to children and adults. Taken together our results indicate that training in USG-PVA in Neonatology should begin early.
The main limitation of our study is the inherent risk of bias due to self-reporting of data and the relatively high non-response rate [40]. The willingness to respond the survey may be dependent on the personal interest on the topic of USG-PVA. It is possible that many neonatologists who are not interested about USG-PVA have not responded to the survey and that the actual use of UGS-PVA may be even lower than recorded. However, we have received responses from 62% of Spanish NICUs, including community-based hospitals as well as large referral centers so we think that our results are representative of current practice in our country. We think that our ndings are an important rst step to identify the needs and expectations regarding USG-PVA in the NICU. This is critical to elaborate coordinated strategies to implement training in USG-PVA and increase the uptake of this essential procedure in the next years.
In conclusion, despite most neonatologist perceive USG-PVA as very useful in daily practice, currently these techniques are underused. The lack of adequate training is perceived as the most important barrier for implementation of USG-PVA in the NICU. Our study highlights the need to develop and implement speci c training programs in USG-PVA during residency and Neonatology fellowship.