The main results of the study are that (1) the incidence of catheter misplacements was 33%, with half of them due to a tip position proximal to SCV-RA junction; (2) CE-TTE subcostal bicaval view was the best transthoracic echocardiographic approach providing 92% diagnostic accuracy and 79% concordance with CE-TEE; (3) CXR, as well as CE-TTE four-chambers view, had a limited value in the detection of misplacements with a poor concordance with CE-TEE.
A CVC tip positioned at the transition from the SVC to the right atrium, lying parallel to the major axis of the vessel, is considered as the “ideal position” however difficult to obtain. CVC-related venous thrombosis has a high prevalence when the catheter tip is positioned in the superior upper third of the SVC or in the brachiocephalic vein  and represents a major cause of late complications finally resulting in SVC syndrome, infections and pulmonary embolism, occurring mostly in oncologic frail patients [19, 20].
In our study, the incidence of catheter misplacements was higher than previously reported [14, 21]. Possible explanations are the following: first, in most of other studies  chest radiography was choose as reference standard even if inaccurate for tip location ; second, in our center catheters were placed by standard Seldinger technique using anatomical landmarks, without use of intraoperative ultrasound guidance or navigation procedures; third, 20 cm of catheter length were inserted in the right internal jugular vein, because this is the current clinical practice in our center. Altogether, the above methodological issues might have increased the rate of misplacements, but it seems unlikely that they might have affected the results.
TEE is considered the gold standard to assess the correct position of CVC though it is not free from complications related to probe insertion and should not be therefore used for this purpose. The strong concordance between TEE and TTE subcostal bicaval view (k = 0.79) is clinically important. Indeed, TTE subcostal bicaval view is non-invasive, frequently obtainable, and based on direct visualization of the CVC tip at the SVC-RA junction. In seven of our cases, we classified the catheters as misplaced even if their tips were correctly placed at a median distance of 2.5 cm (IQR: 1.5–3 cm) from the CT. This was presumably due to the difficulty of visualizing the CVC tip when lying in the proximal portions of the SVC. However, our results show that the probability of classifying a correctly placed catheter as misplaced is 9.7 time lower than the probability of classifying a misplaced catheter as correctly placed.
The CE-TTE four-chambers view allowed showing the correct position of catheter tip in the SVC through the “bubble-test” ; additionally, a low concordance with CE-TEE (k = 0.17) was observed. CE-TTE four-chambers view misdiagnosed catheter position in 32 patients. These findings may have a twofold explanation. First, bubbles reach the RA even in cases of intravascular misplacements. Second, it may be difficult to differentiate between bubbles originating within the RA from bubbles reaching the RA, because CE-TTE four-chambers view does not allow direct visualization of the SVC-RA junction. Indeed, by this technique, intra-atrial misplacements were more readily detected compared to extra-atrial misplacements and the high number of extra-atrial misplacements depicted by TEE led to a lower sensitivity than previously described .
Evaluation of different patterns of jet flow (laminar or turbulent) and bubbles transition-times have been suggested  to define if the arrival of the bubbles was associated with correct positioning. We decided not to adopt this method, because in our experience transition-times may be dependent on (a) patient hemodynamic, (b) length and diameter of the catheter, (c) CVC insertion point, (d) infusion velocity of contrast media, and (e) assessment difficulties (no commercially available device to synchronize “push-to-bubble time”). A previous study  investigating the “qualitative” bubble test, defined as the complete RA opacization versus “few or no bubbles” detection, reported that the complete opacization of RA occurred in half of the patients with CVC misplacements, claiming its low diagnostic accuracy. Furthermore, another study has questioned the usefulness of the proposed cut-off transition times to confirm central catheters tips position .
Some ultrasound-based tip navigation methods suggested direct visualization of the guidewire during the insertion procedure [24, 25]. These methods allow advancing the guidewire with catheter insertion, and misplacements can be quickly recognized. However, this technique can be applied only during real-time ultrasound-guided cannulations, whereas many physicians still use landmark procedures . On the contrary, tip location methods allow checking the catheter tip when already in place. Thus, tip navigation does not replace tip location, which should be assumed as complementary methods .
CXR also showed low concordance with CE-TEE (k = 0.29), as it led to a catheter position misdiagnosis in 27% of cases. These results can be explained considering that the SCV-RA junction cannot be directly visualized by bedside CXR and catheter tip positioning is presumed on the basis of its projection on other fixed anatomical structures taken as radiological landmarks, none of them being 100% reliable [27, 28]. Likewise, the detection of SVC/RA junction by the shadow of SVC on the RA can be difficult .
In recent years, the potential role of transthoracic ultrasound has been investigated as an alternative to CXR in the detection of catheter positioning [12, 13, 15, 16, 30,31,32], but not all studies confirmed these results, mainly because of methodological differences (detection of tip positioning vs. bubble test or subcostal bicaval vs. apical four-chambers scanning view) and lack of a reliable comparator (TEE). This is the reason why, although the ultrasound-guided CVC insertion has been widely accepted to improve success and safety of the procedures [26, 33, 34], the CVC position confirmation by CE-TTE has not yet obtained widespread dissemination  and in many centers CXR is still considered the reference confirmatory test .
To the best of our knowledge, this is the first study rigorously comparing CXR, TTE with TEE. Our results confirm that CE-TTE via short-axis subcostal bicaval view, but neither CE-TTE via apical four-chambers view nor CXR, represents a reliable tool to identify CVC tip misplacements when compared with CE-TEE. CE-TTE via apical four-chambers view is usually used only to rule-out RA misplacements, but in our population the sensitivity was only 50% and it appears to be unreliable in ruling CVC lying proximally in the SCV. Post-operative bedside CXR is always feasible, but its usefulness is limited by its poor diagnostic accuracy.
Limitations: First, our study was conducted in elective anesthetized and pharmacologically paralyzed patients undergoing cardiac surgery. This controlled setting may have influenced the results of the study making TTE scans easier to acquire especially in the subcostal views. Thus, our results may not be directly applied in other clinical contexts such as awake, non-paralyzed, obese patients, emergency, or trauma. Second, all CVCs were inserted percutaneously without ultrasound guide assistance, thus a perioperative vascular scan of latero-cervical, supraclavicular and infraclavicular fossae to identify errant catheters  was not performed. However, we believe that this limitation did not affect the results of the study because vascular ultrasound, however useful for aiding insertion and detecting misplacements in neck veins , is of limited value in confirming tip placement because not suitable for the imaging of SCV. Third, we did not use right atrial electrocardiography during CVC positioning because by this technique it is only possible to detect the tip within the RA but not its exact location within the vascular system. Fourth, only one skilled cardio-anesthesiologist not involved in the CVC placement performed all the examinations to avoid inter-rater variability. Therefore, this source of variability was not considered, and the study was focused on the variability between techniques only. Finally, although the number of patients exceeded the sample size estimation, further single- or multi-center studies including large numbers of patients are needed to confirm our results.