Journal of Perinatology (2013) 33, 950–953 & 2013 Nature America, Inc. All rights reserved 0743-8346/13 www.nature.com/jp
ORIGINAL ARTICLE
Superiority of targeted neonatal echocardiography for umbilical venous catheter tip localization: accuracy of a clinician performance model AS Pulickal1, PK Charlagorla2, SC Tume3, M Chhabra2, P Narula2 and AM Nadroo2 OBJECTIVES: To compare targeted neonatal echocardiography (TnECHO) and antero-posterior (AP) chest radiographs in the localization of the umbilical venous catheter (UVC) tip and to determine the accuracy of UVC tip localization by TnECHO when performed by a trained cohort of pediatric housestaff physicians. STUDY DESIGN: Prospective, observational study of consecutive neonates requiring UVC insertion, in a tertiary care center, in an 18-month period. Chest radiographs reporting optimal position of the UVC tip were compared with the TnECHO results of the pediatric cardiologist. The latter was also compared with the results of TnECHO performed by the pediatric housestaff physicians. RESULT: Thirty neonates with birth weight ranging between 270 and 4490 g and gestational age ranging between 24 and 44 weeks were enrolled. Nine patients (27%) required UVC tip repositioning as the cardiologist performed TnECHO revealed sub-optimal tip position despite optimal position on chest radiography. Among them, four had the UVC tip in the right atrium and five in the left atrium. Compared with the cardiologist, the housestaff physicians had reported TnECHO with a high-accuracy rate (area under the receiver operating characteristic curve ¼ 0.81). CONCLUSION: TnECHO is superior to chest radiography for identifying malpositioned catheters. TnECHO performed by pediatric housestaff physicians with basic training, demonstrated high-accuracy rates. Journal of Perinatology (2013) 33, 950–953; doi:10.1038/jp.2013.96; published online 22 August 2013 Keywords: ECHO; UVC; neonate; X-ray; NICU
INTRODUCTION Umbilical venous catheters (UVCs) are commonly used in preterm and critically sick term infants who require central venous access for fluid resuscitation, medication infusion and parenteral nutrition. However, the placement of the UVC is associated with significant complications. Known cardiac complications of this technique include cardiac arrhythmia, myocardial perforation, pericardial effusion and cardiac tamponade.1–5 Rarer complications include endocarditis and intra-cardiac mycetoma.6,7 Noncardiac complications include intracranial thrombosis, systemic and pulmonary embolism, pulmonary infarction and hemorrhage and extravasation of parenteral alimentation into the pleural cavity.8–10 Antero-posterior (AP) chest radiography is the current standard imaging technique used to localize the tip of the UVC; the landmarks being the thoracic vertebral bodies and the diaphragm. However, case reports suggest that complications of UVC insertion could occur even in situations where chest radiographs had confirmed optimal position of the UVC.11 In view of this phenomenon, attempts have been made to utilize other modalities to confirm UVC tip position. Ades et al. showed that there was poor correlation between the thoracic level by AP chest radiograph and catheter tip location by echocardiography (ECHO).8 Indeed, if ECHO was demonstrated to be superior to chest radiographs for the localization of the UVC tip, in order for it to be
performed at the bedside following UVC placement, the technique would have to be mastered by frontline neonatal clinicians who would be able to perform it round-the-clock without the requirement of pediatric cardiologists. We hypothesized that targeted neonatal echocardiography (TnECHO) is superior to AP chest radiographs and that with basic training, frontline neonatal clinicians including pediatric housestaff physicians would be able to accurately localize the UVC tip in order to make the technique feasible. The aims of this study, therefore, were twofold: to compare TnECHO, specifically for UVC tip localization, to AP chest radiographs and to determine the accuracy of housestaff physicians’ interpretation of TnECHO for UVC tip localization in comparison with the pediatric cardiologist.
MATERIALS AND METHODS Study design This prospective, observational study enrolled all infants, admitted to the Neonatal Intensive Care Unit (NICU) at New York Methodist Hospital between December 2009 and June 2011, who required placement of a UVC as part of their standard clinical care. Informed consent was obtained from the parents of the participants. Neonates with major congenital anomalies and those with the UVC at a different location other than the inferior vena cava (IVC; for example, hepatic vein), diagnosed by the initial radiograph, were excluded. Neonates whose parents refused consent were
1 Division of Neonatology, Department of Pediatrics, Women and Infants Hospital, Alpert Medical School of Brown University, Providence, RI, USA; 2Division of Neonatology and Pediatric Cardiology, Department of Pediatrics, New York Methodist Hospital, New York, NY, USA and 3Division of Pediatric Critical Care Medicine, Department of Pediatrics, Comer Children’s, Hospital, University of Chicago, Chicago, IL, USA. Correspondence: Dr AM Nadroo, Department of Clinical Pediatrics, Weill Cornell Medical College or Division of Neonatology and Pediatric Cardiology, Department of Pediatrics, New York Methodist Hospital, 506, 6th Street, Brooklyn, New York, NY 11215, USA. E-mail:
[email protected]. Received 26 March 2013; revised 24 June 2013; accepted 25 June 2013; published online 22 August 2013
TnECHO for UVC tip position AS Pulickal et al
951 also excluded from the study. This study was approved by the Institutional Review Board of New York Methodist Hospital.
Study techniques The UVC was inserted by a neonatology provider using standard clinical techniques. The length of the UVC insertion was calculated as [(0.5 UAC catheter insertion length) þ 1 cm]. UAC insertion length was calculated using the formula [(3 Birth weight (in kg)) þ 9 cm] as proposed by Shukla and Ferrara.10 Data collected included gestational age, birth weight, length, shoulder–umbilical length, reason for UVC insertion and admitting diagnosis. After UVC insertion, an AP chest radiograph was obtained and based on preliminary interpretation by the neonatologist, the catheter was repositioned, if necessary. A repeat chest radiograph was performed if the catheter was repositioned. The neonatologist inserting the catheter was blind to the subsequent TnECHO results. TnECHO was then performed by a pediatric housestaff physician and a pediatric cardiologist independently within 24 h of catheter placement without any further changes made to the UVC position. Catheter length at the umbilical cord edge was noted before each radiograph and TnECHO in order to confirm non-migration of the catheter in either direction. The same cardiologist performed and interpreted all TnECHO reports in the study. The cardiologist’s report on UVC tip location was considered authoritative. Chest radiographs performed in the closest temporal relation to the TnECHO studies were chosen for analysis. Chest radiographs reporting optimal position were compared with the reports of the TnECHO as interpreted by the pediatric cardiologist. Landmarks used for the catheter tip location on chest radiograph were relative to the right cardiac margin, the diaphragm, the superior vena cava, the pulmonary hilum and the vertebral body. Ideal position of the UVC was defined as the catheter-tip being visible between the 9th and 10th thoracic vertebrae on a chest radiograph and at the level of diaphragm. The position of the UVC was considered too high if the tip of the catheter was in the right atrium or beyond or higher than the 9th thoracic vertebra and too low if the tip was below the 10th vertebra.
Training protocol TnECHO was performed by one of the three pediatric housestaff physicians who had undergone a two-tier training protocol before participation in the study. In the first phase, the pediatric cardiologist led individual sessions for each of the housestaff physicians, each lasting approximately 2 h, during which the principles of ultrasound techniques including ECHO were explained. Subsequently, in the same session, the cardiologist demonstrated the two views (subxiphoid and parasternal short axis) followed by hands-on training for the housestaff physician. Performance of TnECHO in that session did not count toward the housestaff physician’s total number of TnECHO required for participation in the study. In the second phase of the study, the physician performed TnECHO on newborn infants, both term and preterm, under the direct supervision of the pediatric cardiologist on different days. Accurate identification of the anatomic landmarks necessary for the conduct of the study (liver, IVC, IVC–right atrium junction and four
cardiac chambers), as certified by the cardiologist, on five consecutive neonates was the competency outcome required for the housestaff physician to participate in the study. The average time period required for the housestaff physicians to achieve the competency was 2 weeks.
TnECHO protocol Two-dimensional TnECHO was performed using the Philips 5500 ECHO machine (Philips, Andover, MA, USA). For UVC tip localization, both the parasternal short axis and subxiphoid views were obtained and recorded on an optical disc. In order to enhance the feasibility of housestaff physicians’ successful performance of TnECHO for UVC localization, the study was designed to use the minimum number of echocardiographic views. With this in consideration, it was determined that the combination of the subxiphoid and parasternal short axis views would be optimal for the conduct of the study. Absence of the UVC tip in the right/left atrium on both views, with or without visualization of the same in the IVC was thought to negate the possibility of a ‘high’ position (that is, in the right atrium/left atrium). TnECHO was performed after the standard AP chest radiograph was done. Both the cardiologist and the housestaff physician were blind to the AP chest radiograph results at the time of performing and interpreting TnECHO results. In addition, both housestaff physicians and the cardiologist were blind to each other’s TnECHO interpretations. Readjustment of the UVC was done by the neonatology provider if the TnECHO performed by the pediatric cardiologist demonstrated sub-optimal position.
Statistical analyses Descriptive statistics including simple correlation was performed to interpret the accuracy of AP chest radiographs in comparison to the cardiologist-reported TnECHO results. Comparative characteristics to include sensitivity, specificity, positive predictive value and negative predictive value were calculated for the housestaff physicians’ results with the cardiologist’s results as the comparator standard. Receiver operating characteristic (ROC), which is a two-dimensional graph in which the true positive rates of an operator as compared with a gold standard are plotted on the y axis and false positive rates on the x axis, was constructed to determine to describe the validity of the pediatric housestaff physicians’ interpretation of TnECHO with that of the cardiologist. In addition, the area under the ROC curve, which is a statistical method to reduce the ROC performance to a single scalar value representing expected performance in the case of binary variable comparisons between two operators (optimal vs suboptimal UVC position between the cardiologist vs housestaff physicians), was also determined.
RESULTS Patient demographics Thirty neonates were enrolled during the study period. The most common diagnoses in patients with UVC insertion were
Figure 1. (a) Targeted neonatal echocardiography of a study participant demonstrating the umbilical venous catheter (UVC) tip location in the left atrium. Concurrence of results between the housestaff physician and pediatric cardiologist noted in this case. (b) Closest temporal antero-posterior chest radiograph of the same participant in a with UVC tip reported as being in optimal position at the level of the diaphragm. LA, left atrium; RA, right atrium. & 2013 Nature America, Inc.
Journal of Perinatology (2013), 950 – 953
TnECHO for UVC tip position AS Pulickal et al
952
Figure 2. Targeted neonatal echocardiography (TnECHO) reports of pediatric residents and pediatric cardiologist for all participants in the study demonstrating concurrence of umbilical venous catheter (UVC) tip localization in 23 of 30 study cases (77%). In the seven cases with discordant results, there were only two cases (participants 5 and 7, 6% of study population) wherein there was a large discordance (inferior vena cava (IVC) position vs left atrium (LA)) between the two TnECHO reports. y Axis legend: IVC, right atrium (RA), LA.
prematurity and respiratory distress syndrome. Gestational ages and birth weight of the participants ranged from 24 to 44 weeks and 270 g to 4490 g, respectively. TnECHO results Among the UVC tips that were reported in the optimal position on chest radiographs, 9 out of 30 (27%) were malpositioned on TnECHO performed by the cardiologist and needed to be repositioned. Four were localized in the right atrium and five in the left atrium. Figure 1a demonstrates a malpositioned UVC tip in the left atrium of a patient in whom the AP chest radiograph was reported as normal (Figure 1b). Training results Compared with the cardiologist, the housestaff physicians had accurately reported 23 out of 30 ECHO studies (77%, Figure 2). Among the seven that were not accurately reported (23%), six were reported by the housestaff physicians as having the UVC tip positioned in the right atrium in contrast to its location in the IVC as per the cardiologist. Only one housestaff physician report suggested the UVC tip to be in the left atrium instead of the right atrium as reported by the cardiologist. Comparative analysis revealed a sensitivity rate of 86%, specificity rate of 75%, positive predictive value of 90% and a negative predictive value of 67% based on the housestaff physicians’ ability to distinguish on TnECHO, a malpositioned catheter from an optimally positioned catheter. Similarly, the ROC curve suggested high housestaff physician accuracy rates with area under the ROC curve ¼ 0.81 (Figure 3). DISCUSSION In this study, we demonstrate that TnECHO for UVC tip localization is superior to AP chest radiographs. To our knowledge, this is the first study that determined the accuracy of TnECHO performance by front-line neonatal clinicians for the purpose of UVC tip localization. UVC placement has become one of the more commonly performed techniques in the NICU as the first trial of polyethylene tubing as an UVC in a newborn with erythroblastosis was done in the Boston Lying-In Hospital in 1946.12 Despite AP chest radiograph having been the consistent gold standard for determining UVC tip position, earlier studies had suggested that this technique might be suboptimal. Journal of Perinatology (2013), 950 – 953
Figure 3. Receiver operating characteristic (ROC) for housestaff physician-performed targeted neonatal echocardiography (TnECHO) compared with cardiologist-performed TnECHO as gold standard demonstrating high accuracy with an area under the ROC curve ¼ 0.81.
In 1995, Greenberg et al. showed that among 60 UVCs placed with radiological guidance, 36 (56%) were repositioned because of unsatisfactory location, thus requiring more than one radiograph.13 All catheters at T6 and above were in the heart; 18 (58%) of 31 catheters were located in the left atrium. At T7, 14 (82%) of 17 catheters were incorrectly positioned with the tip in the right atrium. At T8–T9, 34 (90%) of 38 catheters were correctly positioned at the right atrium–IVC junction, 100% at T9. The catheters at T11 and below were all in the liver proximal to the ductus venosus. George et al. showed that in only 6 of 15 observations did the radiographic location correspond with the actual position confirmed by contrast ECHO; in 6 patients, radiographic position was erroneous and in 3 others, the positions were unclear.14 According to Reval et al., the detection of malpositioned umbilical catheter by radiography had a sensitivity of 45%, specificity of 87%, a positive predictive value of 77%, negative predictive value of 66% and an accuracy of 66%.9 Ades et al. showed that there was poor correlation between the thoracic level by AP chest radiograph and catheter location by ECHO.8 Correct positioning of the UVC tip is critical in order to prevent the rare, but life-threatening, complications of UVC insertion. Nadroo suggested that catheters should not be placed & 2013 Nature America, Inc.
TnECHO for UVC tip position AS Pulickal et al
953 in the right atrium because of a risk of pericardial effusion and cardiac tamponade.15 Most cases of catheter-related pericardial effusion, cardiac tamponade or sudden death were secondary to initial placement of the catheter tip in the right atrium, or migration from the superior or IVC.16 It is in the context of these findings that alternate modalities to assure optimal localization of the UVC tip needed to be studied and used. Our study confirms that TnECHO is superior to AP chest radiographs for the purpose of UVC tip localization. Among the study population in which the AP chest radiograph reported the UVC to be in optimal position, more than a quarter of the patients had their UVC readjusted because of suboptimal position on TnECHO. These data suggest that TnECHO be preferred to chest radiographs for UVC tip localization. Previous studies have suggested that using TnECHO for UVC tip positioning also offers the advantages of it being performed in real-time enabling optimal and efficient readjustment of the catheter and the prevention of neonatal exposure to radiation by avoiding the need for multiple chest radiographs for tip localization. More importantly, optimal tip positioning using TnECHO could lead to a reduction in the number of significant complications associated with UVCs. Pediatric cardiology teams perform ECHO in most units. If this technique has to be routinely used for UVC tip localization in the NICU, it must be feasible that a front line clinician with optimal training should be able to perform the TnECHO with reasonable accuracy. Our study provides pilot data on the accuracy of a cohort of neonatal providers other than a cardiologist, who underwent a competence-defined training protocol in order to test the feasibility of this approach in the NICU. In this study, with a concise training protocol, pediatric housestaff physicians were able to interpret UVC tip position with high-accuracy rates as compared with the cardiologist. Based on these encouraging data, we speculate that with improved composite experience, involving hands-on training and experiential practice, frontline neonatal clinicians would be able to perform and interpret TnECHO for UVC tip localization with higher accuracy. In addition, it is anticipated that cost-effectiveness and health benefits, including decreased rates of complications and reduced radiation exposure, would be important aspects of a larger prospective safety. The feasibility of the participation of a more representative neonatal provider group including neonatal nurse practitioners, fellows and attendings besides pediatric housestaff physicians could be tested in such a study. In conclusion, we establish that TnECHO is superior to AP chest radiography for UVC tip localization and demonstrate the feasibility of a neonatal clinician performance model of TnECHO with high accuracy, following a competence defined, concise training protocol.
& 2013 Nature America, Inc.
CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGEMENTS We thank the study participants and their parents. We acknowledge Richard Tucker, Department of Pediatrics, Women and Infants Hospital of Rhode Island for the kind assistance with the statistical analyses in the study and Mandy Zhao, ECHO Technologist, Department of Pediatric Cardiology, New York Methodist Hospital, for her technical assistance with the study.
REFERENCES 1 Megha M, Jain N, Pillai R. Pericardial tamponade in a newborn following umbilical catheter insertion. Ind Pediatr 2011; 48: 404–405. 2 Alabsi S. Neonatal cardiac tamponade and pleural effusion resolved with chest tube placement. Neonatal Netw 2010; 29: 347–351. 3 Sehgal A, Cook V, Dunn M. Pericardial effusion associated with an appropriately placed umbilical venous catheter. J Perinatol 2007; 27: 317–319. 4 Traen M, Schepens E, Laroche S, van Overmeire B. Cardiac tamponade and pericardial effusion due to venous umbilical catheterization. Acta Paediatr 2005; 94: 626–628. 5 Van Niekerk M, Kalis NN, Van der Merwe PL. Cardiac tamponade following umbilical vein catheterization in a neonate. S Afr Med J 1998; 88(Suppl 2): C87–C90. 6 Gonza´lez Die´guez CC, Ba´rrios Alonso V, Torrealday Taboada H, Tamariz Martel A, Hellı´n Sanz T, Brito Pe´rez JM et al. Intracardiac mycetoma induced by central catheterization. An Esp Pediatr 1992; 37: 63–65. 7 Rastogi A, Luken JA, Pildes RS, Chrystof D, LaBranche F. Endocarditis in neonatal intensive care unit. Pediatr Cardiol 1993; 14: 183–186. 8 Ades A, Sabble C, Cummings S, Cross R, Martin G. Echocardiographic evaluation of umbilical venous catheter placement. J Perinat 2003; 23: 24–28. 9 Reval CR, Gonzalea E, Bhat AM, Pearlman SA, Stefano JL. Umbilical venous catheter: evaluation of radiographs to determine position and associated complications of malpositioned umbilical venous catheters. Am J Perinat 1995; 12: 201–204. 10 Shukla H, Ferrara A. Rapid estimation of insertion length of umbilical catheters in newborn. Am J Dis Child 1986; 140: 786–788. 11 Chatel-Meijer MP, Roques-Gineste M, Fries F, Bloom MC, Laborie S, Lelong-Tissier MV et al. Cardiac Tamponade secondary to umbilical venous catheterization accident in premature infant. Arch Fr Pediatr 1992; 49: 373–376. 12 Diamond LK, Allen FH, Wo T. Erythroblastosis fetalis: treatment with exchange transfusion. N Engl J Med 1951; 244: 39–49. 13 Greenberg M, Movahed H, Peterson B, Bejar R. Placement of umbilical venous catheters with use of bedside real time ultrasonography. J Pediat 1995; 126: 633–635. 14 George LJ, Deane W, Kirkpatrick E, Searle W, Turner SJ, Peppelbaum A. Umbilical vascular catheters: localization by two-dimensional echocardio/aortography. Pediatr Cardiol 1982; 2: 237–243. 15 Nadroo AM. Death as a complication of peripherally inserted central catheter in neonates. J Pediat 2001; 138: 599–601. 16 Nadroo AM, Roland BG, Jing L, Robert SG, Holzman LR. Change in upper extremity position caused migration of peripherally inserted central catheter in neonates. Pediatrics 2002; 110: 131–136.
Journal of Perinatology (2013), 950 – 953