439 - Use of Ultrasound to Determine Accuracy of Common Measurements for Placing Umbilical Venous Catheters in Newborn Infants

Publication Number: 3424.439

Chelsey A. Bithell, Georgetown University School of Medicine, Chevy Chase, MD, United States; Jiling Chou, MedStar Health Research Institute, Columbia, MD, United States; Suhasini UM. Kaushal, Med, Washington DC, DC, United States; Jayashree Ramasethu, MedStar Georgetown University Hospital, Washington, DC, United States

Background: Umbilical venous catheter (UVC) depth is estimated before placement using formulas, including the Shukla ([3×weight(kg)+9]/2+1, Am J Dis Child 1986) or Gupta (umbilicus to nipple distance (cm)-1, J Perinatol 2015) formulas. UVCs often require adjustment for appropriate placement. Studies to determine accuracy of these formulas were based on x-rays (XR). Ultrasound (US) is more accurate at determining UVC tip placement compared to XR, but is not available in all units.

Objective: Compare the accuracy of the current formulas for estimating UVC insertion depth, using US to determine placement confirmation rather than XR.

Design/Methods: UVCs were placed in the NICU, with XR and US performed within 12 hours of placement. Final correct length based on US was compared to the lengths derived by Shukla and Gupta formulas. The formulas were assessed for agreement with US using the Lin's concordance correlation coefficient (CCC) and Bland-Altman analysis for total cohort and birth weight (BW) subgroups. We used non-parametric bootstrap approach to compare the paired CCCs. The sample size required to show a meaningful level of agreement by CCC at 0.05 significance level was 28. One-sided one-sample t-test was applied to compare the accuracy of the two formulas, defining clinically meaningful difference as >0.5 cm.

Results: Twenty-nine infants were enrolled, 8 of whom weighed ≤ 1500g (Table). Figure 1 shows the CCC between the two formulae estimates and US confirmation - there was no significant difference. Shukla CCC was 0.90 (95% CI: 0.81-0.95) and Gupta CCC was 0.82 (0.68-0.90). In subgroups, both formulas demonstrated lower agreement with wide CI: Shukla ≤1500g [0.47(0.01-0.77)] and >1500g [0.84(0.66-0.93)]; Gupta ≤1500g [0.6(0-0.88)] and >1500g [0.7(0.47-0.85)]. The Shukla formula showed an overall mean difference of 0.11cm relative to US (95% limits of agreement (LOA) −1.32-1.54cm), underestimating the ≤1500g subgroup [-0.31(-1.35-0.73)] and overestimating the >1500g subgroup [0.27(-1.18-1.73)] (Figure 2). The Gupta formula overestimated the depth by 0.73cm (-0.79-2.26 cm), with similar results in the >1500g subgroup [0.86(-0.73-2.45)] and a smaller bias in the ≤1500g subgroup [0.41(-0.79-1.6)]. There was no evidence to state a clinically meaningful difference of greater than 0.5cm between the two formulas.

Conclusion(s): We found a similar level of agreement with US for each formula. Interpretation of the results is limited by small subgroup sizes, narrow measurement ranges, and potential aggregation effects. A larger study is needed to more definitively evaluate the accuracy and clinical utility of each formula.

Table: Data summary and accuracy outcomes for total cohort and by birth weight ≤1500g and >1500g subgroups


Figure 1. Scatterplots demonstrating CCCs
Scatterplot of UVC estimated depth by Shukla (1a, 1c) and Gupta (1b, 1d) formula and US confirmation depth, with linear model fitting smoothing line and shaded area indicated 95% confidence interval.

Figure 2. Bland Altman Plots
Bland-Altman plots with mean difference (diff.) and limits of agreement (LOA) between Shukla and US (2a, 2c) and Gupta and US (2b, 2d).