Role of a radiopaque agent and surveillance radiographs for peripherally inserted central catheters in newborn infants

PICC insertion and maintenance was practised as per the unit’s guideline. PICC use was reserved for infants requiring total parenteral nutrition or prolonged (>7 days) parenteral antibiotic therapy. In our unit, three types of polyurethane PICCs were available for use: single lumen Premicath® 1Fr (28G), single lumen Nutriline® 2Fr (24G) and double lumen Twinflo® 2Fr (24G) (all three from Vygon Gesellschaft mit beschränkter Haftung & Compagnie Kommanditgesellschaft, Aachen, Germany). For infants  <1,000 g at birth, a 1-Fr PICC was used. All PICCs were inserted by an operator skilled in the technique (either a neonatal nurse practitioner, senior registrar, neonatal fellow or consultant). Where possible, two team members were present for the PICC insertion procedure; one was the primary proceduralist and the second assisted with the procedure. PICCs were inserted through either the basilic or cephalic vein into the superior vena cava or through the long saphenous vein into the inferior vena cava. External anatomical landmarks were used to determine the length of the PICC insertion. For the upper limb, this length was the distance in centimetres between the insertion site and the jugular notch of the sternum. For the lower limb, the length in centimetres was the distance between the insertion site and the xiphisternum [17]. Excess catheter beyond the insertion site was looped over the skin, ensuring that the catheter was not kinked. The PICC was secured to the skin using Steri-Strip™ adhesive tapes (3 M Company, St Paul, MN). A clear semi-permeable dressing was placed over the insertion site.

For most patients, the PICC tip position was then confirmed by injecting a small amount (0.2 mL to 0.3 mL) of a radiopaque agent (Omnipaque 240, Iohexol equivalent to iodine 240 mg/mL, GE Healthcare, Chicago, IL) using a 10 mL syringe (as per the manufacturer’s instructions), while simultaneously taking the radiograph. After taking the radiograph, the radiopaque agent was flushed into the patient. The dead space for these PICC lines ranges from 0.09 mL (in 20 cm Premicath®) to 0.2 mL (in 30 cm Twinflo®). The decision to use a radiopaque agent for PICC tip location was not based on random allocation but at the discretion of different providers. For this study (as shown in Figs. 2 and 3), a central ideal position of the PICC tip (in a central vein, but outside the cardiac silhouette) was defined as follows: (i) for PICCs inserted in the arm: PICC tip between the 3rd and the 5th thoracic vertebrae, which corresponds to the superior vena cava and the cava-atrial junction and (ii) for PICCs inserted in the leg: PICC tip between the 8th and the 12th thoracic vertebrae, which corresponds to the inferior vena cava. Where the PICC tip was between the medial third of the clavicle and 2nd thoracic vertebra (subclavian or brachiocephalic vein) or between the 1st and the 3rd lumbar vertebrae, this was referred to as a central acceptable position. Both central ideal and central acceptable positions were considered acceptable regardless of the infusate to be used (usually we administer total parenteral nutrition or antibiotics through the PICCs). All positions of the PICC tip beyond the central ideal or central acceptable were considered as non-central in position. This also included location of the PICC tip in a peripheral vein. After PICC insertion, in order to ascertain PICC migration, weekly surveillance radiographs were performed for all infants, without using a radiopaque agent. Radiographs on the day of insertion were labelled as T0, a week after insertion as T1 and so on. Radiographic images obtained at the time of PICC insertion and for weekly surveillance for all patients were reported by the institution’s radiologist on service. The radiographs were acquired using MobileDaRt X-ray System (Shimadzu Corporation, Kyoto, Japan) and the images were accessed using Sectra UniView® web viewer (Sectra AB, Linkoping, Sweden). In Sectra UniView, the quality of images was optimized by using zoom in, invert and contrast functions.

The outcomes were assessed with univariate and multivariate generalized estimating equations. The models for first outcome were adjusted for potentially confounding variables such as gestational age, birth weight, limb of insertion and side of insertion. The models for second outcome were adjusted for fixed effects of day of radiograph, side, limb, year of birth and interaction term between side and limb. Odds ratios (OR), 95% confidence intervals (CI) and P-values are reported. The models in the outcomes were additionally adjusted for an interaction parameter between side and limb. The goodness-of-fit of the interaction parameter was assessed by a Wald test, with P-values calculated by a Monte Carlo simulation, and measured by the quasi-likelihood information criterion. Throughout, generalized estimating equations handled missing data automatically by adjusting appropriate variance and covariance estimates. A pragmatic study duration of 5 years was chosen for sample size calculation based on the NICU’s admission rate.

We identified 601 infants who were born between 1 January 2016 and 31 December 2020. The median gestational age of the cohort was 29 weeks (IQR 27–31) and median birthweight was 1,200 g (IQR 960–1,455). Male infants represented 54% (326/601). Over the duration of the study, we screened 7,278 infants who were admitted to the NICU for their eligibility. Of these, 6,677 infants were excluded: 6,588 infants did not need a PICC and 89 infants either had a PICC insertion at another hospital and were transferred to our unit or PICC was inserted in our unit, but these infants were transferred to another hospital before PICC removal. For the final analysis, we included 676 PICC insertion episodes in 601 infants. The most common indication for PICC insertion (546/676, 80.8%) was for the administration of parenteral nutrition in view of prematurity. A single lumen PICC (550/676, 81.4%) and a 1-Fr catheter (481/676, 71.2%) were most commonly used. The median PICC in situ time was 9 days (IQR 6–14). Of the 676 PICC episodes, 143 (21.2%) were placed in the arm, 532 (78.7%) in the leg and one in a scalp vein. Of the 143 PICCs inserted in the upper limb, 86 (60.1%) were inserted in the right arm (basilic or cephalic vein in the cubital fossa) and 57 (39.9%) in the left arm. Of the 532 PICCs inserted in the leg, 401 (75.4%) were placed in the right leg and 131 (24.6%) in the left leg. For the PICC tip positions, 563/675 (83%) were central. Of the central PICC tip positions, 103/563 (18.3%) were inserted in the arm and 460/563 (81.7%) were inserted in the leg.

The cohort was dichotomized based on the use of a radiopaque agent. Of the 676 PICC insertion episodes, one episode was excluded because the PICC tip position could not be confirmed as it was obscured by a monitoring lead. Of the 675 PICC insertion episodes, a radiopaque agent was used for 590 (87%). Most baseline clinical characteristics were similar between the two groups (Table 1) except for birthweight and gestational age. There was no difference between the proportion of central PICC tip position (central ideal and central acceptable) based on radiopaque agent use status (radiopaque agent used group: 83.1% compared to no radiopaque agent used group: 85.9%, P=0.51). For non-central lower limb PICC positions, using a radiopaque agent identified PICC tip in the ascending lumbar vein (6/72, 8.3%) (Fig. 3). For all infants where a radiopaque agent was not used, the PICC tip position could be easily identified on the radiographs irrespective of central or non-central position and PICC caliber (Fig. 4). Significantly more radiographs were required in the radiopaque agent group to confirm the PICC tip position compared to no radiopaque agent used group (Table 1). In these cases, the radiopaque agent jet was seen beyond the PICC tip (Fig. 5) and these PICCs required repositioning to optimize their location.

Binary univariate and multivariable logistic regression models were performed for the clinical variables associated with PICC tip position. The birth year and limb used for PICC insertion when adjusted for gestational age and birthweight were the two significant variables. The OR for a central position of the PICC tip was higher for PICCs inserted in 2020 compared to 2016 (adjusted [a]OR 1.49, 95% CI 1.26–1.76). The OR for a central position (central ideal and central acceptable) of the PICC tip was lower for left arm (aOR 0.39, 95% CI 0.19–0.76), right arm (aOR 0.37, 95% CI 0.21–0.66) and left leg (aOR 0.55, 95% CI 0.32–0.95) compared to right leg (arbitrarily chosen as the reference) after adjusting for gestational age, birthweight and infant’s birth year.

For the second aim of our study, from the 676 PICC insertion episodes, one was excluded as it was inserted in a scalp vein. Additionally, 32 PICC insertion episodes were excluded as a repeat radiograph was not performed after PICC repositioning, therefore, their true baseline position at T0 could not be verified. A total of 643 PICC episodes were included for the final analysis. There were 359, 123, 38 and 16 PICC lines screened on day 7 (T1), day 14 (T2), day 21 (T3) and day 28 (T4), respectively. The trend in PICC migration over time based on side and limb of PICC insertion is shown in Fig. 6. Most PICCs, irrespective of their insertion sites, migrated over the in situ duration. Inward migration was identified in 23 out of 643 PICC episodes (3.6%) at day 7 (T1). Of these, 18 PICCs were inserted in the arm and only five of them required repositioning. It was also noted that in 11 out of 18 (61.1%) PICCs the arms were lifted above the head on the follow-up radiographs, therefore likely contributing to an inward PICC line migration. No PICCs migrated inwards beyond day 7. As shown in Fig. 6, PICCs migrated outward over the in situ duration. If the PICC tip was identified to be in a non-central position on follow-up radiographs, its use was at the discretion of the attending physician. Where the PICC was left in situ, it was closely monitored for the occurrence of complications. PICC tip could be identified on all weekly surveillance radiographs, whether a radiopaque agent was used or not at the time of PICC insertion. The adjusted OR for PICC tip remaining in a central position were lower the longer the PICC remained in situ (adjusted OR 0.93; 95% CI 0.92–0.95). There was no difference in PICC migration between left- and right-sided PICCs (independent of the limb of insertion). PICCs in the leg had greater odds of being centrally located compared to PICCs in the arms (aOR 1.20, 95% CI 1.13–1.26). We examined the interaction between the side and limb of PICC insertion and found that PICCs inserted in the left leg had a lower probability of being centrally located compared to PICCs inserted in the right leg (aOR 0.91, 95% CI 0.83–0.99). PICCs inserted in 2020 had a higher probability of remaining in a central position compared to PICCs inserted in 2016 (aOR 1.21, 95% CI 1.16–1.26). The year of birth (insertion) impacted the position of the PICC tip independent of in situ time.