Discussion

The primary finding of our study is that our cohort of neonates and children with congenital or acquired heart disease had an overall incidence of successful peripherally inserted central catheter placement of 94.5%. Our success rate is similar to previously reported paediatric literature (95–100%). ^{Reference Gonzalez and Cassaro1,Reference Conlon, Himebauch and Cahill5} General anaesthesia use during the insertion and year of peripherally inserted central catheter placement seemed to be significant contributors to this success. Our secondary finding is that the incidence of line complications in our cohort reached 28.6%. Previous paediatric studies noted broad complication rates ranging from 2.2 to 31.3%. ^{Reference Gonzalez and Cassaro1,Reference Jain, Deshpande and Shah6,Reference Tan, Hess and Diaz7}

While studies note that physician experience increases procedure success and decreases complication odds, only a few examine this specifically as an effect of time and quantify complication rate. ^{Reference Alexandrou, Spencer, Frost, Mifflin, Davidson and Hillman8,Reference Costello, Morrow, Graham, Potter-Bynoe, Sandora and Laussen9} While one recent paediatric study noted decreased peripherally inserted central catheter complications from 22 to 8% over 6 years, this study did not focus on paediatric cardiac patients. ^{Reference Alexandrou, Spencer, Frost, Mifflin, Davidson and Hillman8} In our study, we found increasing rates of successful line placement and decreasing infection rates with each subsequent year. However, some complications, such as migration/malposition and dislodgement, continued to rise over time. This increase could be attributable to the vascular complexity of patients requiring long-term usage of these lines. Sedation and analgesia are often used in paediatric patients to place peripherally inserted central catheters, especially in the younger population. ^{Reference Westergaard, Classen and Walther-Larsen2,Reference Pitts10} One study noted that successful placement significantly increased in children under the age of 5 years with the use of sedation and analgesia for placement; however, only 7.3% were cardiac patients. ^{Reference Pitts10} Uniquely, we found that general anaesthesia, unlike other sedation methods, was significantly associated with successful line placement in our cohort. This differs from a study that found no difference, regardless of analgesic modality. ^{Reference Costa, Bueno, Oliva, de Castro, de Camargo and Kimura11} However, the use of general anaesthesia for line placement in paediatric cardiac patients must be individually evaluated given the high risks for life-threatening complications with anaesthesia in this patient population. ^{Reference Laing, Burgoyne, Muncaster, Taranath and Taverner12}

In our study, thrombosis and occlusion occurred in 9.45% of all peripherally inserted central catheters (i.e. accounted for 33% of the complications), with about one-third of those requiring line removal. Previous paediatric studies demonstrated peripherally inserted central catheter occlusion secondary to thrombotic and non-thrombotic origin to occur at a rate of 1.35 per 1000 catheter days. ^{Reference Shah, Kalyn and Satodia13–Reference Fratino, Molinari and Parodi15} Reported risk factors included catheter tip position (i.e. increased thrombosis risk in superior portion of superior vena cava, as compared to inferior portion and right atrium), small diameter veins, malignancy, line characteristics (e.g. triple lumen, narrow lumen), and type of medication (e.g. calcium, heparin, total parenteral nutrition, etc.). ^{Reference Westergaard, Classen and Walther-Larsen2,Reference Liem, Yanit and Moseley14,Reference Fratino, Molinari and Parodi15} Our study demonstrated thrombosis/occlusion to be significantly associated with an increased number of lumens. Mechanisms explaining this finding include more manipulation, leading to increased bacterial seeding and decreased blood flow, leading to thrombosis and occlusion from mechanical and inflammatory aetiologies. ^{Reference Baxi, Shuman and Scipione16–Reference Chopra, Fallouh and McGuirk18} Multiple lumen lines are also most often required and utilised in patients with more complicated diseases, and this might represent a population at higher risk for overall complications. Most thrombotic events are asymptomatic, so symptom onset is not a reliable way of detecting venous thromboembolism. ^{Reference Dubois, Rypens, Garel, David, Lacroix and Gauvin19} Incidence of venous thromboembolism associated with paediatric peripherally inserted central catheters ranges from 2.6 to 9%. ^{Reference Westergaard, Classen and Walther-Larsen2,Reference Dubois, Rypens, Garel, David, Lacroix and Gauvin19} In our cohort, 28 of the 279 thrombotic events (10%) had venous thromboembolism and required anticoagulation (n = 5, 17.9%), thrombolysis (n = 10, 35.7%), or line removal (n = 13, 46.4%).

Neonatal studies have reported rates of peripherally inserted central catheter infiltration ranging from 4.7 to 12.5%. ^{Reference Bashir, Swarnam, Vayalthrikkovil, Yee and Soraisham3,Reference Jain, Deshpande and Shah6} These studies demonstrated that infiltration was associated with peripheral catheter position and upper extremity placement. ^{Reference Bashir, Swarnam, Vayalthrikkovil, Yee and Soraisham3,Reference Jain, Deshpande and Shah6} In our study, infiltration occurred in 5.2% of all peripherally inserted central catheters (i.e. accounted for 18.5% of the complications), requiring line removal in about one-third of those events. We also found significantly increased infiltration with peripheral tip location and with the infusion of maintenance intravenous fluids. This could be due to fluid composition and high-flow infusion rates. ^{Reference Park, Jeong and Jun20,Reference Singh, Kalyan, Kaur, Jayashree and Ghai21} Furthermore, our study demonstrated significantly decreased infiltration when lines were placed by the interventional radiology team, likely due to the more precise positioning under fluoroscopy.

Catheter malposition can be defined as primary (e.g. initial malposition secondary to patient moving or vascular anomalies, etc.) and secondary or migration (e.g. despite correct initial placement, catheter migrates secondary to rapid infusion rates, increased physical exertion, or rapid ventilation, etc.), both with a functional line. ^{Reference Gonzalez and Cassaro1} This differs from dislodgement (i.e. accidental line movement or removal resulting in a non-functional catheter). ^{Reference Gonzalez and Cassaro1} Dangerous sequela related to malposition or dislodgement can include cardiac tamponade, arrhythmias, arterial puncture, hemothorax, pneumothorax, etc. ^{Reference Westergaard, Classen and Walther-Larsen2} In our cohort, migration/malposition occurred in 3.5% of the peripherally inserted central catheters (i.e. accounted for 12.4% of complications), with 42% of these patients requiring line removal. We found decreased migration/malposition when lines were placed in the interventional radiology suite, likely due to the frequent use of sedation to prevent patient movement and fluoroscopy for precise positioning in this location. Interestingly, a recent study examining peripherally inserted central catheter placement at the bedside with ultrasound compared to the interventional radiology suite found no difference in success or complications but found a significantly faster placement and less cost when placed at the bedside. ^{Reference Conlon, Himebauch and Cahill5} A recent study also found decreased central line-associated bloodstream infections rates and odds of occurring when peripherally inserted central catheters were placed in the interventional radiology suite compared to other hospital locations. ^{Reference Koo, Cooper, Monroe, Reis, Shivaram and Zerr22} Peripherally inserted central catheter dislodgement has reported rates varying from 2.4 to 17.5% and seems to be associated with age < 5 years old, increased activity level, securement with tape, and improper tip positioning. ^{Reference Tan, Hess and Diaz7,Reference Pittiruti, Hamilton, Biffi, MacFie and Pertkiewicz23} In our cohort, dislodgement occurred in 5.8% of the lines (i.e. 20.4% of the complications) and required line removal in more than half of the cases. Dislodgement was significantly associated with the use of total parenteral nutrition in our study. Dislodgement with total parenteral nutrition has not been described and its biological plausibility is unclear. Possible mechanisms for this association are the use of inappropriate high pressures to flush the line due to intraluminal precipitate of lipid aggregates, inappropriate securing at the time of insertion or inadequate care of the exit site in these patients. ^{Reference Pittiruti, Hamilton, Biffi, MacFie and Pertkiewicz23}

Central line-associated bloodstream infections from peripherally inserted central catheters are most often from coagulase-negative Staphylococci, Staphylococcus aureus, enterococci, and Candida species, with infections manifesting from site cellulitis to systemic infection. ^{Reference Gonzalez and Cassaro1,Reference Westergaard, Classen and Walther-Larsen2} The reported incidence rates of these catheter infections vary from 0.2 to 6.4 per 1000 catheter days. ^{Reference Westergaard, Classen and Walther-Larsen2,Reference Shah, Kalyn and Satodia13} In our study, infection occurred in 2.9% of peripherally inserted central catheters (i.e. accounted for 10.4% of the complications), with gram-positive cocci being the most frequent microorganism group isolated. Only 20.9% of these infections were cleared with antimicrobials (repeat blood cultures yielded negative results, along with clinical and laboratory resolution of the infection), with the remaining 79.5% requiring catheter removal. The peripherally inserted central catheter placement-to-infection time has been found to be 21 days in regular paediatric age groups and 35 days in neonates. ^{Reference Barrier, Williams, Connelly and Creech24–Reference Sengupta, Lehmann, Diener-West, Perl and Milstone27} Median placement-to-infection time in our cohort was shorter, 8 days (interquartile range 5–18); however, we found 21% decreased odds of infection with each additional year of line placement. In the literature, factors associated with peripherally inserted central catheter infection are young age, smaller-sized and double-lumen catheters, polyurethane composition, femoral vein access site, incorrect tip positioning, multiple catheters, and a greater number of daily doses of medication administered. ^{Reference Barrier, Williams, Connelly and Creech24,Reference Bourgeois, Lamagna and Chiang28} We examined the effect of age, size and lumens, access site, and tip position and did not find statistical significance.

There is scant data on peripherally inserted central catheter damage in the literature, with incidence ranging from 0.6 to 1.9%. ^{Reference Chow, Friedman and Macarthur29,Reference Paulson and Miller30} Some reports have demonstrated peripherally inserted central catheter damage due to excessive syringe pressure when giving medications, traction on the catheter-hub junction, and iatrogenic aetiologies. ^{Reference Pittiruti, Hamilton, Biffi, MacFie and Pertkiewicz23,Reference Bashir, Bhat, Manzoor, Bashir and Ahmad31,Reference Graham, Keldermans, Klemm, Semenza and Shafer32} In our study, damage occurred in 1.5% of all peripherally inserted central catheters (i.e. accounted for 5.3% of the complications) and was significantly associated with the use of sedation and analgesia. There is not published data on this association or knowledge of the mechanism explaining the cause-effect. Chemical damage to the line material due to the use of solvents, or the use of medication boluses with inappropriate high pressures on the line wall could explain this finding. ^{Reference Pittiruti, Hamilton, Biffi, MacFie and Pertkiewicz23} Only one-third of those damaged lines were successfully repaired, while the remaining ones required removal.

Our study has a few limitations. First, it is a retrospective database review of a single-center experience with limited data including lack of primary cardiac diagnosis, so the results cannot be extrapolated to all paediatric cardiac patients. Second, we could not adjust for factors associated with peripherally inserted central catheter complications such as the severity of illness, comorbidities, and diagnosis. Third, the absence of standardised line placement and monitoring guidelines in our patients may be an important confounder. Finally, some of the identified associations may not have clinical plausibility.

In conclusion, our study demonstrated that peripherally inserted central catheters in paediatric cardiac patients are feasible and safe, with an incidence of complications similar to other paediatric populations. Identifying factors associated with complications prospectively for this population subset will help shed light on ways further to decrease rates of line-related complications and improve successful insertion.