Incorporating anti-infective drugs into peripherally inserted catheters does not reduce infection rates in neonates

Purpose This study assesses whether peripherally inserted central venous catheters (PICC), impregnated with anti-infective drugs, reduce the rate of infections in neonates compared with unimpregnated catheters. Methods A retrospective analysis was conducted on electronic patient records of neonates born between August 2014 and May 2020, who had PICCs inserted, either standard (S-PICC) or with anti-infective drugs (A-PICC). Catheter-related bloodstream infections (CRBSI) were diagnosed based on clinical symptoms, laboratory results, and mentioning of infection in the patient record. Data on dwell time, mechanical ventilation, insertion site, maximum C-reactive protein (CRP) concentration, and anti-infective drug use were analyzed. Results A total of 223 PICCs were included. The infection rates were A-PICC (18.9%) and S-PICC (12.5%), which were not significantly different (p = 0.257). A-PICCs had significantly longer dwell times than S-PICCs (median 372 vs. 219 h, p = 0.004). The time to infection was not different between the groups (p = 0.3). There were also no significant differences in maximum CRP, insertion site abnormalities, or anti-infective drug use between the groups. Conclusion This retrospective study did not find a significant reduction in infection rates by using PICCs containing anti-infective drugs in neonates. Current antibiotic impregnations do not seem to be effective in preventing blood stream infections.


Introduction
Peripherally inserted central venous catheters (PICC) remain an important tool to ensure adequate parenteral nutrition and application of intravenous medications in small preterm infants for bridging the time until full enteral nutrition has been established (1-5).PICCs are introduced into peripheral veins-most commonly Vena saphena, Vena basilica, Vena cephalica, and the veins of the dorsal hand (1, 2, 6).They are advanced until the tip is located in one of the caval veins (2,3,7).Catheter-related blood stream infections (CRBSI) are a common and serious complication (5,8).
PICCs impregnated with the anti-infective drugs rifampicin and miconazole were designed to reduce CRBSI (9).No recommendation has been made on their routine use in neonates due to a lack of evidence (10).Only the Infusion Nurse Society of the USA suggests using them in high-dependency patients (11).Flemmer et al. spoke at the 42nd conference of the German Society for Neonatology and Pediatric Intensive Care about a reduction in bacteriological complications (12).The multicenter randomized PREVAIL trial did not find a significant difference regarding infection rates (13).Similarly, a monocentric retrospective study at Doha (Qatar) including several catheter types did not find a benefit of anti-infective impregnated catheters regarding culture proven bloodstream infections (14).The following study made use of a large database of patient records accumulated at our university hospital to assess infection rates with PICCs impregnated or non-impregnated with antiinfective drugs in the clinical routine.

Setting
This study was conducted at the University of Leipzig Medical Center, a large tertiary care perinatal center in the German state of Saxony.Due to the retrospective nature of this study, neither treatment nor outcomes were influenced by this investigation.Anti-infective drug-incorporated PICCs (A-PICC, Premistar ® , Vygon) were automatically identified from electronic patient records of all patients born between August 2014 and May 2020.This interval was chosen because both types of catheters were inserted during this time.An approximated number of the standard, i.e., drug-free, standard PICCs (S-PICC, Premicath ® , Vygon) were randomly selected.PICC that were started during infection by other endovascular devices were excluded as were ambiguously documented PICC.We analyzed all microbiology studies taken during dwell time of the studied PICCs from catheter tips and blood cultures drawn peripherally.

Definition of CRBSI
Based on the National Health Safety Networks (NHSN) (7) our definition of CRBSI considered three criteria: (i) clinical symptoms, (ii) laboratory results, and (iii) mention of an infection in the patient record (15).
Any PICC was considered infected if two of these three criteria were positive within 7 days (7): (I) Clinical symptoms were considered positive if at least five of the following eight parameters were present (7,15,16)

Data acquisition and statistics
Our hospital's Data Integration Center extracted the data from our electronic patient data management system.Additional data were extracted from the hospital information and quality assurance systems.We used the R software environment version 3.6 (17) with the packages survival (18), survminer (19), and ggpubr (20) for time-to-event analyses.If not stated otherwise, discrete variables were compared with Welch's two sample t-test or Wilcoxon's rank sum test if normality could not be assumed, categorical variables, with either χ² test, or, if the former was inappropriate, Fisher's exact test against the canonical α = 0.05.The results are reported as median [IQR] unless noted otherwise.Data collection and analysis were approved by the institutional review board of the medical faculty of the University of Leipzig (017/20-ek).

Catheter ingrowth
To prevent ingrowth, a slight retraction was performed in 31 PICCs during dwell time; there was no difference between the groups (p = 0.257) or presence of catheter infection (p = 0.287); but PICCs were described as difficult to remove in four cases, all of them had a catheter-related infection (p = 0.013).

Insertion site
Abnormalities, such as redness, at the insertion sites were not different between A-PICC and S-PICC (p = 0.773).Catheter locations were not different (see Table 2), but infections were more common in catheters inserted at the lower extremities (χ 2 (1) = 4.104, p = 0.043).
Only in three A-PICCs bacteria resistant to the incorporated rifampin were found (p = 0.049).The resistance to miconazole was not studied.

Discussion
In this retrospective analysis, infection rates in A-PICC and S-PICC were similar, with a slight trend toward A-PICC.This is in line with the PREVAIL (13) trial and studies by Klemme et al. (21) and Bayoumi et al. (14).None of them detected differences between A-PICC and S-PICC.If the slight trend toward more infections with A-PICC was indeed an effect, the number needed to harm in our population would result in a high value of 15.58.In contrast, catheters incorporating minocycline and rifampicin had been found to have less infections in older children (22).
Flemmer et al. (12) were the only ones to report a reduction in bacteriological complications by A-PICCs in neonates and preterm infants.Unlike their study, bacterial colonization was not tested in every PICC included in our analysis.
The dwell times of PICCs were considerably longer than reported in the literature (our study 12.3 days, other studies 8.2 and approximately 6.5 days) (13,21).There were no rules prescribing maximum dwell times in our unit.While the removal of infected PICCs was more often complicated, no new  Positive cultures were not mandatory for defining an infection, because otherwise many infections would be missed owing to the low yield of blood cultures (24).To maintain specificity, we mandated more clinical and laboratory signs to be present during a fixed time interval as suggested by our national German guideline for counting an episode as an infection (15).We feel this is justified as the decision to start anti-infective treatment is in clinical routine often based on clinical criteria and laboratory results like CRP and interleukin 6.Their combination has been ascribed a high positive predictive value (15).Waiting for microbiology results would delay treatment unacceptably.We relinquished from assessing procalcitonin (PCT) since IL-6 and CRP have been shown to be superior (25,26).
Preterms with lower birth weights are at an increased risk of late-onset sepsis (LOS) (27) and CRBSI (28); longer dwell times have been associated with a rise in sepsis (27) and CRBSI (28, 29) risks.Interestingly, changing central lines after a fixed time is not recommended, because the replacement line has the same infection risk (29).
Catheter tips were not routinely sent for cultures and documentation of catheter removal is often scant.Any discussion on the microbiology from catheter tips must be interpreted after taking into account their unclear role in the diagnosis of CRBSI (30) and the scarcity of studies included in this analysis.Catheter tip cultures were not assigned a role in the definition of infection by their results or in clinical practice.Arguments have been put forward to forgo them altogether (31).Only with a synchronously drawn blood culture might their results be interpretable (30).Our data are hence not suitable for deriving clear statements on colonization.Staphylococcus epidermidis was the most common bacterial species identified at the catheter tip with no difference between PICCs considered infected vs. not infected.Such coagulase-negative staphylococci (CNS) have been described as most abundant there (32) and as a relevant cause of LOS (27) and CRBSI (33) in the literature.
Vancomycin is the drug of choice for CNS infection in the preterm; Rodriguez-Guerineau et al. found it especially effective if combined with rifampicin (16).In our department, vancomycin is being used frequently as an anti-infective prophylaxis if catheters are used beyond a certain dwell time.We were unable to distinguish such prophylaxis from actual treatments in our data; it is thus unclear whether prophylaxis actually prevented infection.
We identified multiple limitations in our data.First, the retrospective design: since the decision between A-PICC and S-PICC was made by the physician inserting it, a sum of subjective judgements may have caused A-PICCs being used in children weighing less and having a lower gestational age, possibly because A-PICCs might have been attributed increased safety for longer dwell times, which might have been considered especially advantageous in smaller infants.Since these differences may be confounding variables, some findings in our data may be difficult to interpret.
A-PICCs in our study were inserted in patients weighing significantly less, but were also used longer.The trend toward slightly higher infection rates in the A-PICC group may hence be rather due to the infection risk being generally higher in more immature infants.
Due to low numbers of infection in both groups, estimated power is 25.04%, which is too low to accept the Null Hypothesis, i.e., to demonstrate practical equivalence.A-PICCs and S-PICCs were inserted in patients different in weight and gestational age to which the literature ascribes different CRBSI risks.This lowers the informative value significantly, especially with regard to comparability with other studies.Our infection definition hinges on infection symptoms and laboratory studies, which we feel is representative of clinical practice.The number of infections in our study would have been severely reduced if we had kept to the definition by NHSN demanding a positive blood culture, because our department orders few microbiological studies.It seems unlikely that in our case more microbiological studies would have helped increase the number of infections since only 4%-12% return positive (34)(35)(36)(37).Missing a large number of infections, however, would also not improve the conclusions.Of note, there is a high variability in peripheral venous blood culture positive rates: Klemme et al. had no positive blood cultures, whereas the PREVAIL trial reported 61.3%.Limitation of the analysis to catheter dwell time (plus 48 h) and exclusion if any other site of infection had been documented was further reducing the number of infections.Despite all of these shortcomings, we feel that our definition is very close to what is happening in the clinical routine, where catheter infection is most often diagnosed by ruling out other explanations, and antibiotic treatment is started long before microbiology results are available.
If our observations were used to estimate the sample size of a trial with an ɑ = 0.05 and a power of 90%, about 700 PICCs would be needed in both arms.
Finally, because of the few infections in our study, a clear benefit from repeatedly retracting longer-dwelling PICCs was not discernible.
In spite of these limitations, this manuscript is a description of the current use and complications of PICCs in our department with an infection definition that resembles routine rather than an artificial study environment.

Conclusion
In line with other studies on this topic, we could find no clear advantage regarding infections by incorporating anti-infective drugs into peripherally inserted central catheters.
If one were still to stipulate more evidence, for deciding whether to use A-PICCs, a larger trial would be required.

TABLE 1
Patient characteristics by catheter type used (comparisons are χ²-test for categorical and t-test for discrete variables).PICCs vs. 11 of 36 A-PICCs returned positive, so did 12 of 42 of those with no infection vs. 6 of 17 of the infected; Klebsiella oxytoca, Lactobacillus fermentum, Enterobacter cloacae, Klebsiella pneumoniae, Bacillus subtilis, Bacillus cereus, and Serratia marcescens were found.Skin flora, namely, Staphylococcus epidermidis, Staphylococcus haemolyticus, and Staphylococcus hominis, were significantly more often detected on infected catheters (p = 0.005).Candida antigen studies were not different between the groups.

TABLE 2
Insertion site of catheters grouped by infection and anti-infective incorporation.Infection was more common in catheters in the lower extremities (χ2(1) = 4.104, p = 0.043).

TABLE 3
Use of antibiotics by either catheter type and infection and comparisons of dwell times, times to infection, and time to infection or removal with and without anti-infectives.