Antimicrobial-impregnated central venous catheters for prevention of neonatal bloodstream infection (PREVAIL)

Summary Background Bloodstream infection is associated with high mortality and serious morbidity in preterm babies. Evidence from clinical trials shows that antimicrobial-impregnated central venous catheters (CVCs) reduce catheter-related bloodstream infection in adults and children receiving intensive care, but there is a paucity of similar evidence for babies receiving neonatal intensive care. Methods This open-label, parallel-group, pragmatic, randomised controlled trial was done in 18 neonatal intensive care units in England. Newborn babies who needed a peripherally inserted CVC (PICC) were allocated randomly (1:1) to receive either a PICC impregnated with miconazole and rifampicin or a standard (non-antimicrobial-impregnated) PICC. Random allocation was done with a web-based program, which was centrally controlled to ensure allocation concealment. Randomisation sequences were computer-generated in random blocks of two and four, and stratified by site. Masking of clinicians to PICC allocation was impractical because rifampicin caused brown staining of the antimicrobial-impregnated PICC. However, participant inclusion in analyses and occurrence of outcome events were determined following an analysis plan that was specified before individuals saw the unblinded data. The primary outcome was the time from random allocation to first microbiologically confirmed bloodstream or cerebrospinal fluid (CSF) infection between 24 h after randomisation and 48 h after PICC removal or death. We analysed outcome data according to the intention-to-treat principle. We excluded babies for whom a PICC was not inserted from safety analyses, as these analyses were done with groups defined by the PICC used. This trial is registered with ISRCTN, number 81931394. Findings Between Aug 12, 2015, and Jan 11, 2017, we randomly assigned 861 babies (754 [88%] born before 32 weeks of gestation) to receive an antimicrobial-impregnated PICC (430 babies) or standard PICC (431 babies). The median time to PICC removal was 8·20 days (IQR 4·77–12·13) in the antimicrobial-impregnated PICC group versus 7·86 days (5·00–12·53) days in the standard PICC group (hazard ratio [HR] 1·03, 95% CI 0·89–1·18, p=0·73), with 46 (11%) of 430 babies versus 44 (10%) of 431 babies having a microbiologically confirmed bloodstream or CSF infection. The time from random allocation to first bloodstream or CSF infection was similar between the two groups (HR 1·11, 95% CI 0·73–1·67, p=0·63). Secondary outcomes relating to infection, rifampicin resistance in positive blood or CSF cultures, mortality, clinical outcomes at neonatal unit discharge, and time to PICC removal were similar between the two groups, although rifampicin resistance in positive cultures of PICC tips was higher in the antimicrobial-impregnated PICC group (relative risk 3·51, 95% CI 1·16–10·57, p=0·018). 60 adverse events were reported from 49 (13%) patients in the antimicrobial-impregnated PICC group and 50 events from 45 (10%) babies in the standard PICC group. benefit harm miconazole PICCs compared standard PICCs types of antimicrobial impregnation PICCs alternative


Introduction
Bloodstream infection is the most common serious complication associated with the use of central venous catheters (CVCs) in newborn babies. Microbial pathogens adhere to the catheter material and secrete a biofilm 1 that protects them from circulating antimicrobial drugs, enabling sustained colonisation. 2 CVC removal is often needed to clear the infection.
Catheter-related bloodstream infection has been reported to occur in up to 30% of neonates, with the highest rates in babies who are born very prematurely (before 32 weeks of gestation). 3,4 The organisms most frequently isolated from preterm babies are coagulasenegative staphylococci, Gram-negative bacilli, other Grampositive cocci (Staphylococcus aureus and enterococci), and fungi (predominantly Candida species). 5 Bloodstream infection increases the risk of death and serious morbidity in very preterm babies, 6,7 and is associated with long-term adverse neuro developmental outcomes. 8,9 Use of antimicrobial-impregnated CVCs is recommended in US and UK national guidelines for patients at high risk of infection. 10,11 A randomised trial involving 1485 children (aged 1 week to 18 years) receiving paediatric intensive care in the UK showed that use of antimicrobial-impregnated CVCs versus standard (nonimpregnated) CVCs reduced bloodstream infection. 12 However, there are no recommendations for newborn babies because few antimicrobial-impregnated catheters are narrow enough for preterm babies and there is a paucity of evidence from adequately powered randomised trials. 13,14 The PREVAIL trial aimed to address this evidence gap by determining the effectiveness of an antimicrobial-impregnated CVC licensed for newborn infants. We compared use of a miconazole and rifampicin-impregnated CVC with a standard (nonimpregnated) CVC for reduction of bloodstream infection, morbidity, and mortality in newborn babies receiving intensive care.

Study design and participants
This open-label, parallel-group, pragmatic, randomised controlled trial was done in 18 neonatal units in England. Full details of the amended trial protocol (version 5.0; published April 26, 2017), research ethics approval (from Yorkshire and the Humber-Sheffield-Health Research Authority reference no 14-YH-1202), and statistical analysis plan are available online.
All babies requiring a narrow-gauge (French gauge 1) peripherally inserted CVC (PICC) were eligible to participate. The reason for insertion was not requested by the trial team, but PICCs are usually used for parenteral nutrition and drug administration-70% of babies in neonatal units born before 32 weeks of gestation have a PICC inserted (Fraser C, National Neonatal Research Database, England, personal communication). Babies who had previously entered this trial and those who had a known allergy or hypersensitivity to rifampicin or miconazole were excluded.  concluded that, given the paucity of evidence, a large, simple, and pragmatic randomised controlled trial of this intervention was needed to guide policy and practice.

Added value of this study
The PREVAIL trial showed that use of miconazole and rifampicin-impregnated, percutaneously inserted CVCs compared with use of standard CVCs did not reduce the risk of any bloodstream infection or catheter-related bloodstream infection, other morbidity, or mortality in newborn infants. To our knowledge, this is the largest trial to date of this intervention and the validity is enhanced by the methodological quality and power. The findings of this study are broadly applicable to newborn infants cared for in facilities in well-resourced health-care services.

Implications of all the available evidence
The PREVAIL trial findings contrast with those of previous randomised controlled trials of antimicrobial-impregnated CVCs, which reported substantial reductions in bloodstream infection in older children and adults. A possible explanation for this difference is that the trials involving children and adults assessed CVCs impregnated with minocycline and rifampicin rather than the miconazole and rifampicin combination used in this trial. Rifampicin might be more effective when combined with a synergistic antibacterial (minocycline) rather than an antifungal (miconazole) and a simple, pragmatic randomised controlled trial of minocycline and rifampicin-impregnated percutaneously inserted CVCs in newborn infants might now be warranted.
See Online for appendix The parent or legal representative of the baby gave written informed consent for inclusion in the trial.

Randomisation and masking
Participants were randomly assigned (1:1) to receive either an antimicrobial-impregnated PICC or a standard PICC, by use of a secure web-based randomisation programme by the principal investigator or a delegated other individual at the site. Enrolment was done by the principal investigator or a delegated other individual at the site. The randomisation programme was centrally controlled by the Clinical Trials Research Centre (University of Liverpool, Liverpool, UK) to ensure allocation concealment. Randomisation sequences were computer-generated by an independent statistician in random blocks of two and four and stratified by site. Masking of clinicians to PICC allocation was impractical because rifampicin caused brown staining of the antimicrobial-impregnated PICC. However, participant inclusion in analyses and occurrence of outcome events were determined by following an analysis plan that was specified before individuals saw unblinded data.

Procedures
Trial participants were randomly allocated to receive either a miconazole and rifampicin-impregnated PICC (Premistar; Vygon, Swindon, UK) or standard (nonimpregnated) PICC (Premicath; Vygon). PICC insertion was done according to standard unit policy and practice. Miconazole is an anti-fungal agent, which is effective against systemic fungal infection. Rifampicin is an antibacterial agent that has been previously evaluated in rifampicin and minocycline CVC impregnation in adults and children. Vygon reported continuing elution from the CVC impregnated with rifampicin and miconazole over 21 days. 15 The antimicrobial-impregnated PICC was marketed after appropriate certification under the Conformité Européenne process in December, 2012 (certificate number Z/12/02895).
The allocated PICC was inserted within 48 h of random assignment, and thereafter a standard PICC was used. Infection outcomes were captured for all babies until 48 h after PICC removal or following the last unsuccessful PICC insertion or random assignment (if insertion was not attempted). Follow-up for secondary clinical outcomes continued until the infant was discharged to go home, death, or 6 months after randomisation, whichever occurred first. Follow-up for all deaths continued until 6 months after randomisation.

Outcomes
The primary outcome was the time from random allocation to first bloodstream or cerebrospinal fluid (CSF) infection, defined as a microbiological culture of a bacteria or fungus from the blood or CSF sampled for clinical reasons. We use the term bloodstream infection to mean this combined outcome. The time window for sampling for primary and secondary outcomes was 24 h after randomisation until 48 h after PICC removal or death (or 48 h after randomisation if PICC not inserted). We imposed a-priori decision rules to avoid counting preexisting bloodstream infection. We excluded microbial cultures within the assessment time window if the same organism was isolated from blood or CSF and samples were taken less than 14 days apart, or if a different organism was isolated and samples were less than 24 h apart. Multiple infection episodes within the time window were considered as distinct infection episodes if positive samples for each episode involved the same organism and occurred more than 14 days apart or involved different organisms and occurred more than 24 h apart.
Secondary outcomes related to infection were as follows: the type of organism isolated from bloodstream infection meeting primary outcome criteria, the rate of bloodstream infection (including recurrent bloodstream infection) per 1000 days with PICC, occurrence of one or more bloodstream infections, rate of catheter-related bloodstream infection (defined by isolation of the same organism from the PICC tip and blood or CSF) per 1000 days with PICC, rifampicin resistance in any isolate from blood or CSF culture, rifampicin resistance in any isolate from PICC tips, and rifampicin resistance in any isolate from blood or CSF culture or from the PICC tip (this was added as an additional analysis in version 2.0 of the statistical analysis plan, which was approved on March 7, 2018, before database lock [study closure]).
Outcomes measured to detect potential biases in sampling or treatment on the basis of knowledge of PICC allocation were as follows: rate of blood or CSF culture sampling per 1000 days with PICC, duration of antimicrobial exposure from randomisation up to 48 h after line removal, and time to PICC removal.
Clinical secondary outcomes were as follows: chronic lung disease, defined as requiring respiratory support (mechanical ventilation or continuous positive pressure via endotracheal tube or nasal tube) or supplemental oxygen at 36 weeks' postmenstrual age, necrotising enterocolitis (Bell stage 2 or 3), treatment for retinopathy of prematurity (medical or surgical), abnormalities on cranial ultrasound (periventricular leukomalacia or intracranial haemorrhage; worst grade of 1 to 4 used in analyses), time from random allocation to full milk feeds (150 mL/kg per day), total duration of parenteral nutrition from random allocation until discharge from neonatal care, and death before discharge to go home from neonatal care.
Death within 6 months of randomisation and time to death were recorded from linked death registration data. We recorded occurrence of related adverse events for all babies who had a PICC successfully inserted until 48 h after PICC removal. For all babies who had a PICC successfully inserted, we recorded occurrence of any expected or unexpected adverse events considered to be related to the PICC until 48 h after PICC removal.

Statistical analysis
The sample size calculation for the primary outcome was based on the log-rank test for equality of survival curves, with a 5% significance level and 90% power to detect a difference between the treatment groups. We considered a 50% reduction in bloodstream infection to be conservative, given the results of a network meta-analysis by Wang and colleagues 16 of catheter-related bloodstream infections (mean odds ratio 0·18, upper 95% CI 0·34), and the results of the CATCH trial. 10 To detect a reduction in the proportion of babies with a bloodstream infection from 14% in the standard CVC group, which was expected based on audit data from three participating neonatal units, to 7% in the antimicrobial-impregnated CVC group, 79 events were required from 816 babies (408 in each group); allowing for 5% loss to follow-up, we planned to recruit 858 neonates in total.
We analysed the primary outcome and secondary survival outcomes with the log-rank test. We used Kaplan-Meier curves to present the numbers at risk and Cox regression to calculate hazard ratios (HRs). We analysed binary outcomes with Fisher's exact test and presented relative risks (RRs) with 95% CIs. We analysed continuous outcomes with the Mann-Whitney U test and presented medians with IQRs for each group. We analysed rate outcomes using Poisson regression and presented rate ratios with 95% CIs. Descriptive results only are presented for the type of organisms isolated from bloodstream infections, related adverse events, and serious adverse events.
We analysed outcome data according to the intention-totreat principle. Babies who were randomly assigned but had no PICC inserted were assessed for infection-related outcomes until 48 h after the last attempted insertion or 48 h after randomisation. We excluded babies for whom a PICC was not inserted from safety analyses, as these analyses were undertaken with groups defined by the PICC used.
All statistical tests were two-sided, with a 5% significance level. All analyses were done with SAS software, version 9.4. Results from the primary outcome and safety analyses were validated by independent programming by another independent statistician from the point of raw data.
We did four prespecified sensitivity analyses of the primary outcome as follows: time to serious bloodstream infection, defined as treatment with antimicrobials for 72 h or longer or death during treatment; time from PICC insertion to first bloodstream infection; time to first bloodstream infection, excluding samples obtained via arterial cannulas or CVCs; and time to first bloodstream infection, including clearly path ogenic organisms and excluding skin organisms (eg, coagulase-negative staphylococci). For comparability with published studies we also reported bloodstream infection rates per 1000 days with PICC between randomisation and PICC removal.
After seeing the results of the primary analysis, we specified an additional post-hoc analysis of the primary outcome to investigate whether the treatment effect varied by gestational age at birth (before 28 weeks of gestation or at 28 weeks or more of gestation) using a Cox proportional hazards model, including an interaction between treatment and gestational age.
The study was monitored by an independent Data Monitoring Committee who made recommendations to the Trial Steering Committee. An internal pilot study was done to show the feasibility of recruitment after the first 6 months and an interim analysis of the primary outcome was done after around half of the babies had been randomly assigned.
The trial is registered with ISRCTN, number 81931394.

Role of the funding source
The funder appointed independent members to the Trial Steering Committee and Data Monitoring Committee, approved all protocol amendments, and monitored study progress against agreed milestones. The funder had no involvement in data interpretation or writing of the report. The corresponding author had full access to all outputs from the data in the study and had final responsibility for the decision to submit for publication.  The time from random allocation to first bloodstream or CSF infection was similar between the two groups (HR 1·11, 95% CI 0·73-1·67; table 4, figure 2). The Kaplan-Meier curves crossed when the numbers at risk were low. A time varying coefficient was added to the model to check the assumption of proportional hazards, and this was not significant (p=0·62; data not shown). We observed similar results to our primary outcome findings in our sensitivity analyses (table 4).

Between
We found no evidence of a difference in treatment effect for babies with a gestational age of less than 28 weeks compared with 28 weeks or more (p=0·28; appendix). 46 (11%) of 430 babies in the antimicrobialimpregnated PICC group had one or more bloodstream infections, and three (7%) of these babies had two infection episodes. 44 (10%) of 431 babies in the standard PICC group had one or more bloodstream infections, and one (2%) of these had a second infection episode.
Secondary infection-related outcomes were similar between the trial groups, with the exception of rifampicin resistance from PICC tip cultures (RR 3·51, 95% CI 1·16-10·57; p=0·018). We found no significant difference when comparing the two groups for rifampicin resistance from blood, CSF, or PICC tip cultures combined (RR 1·80, 0·84-3·86; p=0·13; table 4). We noted the organisms isolated during the primary outcome time window, which were pre dominantly coagulase-negative staphylococci in both trial groups (appendix). We also recorded rifampicin-resistant isolates by type of organism (appendix). Measures of blood or CSF sampling are shown in table 4. Fewer than half of the babies in each trial group had one or more blood or CSF samples taken (table 3). The rate of blood sampling for suspected infection was significantly higher in the antimicrobialimpregnated PICC group than in the standard PICC group (97·90 per 1000 days with antimicrobialimpregnated PICC vs 79·64 per 1000 days with standard PICC; rate ratio 1·23, 95% CI 1·05-1·45; p=0·012; table 4). We found no significant differences in the median time to PICC removal (8·20 days [IQR 4·77-12·13] in the antimicrobial-impregnated PICC group vs 7·86 days [5·00-12·53] days in the standard PICC group; HR 1·03, 95% CI 0·89-1·18, p=0·73) or in the median duration of antimicrobial treatment (3 days [IQR 2-6] in both groups; p=0·25; table 4). We found no significant differences in any clinical outcomes measured at discharge from the neonatal unit or in mortality within 6 months of randomisation (table 4).
We recorded summary data of the most frequent adverse events (table 5). 60 events were reported from 49 (13%) babies in the antimicrobial-impregnated PICC group and 50 events were reported from 45 (10%) babies in the standard PICC group. One serious adverse event involving supra ventricular tachycardia following PICC placement was reported in the antimicrobial-impregnated PICC group.    Antimicrobial treatment, days 3·00 (2·00-6·00) 3·00 (2·00-6·00) NA 0·25 Parenteral nutrition, days 11·00 (7·00-19·00) 10·00 (7·00-18·00) NA 0·83 Data are n (%) or median (IQR), unless otherwise indicated. For all outcomes that relate to samples, events were only considered on samples taken between 24 h after randomisation until 48 h after removal. PICC=peripherally inserted central venous catheter. HR=hazard ratio. RR=relative risk. CSF=cerebrospinal fluid. NA=not applicable. *HRs (95% CIs) are given for primary outcome data and time to event data; rate ratios (95% CIs) are given for secondary outcomes pertaining to rate data; and RRs (95% CIs) are given for all other data. †Median time to event not reported as not enough babies had an event. ‡Only includes babies where PICC was successfully inserted (antimicrobial-impregnated PICC: 401 babies; standard PICC: 415 babies). §Outcome not prespecified in protocol but requested by investigators and included in statistical analysis plan before any unmasked data were seen.

Discussion
We found no evidence of benefit or harm from miconazole and rifampicin-impregnated PICCs in babies receiving intensive care. The 95% CI for the primary outcome excluded a 27% reduction or 67% increase in the time to bloodstream infection associated with using an antimicrobial-impregnated PICC compared with a standard PICC, and similar results were found in the sensitivity analyses. We found no differences between the two groups for mortality at 6 months or clinical outcomes recorded at discharge home from the neonatal unit. The strengths of the trial are the large sample size and the multicentre, nationally representative sample of babies admitted for neonatal intensive care, which was adequately powered to detect a halving of the bloodstream infection risk. As 754 (88%) babies participating in the trial were born before 32 weeks of gestation, this trial provides important new evidence for a group at high risk of infection, who have frequent use of PICCs, but for whom trial evidence is lacking. 13,14 The pragmatic trial design, with no additional sampling, and use of a primary outcome on the basis of positive cultures taken as part of clinical practice in response to suspected infection to guide antibiotic treatment, ensured relevance to routine practice. Although blood culture is relatively insensitive, we found no difference in clinical outcomes such as duration of antibiotics or PICC insertion.
We used central web-based randomisation to ensure allocation concealment, achieved almost complete follow-up and assessment of the primary outcome, adhered to a prespecified statistical analysis plan for intention-to-treat analyses, and halted recruitment once the sample size was achieved. Baseline characteristics were well balanced at randomisation. Slightly fewer babies in the anti microbial-impregnated PICC group received the allocated PICC, probably because the randomly assigned PICC had to be inserted within 48 h; thereafter, the standard PICC was used. The proportion of babies in the standard PICC group with bloodstream infection (44 [10%] of 431 babies) was lower than expected (14%) but we had sufficient power to exclude a moderate reduction in the risk of bloodstream infection. We based the sample size and statistical analyses on the proportional hazards assumption. We checked this assumption and believe it is reasonably upheld, but acknowledge that there could be an impact on study power if this is not the case. This study was open label, so clinicians could distinguish the type of PICC. We found a slightly increased rate of blood culture sampling in the antimicrobial-impregnated PICC group, but the proportions of babies with at least one blood or CSF culture or any PICC tip culture were similar and there were no differences in the timing of PICC removal between trial groups.
A limitation of this study was the insufficient power to detect significant differences in rifampicin-resistant organisms isolated from blood or CSF cultures. The low number of resistant organisms was due to few positive cultures, and because only 44-54% of these cultures were tested for rifampicin resistance. The risk of rifampicin resistance in isolates from positive blood or CSF cultures did not differ between the trial groups but was significantly increased in positive tip cultures from antimicrobialimpregnated PICCs. Selection of rifampicin-resistant Gram-positive bacteria during treatment-if rifampicin is used as the sole antibacterial agent-is well recognised. 17 Emergence of resistant organisms was considered by the investigators, the Trial Steering Committee, and the Data Monitoring Committee, but the risk of adverse events arising was viewed as low because the restricted release of rifampicin from the catheter surface would be unlikely to affect bacteria at any site other than the catheter itself. Even if rifampicin-resistant Gram-positive bacteria did cause infection in an indi vidual patient, routine antibiotic use would be unaffected because rifampicin is rarely used for treatment in the neonatal setting.
We found that miconazole and rifampicin impregnation did not reduce bloodstream infection in newborn babies. This result is consistent with findings from a randomised controlled trial in adults and a small randomised  controlled trial in newborn infants published as an abstract (appendix). 14,18,19 However, our findings contrast with evidence of reduced catheter-related bloodstream infection in adults and reductions in any bloodstream infection in children randomised to minocycline and rifampicin-impregnated CVCs versus standard CVCs. 10,20,21 Several explanations could account for these differences. First, miconazole and rifampicin might be less effective than minocycline and rifampicin impregnation.
Miconazole is used to prevent invasive fungal infection in preterm babies, which is rare in the UK, but has a very high mortality. 8 Few babies in our trial had fungal bloodstream infection, consistent with a recent UK study. 22 However, rifampicin might be less effective when used as the sole antibacterial agent combined with miconazole. Rifampicin is more active against Grampositive than against Gram-negative bacteria and has synergistic action against staphylococci when combined with another antibacterial drug-eg, minocyclineespecially against methicillin-resistant strains. 17,23 Second, although the combination of minocycline and rifampicin has been reported as the most effective type of antimicrobial impregnation in systematic reviews, 24,25 it is possible that minocycline and rifampicinimpregnated CVCs might not effectively reduce overall incidence of bloodstream infection or sepsis. 26 A network meta-analysis of seven randomised trials in adults showed beneficial effects of antimicrobial impregnation for catheter-related blood stream infection, 25 but few trials have measured the effect on any bloodstream infection. 10 Catheter-related infection requires the same isolates from blood and CVC tip and could be prone to bias because of inhibition of positive tip cultures by leaching of antimicrobial from the tip during plating out for culture. Only the large CATCH trial 10 in children used any clinically indicated bloodstream infection as the primary outcome and found a 57% reduction in time to infection (appendix). A smaller trial 27 compared catheter-related bloodstream infection in children randomly assigned to minocycline and rifampicinimpregnated CVC or standard CVC and found no difference in the rate of catheter-acquired bloodstream infection, but detected few infection events (three in each group; appendix). 27 Third, reductions in infection rates in neonatal units associated with improved catheter asepsis practices and shorter duration of PICC use might have narrowed the potential for further benefits from antimicrobial impregnation. 28 PICCs also might not be an independent risk factor for infection in sick preterm babies because of their high susceptibility to infection from multiple sources, due to numerous invasive procedures and devices, gut permeability, and immune immaturity. 29 Since 2012, the Premistar PICC has been the only antimicrobial-impregnated PICC available for preterm babies in Europe. Use of Premistar has been reported in Germany and Italy, 14 but, to our knowledge, use in the UK has been limited to the PREVAIL trial. The trial findings do not support the use of miconazole and rifampicin-impregnated PICCs in newborn infants, as we found no evidence that antimicrobial-impregnated PICCs reduced bloodstream infections and such PICCs are more expensive than standard PICCs. However, the serious life-long consequences of bloodstream infection mean that even interventions with relatively small effects might be clinically important. One in ten babies in the PREVAIL trial had a bloodstream infection and some might have serious life-long neurodevelopmental impairment or lung disease as a result. Therefore, more large trials are urgently needed to reduce the risks of bloodstream infection and their long-term consequences. Viewed against existing evidence on the effectiveness of rifampicin and minocycline-impregnated CVCs in adults and children, our findings have implications for the manufacturers of these devices and for clinicians. First, alternative types of antimicrobial-impregnated PICCs should be considered for evaluation in newborn babies, along with other interventions to prevent bloodstream infection in the neonatal intensive care unit. Second, further randomised controlled trials should be considered to determine whether previous evidence on the effectiveness of antibiotic-impregnated CVCs in adults is sustained in the context of effective infection control practices.