Table 1. Recommended practice, supporting evidence and quality and strength of evidence according to GRADE criteria.

Practice	EPIC3 guideline recommendation	Supporting evidence (summary of studies in S1 File)	Quality and strength of evidence using GRADE
1) Use of a care bundle for (a) insertion or (b) maintenance of PICCs and (c) monitoring compliance of care bundles	Use quality improvement interventions including protocols for catheter insertion and maintenance, audit of compliance with practice and feedback to practitioners. Guidelines note that these are commonly implemented as bundles.	A systematic review of CVC bundles in all ages and three reviews of CVC bundles in neonates or children found evidence of a reduction in BSI associated with CVC bundles, but the included studies were of low quality.[24–27] The only RCT identified evaluated a parenteral nutrition care bundle in neonates and found no difference in rates of late onset sepsis between the bundle and standard care.[28] We found six subsequent ‘before versus after’ studies, all of which reported reduced BSI rates after introduction of CVC bundles.[29–34] Only one study (KH) took into account pre-existing BSI trends and adjusted for case mix.[32] This UK study found a small but significant (14%) reduction in BSI rate after introduction of CVC bundles. CVC bundles were consistently associated with a reduction in BSI rates in before vs after studies.	Low quality of the evidence, weak strength of recommendation (observational studies, one RCT found no difference)
2) Routine replacement or removal of PICCs after a specified time period	Do not routinely replace CVCs to prevent infections.	We found no systematic reviews that evaluated routine replacement of CVCs in neonates or children. One systematic review of 12 RCTs in adults concluded that routine replacement of CVCs does not reduce the rate of BSI compared with replacement as needed.[35]	Moderate quality evidence, strong strength of recommendation (RCT evidence in adults is indirect)
3) Use of chlorhexidine-impregnated foam dressing at site of PICC insertion.	Do not use chlorhexidine-impregnated foam dressing in neonates. Only use in adults.	A systematic review of RCTs evaluating the efficacy and safety of antimicrobial dressings in reducing BSI for neonates only identified one small RCT evaluating chlorhexidine-impregnated dressings.[36] Chlorhexidine-impregnated dressings have a moderate effect on catheter colonisation, no significant effect on BSI and a high risk of contact dermatitis.	Moderate quality evidence, strong strength of recommendation (RCT is small and subject to bias but shows risk of harm with no effect on BSI)
4) 2% chlorhexidine in alcohol for skin preparation prior to insertion of PICC.	Apply 2% chlorhexidine in alcohol at the CVC insertion site.	Two reviews, conducted 10 years apart, found no RCTs that showed a significant benefit of chlorhexidine for skin preparatin in neonates on rates of BSI.[37, 38] There are concerns regarding the toxicity of chlorhexidine and alcohol, particularly neurotoxicity in the case of chlorhexidine.[39–42]) A recent RCT in neonates found 10% providone-iodine and 2% chloehexidine in alcohol to be equally effective at reducing CLABSI with no difference in skin damage, but the study was underpowered.[43]	Low quality evidence, weak strength of recommendation (one underpowered RCT and evidence of harm)
5) Cleaning of catheter ports with 2% chlorhexidine in alcohol.	Decontaminate CVC access ports with 2% chlorhexidine in alcohol.	No studies met our inclusion criteria. One before and after study in neonates that did not report pre-existing trends found a reduction in BSI when changing from 70% isopropyl alcohol alone to 2% chlorhexidine in 70% isopropyl alcohol to clean CVC connectors.[44] The evidence in adults cited in the epic3 guidelines should be applicable to neonates as the mechanism for colonisation is unlikely to differ in the two age groups.[45, 46]	Low quality evidence, weak strength of recommendation (no RCTs in correct population)