Discussion

Current practice guidelines indicate that the optimal NC design for CRBSI prevention is unknown.^{Reference Buetti, Marschall and Drees10,Reference Gorski, Hadaway and Hagle11} Although several design features have been associated with BSIs, none have been studied for their relative contribution to the risk.^{Reference Jarvis14} Recent investigation of the clinical impact of NC design maintains that analyses of microbial ingress, antiseptic methods, and biofilm formation in different designs are needed to address CRBSIs.^{Reference Rosenthal1} To our knowledge, this is the most comprehensive study of NC design, intraluminal biofilm formation, and bacterial transfer risk to date. Our study was performed over 5 years with 33 independent experimental runs using 20 nondisinfected NCs to determine differences in bacterial transfer in the flush and intraluminal biofilm colonization in the connector-catheter system, allowing an assessment of contributing design factors.

Since the clinical introduction of NCs, CRBSIs have been linked to their use. Intraluminal biofilms were first detected in NCs harvested from patients and tested at the Centers for Disease Control and Prevention (CDC) in 2001.^{Reference Donlan, Murga and Bell15} Outbreaks have been reported with each design generation^{Reference Rosenthal1} ; moving from the original split septum/blunt cannula to mechanical valves differentiated by negative, positive, and neutral displacement.^{Reference Jarvis14} Early investigation implicated negative and positive displacement mechanical valve NCs to be associated with increased BSIs.^{Reference Jarvis, Murphy and Hall16} Research has since suggested that differences in contamination risk among NCs (even with disinfection) are not specific to the displacement categorization as either positive, negative, or neutral.^{Reference Casey, Karpanen, Nightingale and Elliott17} A recent literature review also reported that BSI rates were variable for different displacement categorizations.^{Reference Rosenthal1}

This clinically simulated in vitro study was designed to represent a common treatment schedule for intermittent intravenous infusion, and was modified from standardized, controlled protocols used to investigate NC biofilm development.^{Reference Donlan, Murga and Bell15,Reference Murga, Miller and Donlan18} Previous investigators mostly used in vitro models to compare bacterial transfer in the flush with variable protocols, some including NC disinfection.^{Reference Casey, Karpanen, Nightingale and Elliott17,Reference Guembe, Pérez Granda, Cruces, Alcalá and Bouza19,Reference Casey, Karpanen, Nightingale and Elliott20}

NC septum disinfection prior to each access is a recognized critical step to prevent microbial transfer.^{Reference Loftus, Patel and Huysman21} However, the most effective disinfection method remains unknown as no study to date has tested against septum surface biofilms, even though significant biofilm accumulation has previously been observed on NC septums and internal surfaces using scanning electron microscopy.^{Reference Ryder, Schaudinn, Gorur and Costerton22} Compliance with clinical NC disinfection is variable (12%–100%),^{Reference Burke, Jakub and Kellar23,Reference Slater, Cooke, Scanlan and Rickard24} with poor adherence to disinfection protocols and techniques.^{Reference Slater, Cooke, Scanlan and Rickard24} Accordingly, microbial burden on NC septum surfaces prior to disinfection has ranged from 10¹ to 10⁵ CFU in up to 67% of NCs.^{Reference Sengul, Guven, Ocakci and Kaya25–Reference Rupp, Yu and Huerta27} This is likely an underestimate because simplistic culturing techniques used in these studies typically yield low retrieval rates.^{Reference Murga, Miller and Donlan18,Reference Verran, Redfern, Smith and Whitehead28} Thus, more vigorous methods of sonication and vortex, as used in this study, are necessary to remove and disaggregate surface-attached biofilm cells.^{Reference Casey, Karpanen, Nightingale and Elliott17,Reference Perez, Williams and Jacob29} Hence, this study was performed without disinfection, and with a high inoculum (10⁶ CFU), to simulate both the microbial burden found clinically and the worst-case scenario of NC disinfection noncompliance. Omission of the disinfection step was also intended to robustly ascertain performance differences among NCs and to compare the intrinsic risk of each NC in the absence of proper disinfection.

Our results demonstrated several key findings, among dispelling displacement categorization as the sole predictor of CRBSIs. First, flow path was the most important design factor; the internal cannula was associated with less bacterial transfer and biofilm formation. Second, neutral displacement and simple hydrodynamics were associated with reduced NC biofilm and bacterial transfer in the flush. Complex designs with internal irregularities can disrupt laminar flow, which can cause high velocity and turbulence, contributing to biofilm formation.^{Reference Nybo, Maneval and Codd30–Reference Busscher and van der Mei34} Third, variation in access portal seal length, surface area, and volume influence differences seen among NCs with similar and dissimilar displacement and hydrodynamic characteristics. This likely explains why all 8 factors simultaneously best predict bacterial transfer and biofilm formation and that displacement classification alone can be misleading. Lastly, biofilm bacteria were most abundant in the NC compared to the hub and lumen, which was also the only significant predictor of bacteria in the flush.

The CDC recommends NC exchange at least every 96 hours or according to the manufacturer’s instructions.^{Reference O’Grady, Alexander and Burns35} Our analyses support this, given the high levels of biofilm on the NC, hub, and catheter lumen by 72 hours. However, our study also demonstrates that some NCs require more frequent exchanges, given that some NCs allow more bacteria through the connecter over time compared to others. Even after 5 days, bacterial counts in the flush among some NCs never reached the high levels that were observed after 24 hours in other NCs (Fig. 5A). The substantial differences in the daily bacterial transfer rate and the increase in bacteria over time for some NC types warrant reconsideration of NC exchange recommendations. However, it is important to note that NC replacement does not affect biofilm in the hub or catheter, as these distal components will continue to harbor biofilm.

Current IDSA Clinical Practice Guidelines recommend culturing long-term catheter insertion site and hub to rule out the catheter as the infection source for CRBSIs,^{Reference Mermel, Allon and Bouza36} but this is rarely practiced due to the complexity and risk of obtaining hub cultures. In a series of studies, researchers investigated the efficacy of insertion-site skin cultures and NC flush sonication cultures in the diagnosis of CRBSI to show that the combination of skin culture and NC quantitative cultures (>1,000 CFU/connector) can be used to rule out catheter colonization and CRBSI.^{Reference Pérez-Granda, Barrio and Cruces37} Our results support these findings because NC biofilm was the best predictor of bacterial transfer in the flush. Moreover, in our study, MicroClave, Clave Neutron, SafeTouch and SafeAccess NCs generated substantially lower rates of bacterial transfer. This finding may explain a reduced relative risk of CLABSI in hospitals using Clave technologies compared to hospitals using other NCs.^{Reference Ryder, Sun and Battle38}

This study had several limitations. First, the study design may not have represented all possible uses of NCs because clinical access frequencies and patterns vary greatly. Second, blinding study personnel to NCs was not possible due to visual distinguishability, which potentially introduces bias, and procedures were standardized to minimize this bias. Third, the use of lower inoculum and/or other organisms may produce dissimilar results. Fourth, although we focused primarily on design factors most likely to influence biofilm formation based on prior research, other factors may exist. Material type would be one to consider; however, it would be difficult to assess because septum, housing, and internal components use different materials in different NCs. Lastly, additional clinical studies are warranted to assess real-world risk for CRBSI with different NC types.

Overall, our findings have demonstrated that NC choice can be an important CRBSI risk reduction strategy. Bacterial transfer and biofilm formation through the connector-catheter systems were statistically significantly different among NCs with different designs. Furthermore, biofilm in the NC was predictive of bacterial transfer in the flush. The NC designs associated with the least bacterial transfer and biofilm formation were NCs with a split septum, minimal seal length, internal cannula, low surface area and volume, neutral displacement, and simple hydrodynamics of the flow path. Although NC disinfection remains critically important, this research highlights the need to choose NCs with the least risk of biofilm formation and bacterial transfer to reduce CRBSIs.