This single-center retrospective study demonstrated promising anti-infectious outcomes of the silver-mixed TIVAP device, as the silver-mixed TIVAP (S group) was found to be the most correlated factor for TIVAP-related infection. The TIVAP-related infection rates were 3.0%, 7.7%, and 6.1%, and the incidence of total infection per 1,000 catheter-days were 0.114, 0.214, and 0.187 in the S group, NS group, and entire group, respectively. Previous investigations reported a TIVAP-related infection rate between 5.6% and 13% and an infection rate ranging from 0.15 to 0.39/1000 catheter-days in oncological patients [9, 10]. Our study showed a lower infection rate in the S group and an equal infection rate in the NS group compared with those in previous reports. The TIVAP devices used in the S and NS groups were made of silicone with the same size and shape; the only difference was the presence or absence of silver mixed in the catheter and tank. Patients who used TIVAPs made of silicone tended to have a lower infection rate than those who used TIVAPs made of polyurethane [11]. Therefore, TIVAP made of silicone mixed with silver may be more tolerant to infection than other products.

In this study, no GNR or Candida infections occurred in the S group, whereas the NS group showed 7 (23%) and 5 cases (16%) of 31 TIVAP-related infections caused by GNR and Candida, respectively. Neutropenia due to intensified antineoplastic chemotherapy, translocation of microorganisms from the gut to bloodstream due to total parenteral nutrition (TPN), and prolonged administration of broad-spectrum antibiotics can lead TIVAP-related infection by GNR and Candida [12]. The antimicrobial potential of silver cations depends on the composition and thickness of the bacterial external envelope. Gram-negative bacteria may be more susceptible to the antimicrobial effect of silver because they have thinner cellular walls than gram-positive strains [13]. This difference in bacterial structure may have reduced the GNR infection rate in the S group.

Several reports are available on silver-coated or silver-impregnated medical devices, such as central venous catheters (CVCs) [14, 15], urethral catheters [5], and implants for hip arthroplasty [6]. Silver-impregnated collagen cuffs significantly decrease the risk of short-term catheter colonization [16], but fail to prevent long-term colonization, possibly because of the early degradation of the cuffs [17]. For CVCs, although a prospective randomized trial showed results supporting the anti-infectious effect of silver-impregnated catheters [15], a meta-analysis disagreed its validity [14]. In the environment of CVCs, the access site is exposed and constantly contacts gram-positive bacteria on the skin, which may have compromised the antimicrobial effect of silver-coated CVCs, because silver has more effective antimicrobial activity against gram-negative infections. Additionally, the silver-mixed TIVAP, which was used in the S group, is speculated to be quite different from silver-coated or silver-impregnated medical devices because the silver-based inorganic antimicrobial agent is kneaded into the silicone material itself. This manufacturing difference may lead to more stable and effective antimicrobial activity in vivo relative to silver-coated or silver-impregnated devices.

Previous investigations have suggested that TPN, age over 65 years, hematological disease, and cancer with an oropharyngeal and pulmonary origin are risk factors for TIVAP-related infection [7, 11, 18,19,20]. Additionally, immunosuppressive status, such as neutropenia and long-term steroid usage, has been reported as a factor associated with infection [20,21,22,23,24]. In this study, the S-NS was the only factor significantly associated with TIVAP-related infections. Multivariate analyses showed no confounding factors with this relationship. Other factors, including TPN, steroid use, age, neutropenia and primary lesions, were not statistically significant.

This study has several limitations. First, this was a retrospective, single-center study with a relatively small sample size. Second, although the observation period for the S group was above the average days of infection in the NS group, a shorter observation period in the S group may have led to an underestimation of the infection rate. Third, more detailed statistical analyses, such as addition of confounding factors, including proficiency of the operators and surgical time; adoption of exclusion criteria, including immune status and origin of malignancy; and employment of propensity matched analyses; were not feasible in this study owing to the limited sample size. Thus, a larger number of study participants is desired.