4.1. CRBSI Rates with PICCs vs. Other Central Catheters
In the present study we have shown that CRBSI rates were similar between PICCs and tunneled central catheters, but PICCs showed fewer infections compared to ports in prospective studies. Previous studies showed that PICCs had fewer infectious complications than other CVC when used for many intravenous therapies, as reported by Maki et al. [
34], whose meta-analysis showed that the rate of CRBSI was 1.0 per 1000 catheter days for PICC and 1.6 per 1000 catheter days for tunneled catheters. A recent meta-analysis that focused specifically on the rate of CRBSI in patients with HPN showed a lower rate of CRBSI in HPN patients using PICCs compared with tunneled central catheters [
11]. However, an analysis of single-arm studies in this same meta-analysis showed that the rate of CRBSI was comparable between both types of catheters. This meta-analysis included both prospective (
n = 3) and retrospective studies (
n = 2), with the consequent risk of bias modifying the results.
Of the two retrospective studies included in the latter meta-analysis, one by Elfassy et al. [
19] included 202 PICCs and 62 tunneled catheters reported a 1.96 and 1.93 CRBSI rates for these catheters, respectively, and the other by Christensen et al. [
10] included 126 PICCs and 169 tunneled catheters and reported 1.63 and 0.56 CRBSI rates for these catheters, respectively. Patient selection was done by identifying all adult patients with intestinal failure requiring HPN by clinic charts and electronic medical records from 2001 to 2008 [
19] or by local archives from 2008 to 2014 [
10] in a retrospective way. In the study by Elfassy et al. [
19], patient data were excluded from the analyses if the date of the line insertion was not documented or if the patients did not attend their clinic for a follow-up. In the study by Christensen et al. [
10], records were included for all patients who were dependent on HPN at that time, patients with finalized treatment, and patients who were deceased. Therefore, these different inclusion criteria may have biased the results. In addition, although in the study by Christensen et al. [
10] all catheters were single lumen ones, in the study by Elfassy et al. [
19], the type was not reported. Moreover, patients in the former study needed a total of 295 catheters over a 6 y period, which shows a higher rate of catheter replacement than in other studies.
Two other recently published retrospective studies were not included in the meta-analysis by Hon et al. [
11] and neither in our present study: one of them included 191 PICCs and only 11 tunneled central catheters with a 0.61 and 0.93 CRBSI rate per 1000 catheter days for these catheters respectively [
21], which could have biased our result in favor of PICCs. The other study included patients identified through medical records during a predefined 3 year period, reporting 123 PICCs and 51 tunneled central catheters with a 1.78 and 1 CRBSI rate per 1000 catheter days for these catheters, respectively [
9]. Although including these four recent published retrospective studies could have increased the statistical power of our meta-analysis, we considered that the risk of bias was high and thus decided to include only prospective studies.
The rate of CRBSI in the three prospective included studies did not differ after meta-analysis between PICCs and tunneled catheters. Although this is a consistent result with the data of the single arm studies of the previous published meta-analysis by Hon et al. [
11], we were not able to analyze the impact of the number of lumens of the catheters on the CRBSI rate. In fact, we have recently reported that CRBSI per 1000 catheter-days showed no difference between the PICCs and Hickman, but the rates were more frequent with multilumen catheters [
7]. This is also consistent with the results of Ross et al. in their recent report, in which patients with double-lumen catheters also had more CRBSI than those with PICCs or single-lumen catheters [
6].
On the other hand, our meta-analysis demonstrates that PICCs have less CRBSI than ports, which is also consistent with our previous results [
7] and also those reported by Ross et al. [
6]. Port systems are totally implantable venous silicone or polyurethane catheters with subcutaneous reservoir chambers made of titanium or ceramic. The port membrane is made of silicone and is only punctured with special port cannulae (non-coring port needles) [
8]. It is recommended that the port needle be replaced every third to seventh day in patients receiving HPN with cyclical nutritional application [
35]. Therefore, the use of the needle and its more frequent manipulation may increase the CRBSI rate compared to PICCs and tunneled catheters.
4.2. Non-infectious Catheter Related Complications
Catheter-related central vein thrombosis is a severe complication of HPN and may be a concern while choosing a type of catheter for delivering parenteral nutrition, especially for the long term. Cuerda et al. [
36] have recently reported a multicenter prospective study of sixty-two patients (31 males, 31 females) aged 50 ± 19 years, who followed-up for a median 363 days, for whom the study found an incidence of catheter-related thrombosis of 0.045 per catheter/year. In this study, the thrombosis rate was not significantly associated with any of the variables analyzed, including the type of catheter [
36]. This is in agreement with the results of our present study, although there was a trend for more episodes of catheter-related thrombosis with PICCs than with tunneled catheters, this was not significant after the meta-analysis.
In fact, of the three included studies in our meta-analysis, only one showed a significant incidence of greater thrombosis with PICCs (Toure et al. [
18]). In this study, the specific care protocol of the catheters was not reported in detail (and the number of single lumen or multilumen catheters was also not reported), and taurolidine-citrate locks were injected in 35% of patients, but for the rest, the flushing compound was not reported. Furthermore, ultrasound was employed to diagnose catheter-related thrombosis only in patients presenting with symptoms like arm swelling, pain, loss of function, and head or neck swelling [
18]. The different care protocols in the three studies may have produced differences in the rate of catheter-related thrombosis, as good catheter care requires an experienced multidisciplinary team trained to identify and aggressively treat catheter-related complications [
36].
We found no differences in the mechanical complications between PICCs and other tunneled catheters, and this result was consistent and homogeneous for the three included studies in our meta-analysis. Therefore, the choice of the type of catheter for HPN need not rely on the risk of catheter-related thrombosis or other non-infectious complications.
4.3. Limitations
As we identified only three prospective studies after the systematic review for inclusion in the meta-analysis, there is a limitation in terms of statistical power. Further, the allocation of patients to one or another type of catheter was not randomized, and the selection of the catheters for each patient was performed on a clinical basis with a consequent risk of bias. In the study by Santacruz et al. [
7], the choice of the CVC was not randomized but based on the patient´s responsible physician, always taking into account the underlying disease, the expected duration of HPN, and the possibility of a safe procedure for obtaining a venous access. In the study by Toure et al. [
18], the choice of CVC was made by the patient’s physician; a PICC catheter was preferred for patients initially expected to have HPN for fewer than 6 months, and Broviac was preferred for those patients initially expected to need HPN for more than 6 months. In the study by Cotogni et al. [
17], the choice of the type of catheter in each patient was largely based on the provider´s preference, since at the time of their study, there was no official hospital policy or recommendation.