Introduction

Cancer requiring chemotherapy is common. The National Chemotherapy Advisory Group estimated ~65,000 programmes/year, and Hospital Activity Data for England reported 425,000 deliveries of chemotherapy for cancer during the year 2009–2010 [1]. Centrally inserted external venous catheters (CIEVCs) and totally implantable ports (TIPs) are devices commonly used in the delivery of chemotherapy to cancer patients. CIEVCs were introduced in the 1970s and modifications such as the Dacron cuff by Hickman in 1979 helped improve their durability. TIPs were introduced in the early 1980s offering an alternative strategy removing the need for an external catheter [2]. Infective and mechanical problems plaque any long-term venous access device, which can lead to interruption of treatment, increased patient morbidity, and the need for premature device removal and replacement [3]. There is a general perception that TIPs carry a lower risk of complications than tunnelled central lines. There may also be other advantages resulting from the absence of an external line, such as patient acceptability and quality of life. However, TIPs are generally more complex to place requiring some tissue dissection and suturing. Each time a TIP is used, it must be needled for access, and the device is more costly than a CIEVC (prices vary from country to country but they are typically 3–10 times more expensive). Currently, there is no good evidence-based guidance to help health care providers and patients choose between these devices, and the decision is often based on local preference, availability of trained staff, cost pressures, duration of chemotherapy, and other unknown factors. This systematic review and meta-analysis aims to compare the complication profile and device-removal rate associated CIEVCs compared with TIPs in patients receiving chemotherapy for cancer.

Materials and Methods

A systematic review was performed according to the principles set out in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses [4]. The PRISMA checklist is provided in Table 2 in Appendix 1.

Eligibility Criteria

All prospective and retrospective studies that met the following criteria were included in the review:

  • study population included patients who received chemotherapy through central venous access devices for the management of solid or haematological malignancies;

  • studies that compared external catheters with internally implanted ports; and

  • clinical outcomes included device-related complications, such as infection, mechanical failure and device removal.

No specific exclusion criteria and no limitations on language of publications were applied.

Data Sources

An extensive search was performed on MEDLINE and EMBASE from inception of the analysis to December 2012. Relevant keywords and permutations of search terms relating to CIEVCs and TIPs were combined with those relating to cancer and chemotherapy; subsequently, validated search filters for randomised controlled trials, and observational studies were applied to the search output [5]. An example of the electronic search strategy (used in MEDLINE) is presented in Appendix 2. In addition, the database search was supplemented by hand-searching of the reference lists of all for the studies meeting the inclusion criteria, and a citation search was performed to identify all articles that cited relevant original studies using the Web of Science database.

Study Selection, Data Items, and Data Extraction

Two reviewers (S. K., O. W.) screened all titles and abstracts of the studies retrieved from the search and excluded duplicates and studies that clearly did not meet the eligibility criteria. Full texts of the remaining studies were retrieved for further review. Data extraction of all included studies was performed according to a predefined protocol. One investigator performed the data extraction, and a second reviewer independently validated the data extraction. Where multiple publications referring to the same population and study time periods were found, only the publication reporting on the most up-to-date results was included. Data were extracted from each study for: (1) study design and follow-up period; (2) characteristics of the patient population; (3) type of intervention and comparator, including details on catheter placement; (4) prophylactic antibiotic use; (5) type of outcomes measured; and (6) all numerical data on device-related complications.

Risk of Bias in Individual Studies

The quality, including potential risk of bias, of the individual studies was assessed. To maintain a consistency of reporting, a validated generic checklist designed for quantitative studies (randomised and nonrandomised) was used to assess the quality of all studies included in the review [6]. Any disagreement relating to inclusion of studies, data extraction, or quality assessment between the reviewers was resolved by consensus.

Data Analysis

Different approaches have been used to estimate the incidence of device-related complications. In some studies, the number of patients was the chosen unit of measure, and the proportion of patients with complications was reported. In other studies, the number of devices was the chosen unit of measure, and the proportion of catheters resulting in complications was reported; this allows all data for patients who underwent multiple catheter placements during the study to be taken into account. In a few studies, catheter-days was used as the chosen unit of measure, and the number of patients who experienced device-related complications, or the total number of complication episodes observed per 1,000 catheter days, was reported; this takes into account the possibility of multiple complications (e.g., repeat infections) associated with one device. For the purpose of analysis, the number of devices was used as the unit of measure, and it was assumed that for studies that only provided data on patients, each patient received only one device during the study.

For each study, individual complication outcomes, including device removal if reported, was summarised by its risk ratio. Where appropriate, meta-analysis was performed based on the random effects model. Pooled risks of CIEVCs relative to TIPs were calculated for each complication outcomes. The results were expressed as odds ratios (ORs) with values >1.0 indicating an increased risk of complications associated with CIEVC. All analyses were stratified by patient population (adults or paediatric) and study design [randomized controlled trial (RCT), prospective cohort study, and retrospective studies]. Heterogeneity between the studies was examined with standard Chi square test. In addition, the I 2 statistic was also calculated. The association between study size and results was examined in funnel plots by plotting ORs against their SE, and asymmetry was measured by the asymmetry coefficient. All analyses were performed using Stata version 11.0 (StataCorp, College Station, TX, USA).

Results

In total, 4,557 publications (1,812 RCTs and 2,745 observational studies) resulted from searching the electronic databases (Fig. 1) with 5 additional observational studies identified through hand-searching. After removing duplicate articles, 2,726 titles and abstracts were screened. Subsequently, the full text of the 39 potentially eligible studies were reviewed, of which 5 RCTs and 25 observational studies met the inclusion criteria and were included in the systematic review (Table 1).

Fig. 1
figure 1

PRISMA flow diagram of study identification and selection

Full size image
Table 1 Characteristics of the included studies
Full size table

In four of the RCTs included in the review, oncology patients were randomised to receive either a CIEVC or a TIP. The primary aim of the fifth RCT [7] was to compare heparin and urokinase as “locking agents” to maintain device patency. Although this study reported outcomes associated with CIEVC and TIP, urokinase was withdrawn from the market by the United States Food and Drugs Administration due to concerns regarding potential adverse outcomes, and this resulted in the trial terminating prematurely. Because patients were not randomised between the two devices, for the purpose of this systematic review, this study was categorised as a prospective cohort study. A second trial was also closed prematurely due to excessive bleeding in the TIP arm [8]. The four RCTs were performed in adults (age >15 years). The devices were inserted by either anaesthetists [8, 9] or by surgeons [8, 10, 11]. The management of the devices regarding prophylactic antibiotics use and the frequency of anticoagulant “locking” varied amongst the trials, e.g., between twice weekly to every 2 weeks for anticoagulant “locking.” The primary outcomes reported in these trials were catheter-associated complications, including infection, mechanical problems, thrombotic occlusion, and complications leading to device removal. Overall, the follow-up periods of the RCTs ranged from 13 to 30 months.

In terms of observational studies, 11 prospective (including the Dillon et al. RCT) [5, 12–21] and 15 retrospective studies [22–36] evaluated the use of venous catheters in different patient populations. 11 studies were in the adult population [14, 23–26, 28–34], of which 4 recruited only gynaecological patients [25, 26, 30, 34], 13 were performed in children (age ≤21 years) [7, 12, 13, 15–22, 35, 36], and 1 was performed in a mixed population (age 3–78 years); for the purpose of analysis, this study population was defined as adult [27]. Devices were inserted exclusively by surgeons in ten studies [12, 20, 23, 24, 28, 29, 32, 33, 35, 36], by surgeons or radiologists in two studies [22, 31], and by an oncologist or attending physician in 3 studies [25, 26, 34]. The primary outcomes of all of the studies were catheter-associated complications and complications leading to device removal with a particular emphasis on infection. Secondary outcomes included device-related anxiety and device acceptability by the patients and number of catheter days.

Assessment of Risk of Bias

Overall, the RCTs were well performed and fulfilled the majority of the quality assessment criteria partially or in full (Fig. 2). The major limitations were small sample sizes, particularly in two RCTs [8, 10], and the absence of any estimates of variance or uncertainty for the main results. Similarly, among the observational studies, ~50 % of the studies failed to report measures of uncertainty for the main results. Although the majority of the studies reported details of the analytical methods, only 40 % of the studies reported an estimate of variance and reported the results with sufficient details, and only 20 % of the studies controlled for potential confounding factors.

Fig. 2
figure 2

Quality assessment of included studies

Full size image

Device-Related Infectious Complications

All studies reported on device-related infectious complications, such as exit site, tunnel, port pocket, and catheter line infections as well as bacteraemia. However, one study did not report separate data for the two devices being compared and was excluded from the meta-analysis [13]. Overall all four RCTs, with the exception of the one terminated prematurely (due to bleeding complications) [8], reported a greater rate of infection in the CIEVC arm compared with the TIP arm. Meta-analysis of the data from the four RCTs showed that CIEVCs were associated with an increased risk of infection compared with TIPs; however, the difference was not statistically significant (OR 1.82; 95 % CI 0.93–3.55). Pooled analysis of the observational data supported the increased risk of infection with external catheters (Fig. 3). In particular, the increase in infection risk was greater among adult patients [OR 8.34; 95 % CI 6.14–11.32 (based on only one prospective study) and OR 3.45; 95 % CI 2.32–5.11 (based on pooled analysis of 12 retrospective studies)] than paediatric patients [OR 2.70; 95 % CI 1.91–3.82 (based on pooled analysis of nine prospective cohorts) and OR 1.74; 0.94–3.21 (based on pooled analysis of three retrospective studies)]. Overall, evidence of significant heterogeneity and moderate inconsistency was observed among retrospective studies of adult patients (p = 0.035, I 2 47 %) and among prospective studies of paediatric patients (p = 0.011, I 2 60 %).

Fig. 3
figure 3

Risk of infection

Full size image

Other Device-Related Complications

Only two of the four RCTs reported other device related complications [8, 11], and the pooled analysis showed that CIEVC were associated with a statistically significant increase in risk of these complications than TIPs (OR 3.27; 95 % CI 1.23–8.70). Similar findings were observed in paediatric patients, based on data from prospective cohort studies (OR 4.00; 95 % CI 2.00–8.01) [7, 17, 18, 21]. Although data from the retrospective studies showed an increased risk of other complications in both adult and paediatric groups in those with an CIEVC these results were not statistically significant, evidence of significant heterogeneity was also observed among retrospective studies of adult and paediatric patients.

Device Removal Due to Complications

Device removal due to infection and other complications was reported in the majority of studies. Most reported a statistically significant increase in the risk of device removal in patients with a CIEVC compared with a TIP. The risk associated with device removal was approximately three times greater in CIEVCs compared with TIPs in the four RCTs (OR 2.82; 95 % CI 1.06–7.52) (Fig. 4). The estimated magnitude of the risk observed was greater in the observational studies of adult patients, and one prospective cohort reported a six-fold increase (OR 6.36; 95 % CI 4.11–9.85). Retrospective studies reported an approximate fourfold increase (OR 3.79; 95 % CI 2.15–6.68). In the paediatric population, prospective cohort studies reported a three-fold increase in risk (OR 3.29; 95 % CI 2.24–4.82), and there was a two-fold increase in retrospective studies (OR 1.93; 95 % CI 0.84–4.43). Evidence of significant heterogeneity and inconsistency was observed among the randomised controlled trials (p = 0.054, I 2 61 %) and observational studies [p = 0.005, I 2 64 % (when pooling prospective studies in paediatrics); p = 0.000, I 2 74 % (when pooling retrospective studies in adults)].

Fig. 4
figure 4

Risk of device removal due to complications

Full size image

Patient Acceptability

Only four studies (two RCTs and two observational cohorts) reported on patient acceptability in relation to the use of these devices. One trial (Carde et al.) reported that although patients generally accepted either device, the patient activity rate (p = 0.02) and hygiene (p < 0.001) were better showed within the TIP cohort [7]. Similarly, Kappers-Klunne et al. [10] commented on a general preference for TIPs due primarily to the absence of an external device and less maintenance. In the paediatric population, one study reported that 9 (7 %) of the CIEVCs were removed at the patient’s request compared with none receiving a TIP [15]. Another study reported greater positive responses and fewer negative responses with TIPs than CIEVC when comparing ease of care, comfort, and overall acceptance [18].

Discussion

During the last three decades, many studies have suggested that CIEVCs are associated with an increased risk of device-related complications compared with TIPs. The validity of this “opinion” has until now not been confirmed by a systematic review or meta-analysis and may explain why the use of TIP use varies widely across the world. This systematic review and meta-analysis generally supports the view that there is an increased risk of infection, noninfectious complication, and complication-related device removal among patients with CIEVC compared with those with TIPs albeit with some caveats.

The reported risk of infection varied substantially between individual studies, and this remained the case irrespective of study type and population. One explanation may be the wide variance in the definition of infection across individual studies, e.g., some studies only included catheter-associated bloodstream infections, whereas others included bloodstream infection, bacteraemia, and exit-site infections. The pooled ORs from the meta-analysis of the RCTs were approximately two-fold; however, this did not achieve statistical significance. However, this may be the result of the small samples sizes (largest reported n = 106). Observational studies showed an approximate three-fold increase in the risk of infection among adult and paediatric patients (prospective studies only) with CIEVCs compared with TIPs. However, there is evidence of substantial heterogeneity among these studies.

Due to the limited data available, noninfectious complication described in this study is a composite outcome that included mechanical complications, thrombosis, and bleeding. Therefore, unsurprisingly, the reported risk of noninfectious complications varied between studies, and the pooled analysis was associated with substantial heterogeneity. Nonetheless, it is interesting to note that the risk of noninfectious complications associated with CIEVC use relative to TIP use is similar to that of infection. Similarly, device removal included removal as a result of any complications (i.e., infectious and noninfectious complications). In the adult population, regardless of study type, the evidence is consistent in showing a statistically significant three- to four-fold increase in the risk of device removal among CIEVC patients compared with to those receiving TIPs. Similar findings were found in the prospective studies of the paediatric population but not in the retrospective studies.

The main limitation of this study is the heterogeneity of the evidence base; therefore, the pooled estimate should be interpreted with caution. Although it is reasonable to conclude that the existing studies support that the risks of infectious, noninfectious complications, and device removal are greater with CIEVC compared with TIPs, the precise magnitude of these risks is less conclusive. As noted previously, studies included in this study vary in terms of population (e.g., age group, cancer type, and disease severity) and device-management protocols. An additional potential limitation is the 24-year time span (1988–2012) used for the literature search. However, there has been little significant modification of these devices since their inception. What has changed is the use of imaging to access the target vein and an expanding group of staff groups who now place these devices (anaesthetists, radiologists, nurse practitioners). However, we found no evidence that the age of participants in the studies (a surrogate for the age of the device) had any effect on the relative complication rates.

In the United States, TIP devices are the preferred choice for long-term venous access in oncology patients, whereas in the United Kingdom and continental Europe, CIEVCs play a central role. The decision-making processes behind this are complex and ill understood. In the United Kingdom, CIEVC placement is largely a nurse-led service, whereas TIPs are usually placed by doctors. A move away from CIEVC use to a TIP strategy will have an initial impact on both training and service delivery, which will need to be addressed. Interventional radiologists, anaesthetists, and surgeons will need to embrace change and support other groups (mainly nurse practitioners) in placing TIPs. The other major factor is the cost of the device itself; the purchasing costs of TIPs are substantially greater than the costs of CIEVCs. However, total costs associated with using a certain device should include not only the purchasing cost of the device but also the costs associated with device placement, management of complications, and replacement devices if necessary. Ng et al. [31] performed a simple cost analysis that compared total cost for both devices in the United Kingdom and estimated total costs of £1512/catheter for a CEIVC and £1483 for a TIP. In another study, Ross et al. [18] reported that the costs of both devices are comparable during the short-term, but potential cost savings may be achieved when TIPs are in use for >6 months. However, the decisions on adoption of health care interventions (pharmaceuticals and medical devices) should be informed by an extensive health technology assessment when both clinical benefits and cost-effectiveness are formally assessed.

This systematic review and meta-analysis has shown that CEIVC use is associated with an increased risk of infections, noninfectious complications, and need for device removal due to complications compared with TIPs. However, a formal health technology assessment on the clinical and cost-effectiveness of the use of TIPs compared with CEIVCs is needed to inform policy-makers on the relative value of investing in totally implantable devices compared with external catheters.