Summary of findings
Description of the condition
Since its introduction over three decades ago, the central venous catheter (CVC) has been an essential device in managing patients with both acute, life-threatening illnesses and chronic conditions such as cancer. Compared to peripheral catheter, CVC is inserted deeper into the larger veins of the body, providing a more secured and durable intravenous access. While the CVC can be used for administering various medications, large amounts of fluids, and total parenteral nutrition, there are significant risks to using this device. The major concern with the CVC is colonization by micro-organisms which in turn could lead to catheter-related bloodstream infection (CRBSI), which is associated with increased morbidity, mortality, and healthcare costs (CDC 2011; Cicalini 2004; Saint 2000). While the incidence of CRBSI varies depending upon the patient population evaluated and adherence to recommendations based on up-to-date and high quality clinical evidence, CRBSI remains an important patient safety problem in both developed and developing countries (Norwood 1991; Pronovost 2006; Rosenthal 2006; Saint 2000). For instance, in the United States National Healthcare Safety Network (NHSN) report in 2010 that covered 2473 hospitals, there were nearly 11,000 cases of laboratory-confirmed CRBSI, with estimated CRBSI rates of up to 3.5 per cent (NHSN 2010).
Description of the intervention
Several methods have been evaluated to prevent CRBSI, including the use of maximum sterile barriers, namely cap, mask, sterile gown, gloves and full-sized sterile drapes during catheter insertion (Hu 2004), chlorhexidine gluconate rather than povidone-iodine for CVC site disinfection, and avoidance of the femoral site for catheter insertion (CDC 2011; Chaiyakunapruk 2002; Gnass 2004; Raad 1994). Additionally, modifications of the CVC itself, in the form of antimicrobial impregnation, coating, or bonding, have also been used to prevent CRBSI (Cicalini 2004). Antimicrobial is a general term to describe an agent that either kills or inhibits the growth of micro-organisms, which include bacteria, fungi, viruses or parasites (CDC 2010). Currently, two major types of antimicrobial agents are used as CVC coating: antiseptic and antibiotic. Antiseptic refers to an agent that destroys or inhibits the growth of a range of micro-organisms in general that are present in or on living tissues (e.g. hand washes or surgical scrubs), while antibiotic refers to an agent that acts in similar fashion as antiseptic, but targets selective micro-organisms, especially bacteria, and works generally in low concentrations (McDonnell 1999). Various forms of antiseptic and antibiotic catheter impregnation have been introduced since the late 1980s, including chlorhexidine-silver sulphadiazine (C-SS) and minocycline-rifampicin (MR) impregnation which are the most commonly used and studied (Falagas 2007; Mermel 2001). For the early C-SS-impregnated catheters, impregnation was only applied at the external surface, whereas MR impregnation is applied to both external and luminal surfaces. More recently, second-generation C-SS-impregnated catheters have been introduced, with impregnation both at the external and luminal surfaces of the catheter (Ramritu 2008).
How the intervention might work
It is proposed that these compounds with well-established antimicrobial properties inhibit the colonization of micro-organisms like bacteria on the catheter surface, which in turn prevent the spread of these micro-organisms into the bloodstream (Cicalini 2004). Several other compounds which demonstrated antibacterial activities in vitro, like silver, platinum, carbon and heparin have also been evaluated as CVC impregnation materials in clinical studies (Abdelkefi 2007a; Hanna 2006; Khare 2007). Silver and platinum were found to inhibit bacterial cell growth and division (Jung 2008; Rosenberg 1967), while heparin was proposed to reduce bacterial growth via a prevention of fibrin deposition and thrombus formation in the catheters (Abdelkefi 2007). Carbon nanotubes had in turn been shown to cause cell wall damage to bacteria in direct contact (Kang 2007), and when combined with platinum and silver seemed to enhance their overall antibacterial properties (Narayan 2005). Initial in vitro and animal studies revealed the effectiveness of some of these impregnated catheters against certain common colonizing micro-organisms (Raad 1995; Raad 1996).
Why it is important to do this review
While promising, the effectiveness of these new catheter-based technologies was challenged by the progressive discovery of different types of colonizing bacteria, different factors that facilitate their adherence to the catheter and changes in their sensitivities to antibiotics over time (Raad 2002). Furthermore, the antibacterial activities of these modified catheters have been found to diminish after a period of use (Sampath 2001; Schmidt 1996; Yorganci 2002). Despite official recommendations regarding when these modified catheters should be used (CDC 2011), various systematic reviews yield discrepant findings (Gilbert 2008; Ramritu 2008; Veenstra 1999a; Walder 2002), reflecting a need to provide ongoing up-to-date collective evidence on the clinical impact of these modified catheters to inform current practice and direct future research. The benefits of these modified catheters in different hospital settings, e.g. intensive care units, standard wards and oncology units, also demand evaluation. Although people cared for at home with CVC in place, like cancer patients, constitute an important population in terms of catheter care, it is unrealistic to expect them to be participants in such studies.
In this review, we aimed to assess the effectiveness of antimicrobial impregnation, coating and bonding on CVC in reducing catheter-related infections in adults. We also assessed their safety and cost effectiveness where possible.
We aimed to assess the effectiveness of antimicrobial impregnation, coating or bonding on CVCs in reducing catheter-related infections in adult participants who required central venous catheterization, along with their safety and cost effectiveness where data are available. Specifically, we undertook the following comparisons:
- Catheters with antimicrobial modifications in the form of antimicrobial impregnation, coating or bonding, against catheters without antimicrobial modifications.
- Catheters with one type of antimicrobial impregnation against catheters with another type of antimicrobial impregnation.
- Catheters with any type of antimicrobial impregnation against catheters with other antimicrobial modifications, e.g. antiseptic dressings, hubs, tunnelling, needleless connectors or antiseptic lock solutions.
Additionally, we examined with subgroup analyses the effectiveness, safety and cost effectiveness of these catheters in participants for whom they were intended for use over the short term (< 10 days) versus the long term, in studies using different types of catheter impregnation, (e.g. chlorhexidine-silver sulphadiazine (C-SS) or minocycline-rifampicin (MR)) in the experimental arm, in studies performed in settings with different levels of care (e.g. intensive care unit, standard ward and oncology unit) and in participants with different baseline risks, in studies with consensus definitions of clinical sepsis versus other definitions, in studies that examined these modified catheters with and without co-interventions, and in studies that assessed cost effectiveness in different units of measurement (e.g. different currencies).
Criteria for considering studies for this review
Types of studies
We included randomized controlled trials (RCT), quasi-randomized trials and cluster-randomized trials comparing CVCs with antimicrobial impregnation, coating or bonding with catheters without these modifications. We excluded cross-over studies. We also excluded studies assessing CVCs for haemodialysis, as this is covered by another Cochrane title (McCann 2010).
Types of participants
We included studies with participants cared for in the adult inpatient unit in a hospital setting (ICU and non-ICU) with a CVC in place. We accepted studies that enrolled a participant more than once. We addressed the issues arising from multiple enrolments using the approach detailed under the heading of Unit of analysis issues. We excluded studies of children because there is another Cochrane title that includes neonates and children as participants (Shah 2008).
Types of interventions
The use of CVCs with antimicrobial impregnation, coating or bonding.
The main types of catheter impregnation were as follows:
- Chlorhexidine-silver sulphadiazine (C-SS).
- Minocycline-rifampin/rifampicin (MR).
- Others such as heparin and silver, platinum and carbon impregnation.
The use of standard CVCs of matching material and design without antimicrobial modifications.
We would have included studies comparing catheters with antimicrobial impregnation and the use of catheters with the following modifications or procedures if such studies were available:
- Antimicrobial-impregnated dressings.
- Silver iontophoretic device.
- Antiseptic-filled catheter hubs (including iodinated alcohol or povidone iodine).
- Needleless connectors.
- Antimicrobial lock solutions.
We also included studies comparing one type of impregnation to the other (C-SS versus MR).
Each participant should only have one study catheter at any one time during the study. We attempted to identify whether any participant had multiple catheters concurrently from the descriptions of the participants in the methods and the results sections if such information was available. We did not place any limit on the minimum and maximum catheter indwelling time for each study.
Types of outcome measures
The following outcomes were measured during the indwelling time of the CVCs or at their removal, or in the case of patient-level outcomes such as all-cause mortality and length of hospital stay, throughout the period in which the participants were being observed for the purpose of research whether or not the CVCs were still in place.
- Number of participants with clinically diagnosed sepsis. We used the diagnostic criteria developed from the 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference (Levy 2003), as detailed in Appendix 1. This set of diagnostic criteria contains an extensive list of clinical features and investigation findings, with no clear statement on the minimum number or thresholds required to satisfy a diagnosis of sepsis. Therefore, we accepted various definitions adopted by the author of each study, as long as the items included in their definitions were those contained in this set of diagnostic criteria. We would, however, also accept definitions that were not consistent with this set of diagnostic criteria, provided the authors justified their definitions with validated sources. We would then analyse as subgroups those studies that followed such diagnostic criteria and those that adopted other definitions.
- Number of participants with laboratory-proven catheter-related bloodstream infection (CRBSI), defined as isolate of the same organism from a semi-quantitative or quantitative culture of a catheter segment and from separate percutaneous blood cultures, with no other identifiable source of infection (CDC 2011).
- All-cause mortality.
- Number of participants or catheters with catheter-related local infections, including exit site and tunnel infection: defined as isolate of organism from a semi-quantitative or quantitative culture of a catheter segment, with clinical signs of infection around the insertion site (CDC 2011).
- Catheter colonization: number of participants or catheters with positive catheter cultures: defined as any positive semi-quantitative or quantitative culture from a proximal or distal catheter segment (CDC 2011).
- Number of participants or catheters with resistant organisms from catheter cultures.
- Number of participants or catheters with skin or site colonization: defined as any positive semi-quantitative or quantitative culture from the skin around the catheter site (CDC 2011).
- Mortality from catheter-related bloodstream infection, defined using diagnostic criteria as stated in the Primary outcomes (see number 2).
- Number of participants or catheters with adverse effects: including skin irritation/contact dermatitis, thrombophlebitis, thrombo-embolism and anaphylaxis.
- Number of participants or catheters with catheter failure or premature catheter removal.
- Use of systemic antibiotics: total courses of systemic antibiotics used during hospital stay or number of participants who required systemic antibiotics during the course of the study.
- Length of hospital stay.
- Cost of care, including the costs associated with the material and the number of catheters used or medication given (e.g. antibiotics).
- Quality of life, measured using validated scales like a disease-specific adapted quality of life tool.
We also assessed whether the methods of outcome measurement, in particular the laboratory methods and survey tools such as the Quality of Life instrument have been previously validated by evaluating whether the authors cited relevant literature on the use of such tools. If there were a large number of studies that adopted non-validated tools in measuring their major outcomes, we would further explore the differences in the effect estimates between these studies and the studies that adopted previously validated tools via a sensitivity analysis.
Search methods for identification of studies
See: Cochrane Anaesthesia Review Group (CARG) methods used in reviews.
We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library, issue 3, 2012), MEDLINE (OVID SP) (1950 to March 2012), EMBASE (OVID SP) (1980 to March 2012), and CINAHL (1982 to March 2012) databases. We reran the search to May 2013 shortly prior to the publication of the current version of the review and placed the studies of interest from our updated search in the Studies awaiting classification section. We will deal with these studies when we next update the review.
We employed the search strategy as stated in the Cochrane Handbookfor Systematic Reviews of Interventions (Cochrane Handbook). Our detailed search strategies for MEDLINE (OVID SP), CENTRAL, EMBASE (OVID SP) and CINAHL are displayed in Appendix 2, Appendix 3, Appendix 4 and Appendix 5 respectively.
We also searched for ongoing clinical trials and unpublished studies via Internet searches on the following sites:
We did not apply language or publication restrictions.
Searching other resources
To identify further potential studies, we examined references cited in previous relevant Cochrane reviews, in other relevant studies, review articles and standard textbooks. We assessed handsearch results from the CARG. We also sought relevant information from expert informants on additional published and unpublished studies.
We accepted studies whether published or unpublished, in full article or abstract form, as long as assessment of study quality was possible and where the other inclusion criteria were fulfilled. If studies were published as abstracts, we would contact the study authors for further information if necessary.
We contacted authors of all studies identified to be relevant where possible, to clarify details of reported follow-up studies where necessary, or to obtain any information about long-term follow-up where none had been reported, and to enquire about additional studies potentially suitable for inclusion.
Data collection and analysis
Selection of studies
We used the standard method of The Cochrane Collaboration as described in the Cochrane Handbook Chapter 7, section 2, and referred to the CARG's guidelines where appropriate. Two authors (NML and NC) independently performed the first round of searching for studies that appear to be relevant. Two authors (NML and WP) then screened these studies for inclusion in the review, using predefined inclusion and exclusion criteria in selecting eligible studies and determining their risks of bias, as detailed under Assessment of risk of bias in included studies. We resolved any disagreement by discussion leading to a consensus.
Data extraction and management
We extracted the following data from each included study: study characteristics, information relating to the risks of bias, outcomes assessed and data for each outcome that were relevant to this review. We used a standard data collection form from the CARG for this purpose. One review author (NML) first entered all the data from the included studies. The data were then cross-checked independently by other co-authors (WP and EOR, NAL and SS) for accuracy. Any possible inaccuracy in the data was communicated with the first author (NML), which led to amendments of the data if necessary. We also independently screened for duplicate entry of participants in each study by matching the initial number recruited against the total number along each step in the conduct of the study.
For studies with multiple comparisons, e.g. antimicrobial-impregnated CVCs versus non-impregnated CVCs versus CVCs with non-catheter-related hygiene measures, we included only interventions that were relevant to this review (antimicrobial-impregnated CVCs versus non-impregnated CVCs). If there were more than two intervention groups that were relevant to this review, e.g. antimicrobial A-impregnated CVCs versus antimicrobial B-impregnated CVCs versus non-impregnated CVCs, we combined the intervention groups into a single pairwise comparison (combining antimicrobial A- and antimicrobial B-impregnated groups combined versus non-impregnated group), as detailed in the Cochrane Handbook.
For studies in which data were given only in percentages, we converted the percentages into the nearest round numbers by multiplying the percentages in fraction by the total number of participants analysed in the assigned group.
We assessed the definition of each outcome in the included studies. Some studies contained outcomes that were relevant to this review, but labelled them differently, e.g. a study could assess an outcome that matched our definition for 'catheter colonization', but the study authors labelled this outcome as 'catheter-related infection'. In such cases, we allocated the data from the outcomes concerned to our prespecified outcomes that best matched the study authors' outcomes in definition.
We resolved any disagreements by discussion among the authors.
Assessment of risk of bias in included studies
Two authors (NML and EOR) independently assessed each included study against the following criteria, using the methods as detailed in the Cochrane Handbook:
- Adequacy of sequence generation
- Adequacy of allocation concealment
- Completeness of follow-up and handling of incomplete outcome data
- Blinding of participants and care providers to intervention (medical and nursing staff who insert the catheters and are involved in the participants' day-to-day care) where possible. Statements in the study or clarification by the authors on whether the catheters compared are indistinguishable in appearance, and hence blinding is possible, were sought
- Other issues (e.g. validity, reliability, objectivity, or blinding of outcome measurement, and whether there was extreme baseline imbalance).
The assessors assigned a judgment of low, high or unclear risk of bias for each item.
A detailed description of the criteria is available in Appendix 6.
As an additional measure to assess the risk of performance bias, we looked for evidence in each study that a standard protocol was followed by all groups under study for insertion, use, maintenance and removal of CVCs, and for concurrent use of catheter-related antiseptic measures (including the use of prophylactic antibiotics) and sterile procedures in all groups. We made relevant comments in the corresponding tables for each study.
Measures of treatment effect
For categorical data, we pooled outcome estimates that were measured using the same scales with risk ratios (RR), absolute risk reduction (ARR) and number needed to treat for an additional beneficial outcome (NNTB) for each specific comparison, with their respective 95% confidence intervals (Cochrane Handbook). As the continuous data were provided by single studies, we expressed the results using mean difference (MD) with 95% confidence interval. Where pooled analyses were not possible we report the results of the individual studies separately.
Unit of analysis issues
We assessed unit of analysis issues in the included studies in two possible ways in which they might arise: 1. Multiple enrolments of the same participants from either individually randomized trials or cluster-randomized trials; and 2. Clustering at the level of the enrolled units in cluster-randomized trials.
- Unit of analysis issues might arise if there were multiple enrolments of the same individual following a need for repeated catheterization. We addressed this unit of analysis issue by first assessing each included study for any evidence of multiple enrolments. If we found evidence of this, e.g. the number of catheters exceeding the number of participants, we assessed whether there were any participants with more than one event reported, provided there was sufficient information in the paper for us to do so. We then excluded those with multiple enrolments by entering the data for those who were enrolled only once. However, if such information was not available, we performed our analysis based on whatever data the authors provided, using the total number of catheters as the denominator.
- In dealing with cluster-randomized trials, we would have looked for evidence that the authors had made appropriate adjustments in their analyses in the methods and results. We would also have inspected the width of the standard error (SE) or 95% confidence interval (CI) of the estimated treatment effects. If we had found an inappropriately small SE or a narrow 95% CI, we would have asked the authors of the study to clarify the unit of analysis.
If we found a unit of analysis error that was correctable with the information provided by the authors, e.g. when the included study analysed outcome data for individual participants without adjusting for the effects of clustering, we would have performed our own adjustments. We would do this by adjusting the final estimates of the study, using the methods detailed in the Cochrane Handbook, i.e. by multiplying the SEs of the final effect estimates by the square root of the 'design effect': (1 + (M-1) x ICC), where M was the average cluster size (number of participants in the units being studied) and the ICC was the intracluster correlation coefficient among participants within each unit. We would determine the average cluster size (M) from each trial by dividing the total number of participants by the number of units recruited. We would have sought the best estimate of ICC from reliable resources such as landmark cluster-randomized trials on central venous catheters where such trials were available. We would then have combined the adjusted final effect estimates from each trial with their SEs in the meta-analysis using generic inverse variance methods in Review Manager 5. Should we fail to identify a reliable ICC for the relevant cluster-randomized trials, we would have used the unadjusted estimates as reported by the study authors for our meta-analysis, noting the absence of an appropriate adjustment. We would also have performed a sensitivity analysis to assess how the overall results were affected with and without the inclusion of these studies.
We addressed the unit of analysis issues that might arise from multiple comparisons by combining all the intervention groups into a single combined intervention group to achieve a single pairwise comparison, as detailed under Data extraction and management.
Dealing with missing data
We obtained drop-out rates from each study. We considered a drop-out rate higher than the control group event rate to be significant. If we found a significant drop-out rate with no reasonable explanation, we contacted the authors of the individual studies where possible, to request further data. We also assessed whether an intention-to-treat analysis was performed.
We performed sensitivity analyses to assess how the overall results were affected with and without the inclusion of those studies with a high risk of attrition bias from incomplete outcome data.
Assessment of heterogeneity
We assessed the treatment effects of individual trials and the heterogeneity between trial results by first inspecting the forest plots.
We explored clinical heterogeneity by assessing clinical and methodological characteristics of the included studies (e.g. difference in study quality, participants, intervention or outcome assessment). We only attempted to pool data in a meta-analysis if there was negligible clinical heterogeneity among the selected studies. We would in the process decide whether to exclude some studies altogether from the meta-analysis, if major discrepancies in clinical or methodological characteristics were found, or to include them and perform a sensitivity analysis of the main outcome.
In addition, we used the I² statistic (Higgins 2002) to measure inconsistency in the study results, with a value greater than 40% indicating substantial statistical heterogeneity. If significant statistical heterogeneity was found but the studies were considered suitable to be combined for a meta-analysis based on the clinical and methodological characteristics as detailed above, we relied on the pooled effect estimates provided by the random-effects models.
Assessment of reporting biases
For each study, we compared the outcomes reported in the results against the outcomes listed in the methods section. We also identified some key outcomes that might have been assessed but were not included. We contacted the study authors for clarification where necessary. In studies in which critical outcomes were missing, we sought the study protocol, either from PubMed, the relevant trial registry, the web link provided by the study or directly from the study authors, to establish whether these outcomes had been prespecified. In addition to our description under 'reporting biases' in the risk of bias assessment tables (Characteristics of included studies), we present a matrix highlighting those studies in which there were discrepancies between the major outcomes listed in the methods versus those reported in the results, and also those studies in which there were critical outcomes that were not reported at all (Appendix 7).
Where possible, we also performed a sensitivity analysis taking an outcome that was reported by all studies, and comparing the overall results with and without inclusion of those studies in which key outcomes were missing.
Assessment of publication bias
We screened for publication bias by using a funnel plot if there were sufficient number of studies (at least 10) included in the analysis. If publication bias was suspected, i.e. significant asymmetry was found after a visual inspection of the funnel plot, we included a statement in our results and the summary of findings table with a corresponding note of caution in our discussion.
We followed the procedures of the Cochrane Anaesthesia Review Group. We performed meta-analysis of the included trials with Review Manager 5 (Revman 5.2), using a fixed-effect model, unless significant statistical heterogeneity was found, as detailed under the previous heading of 'Assessment of heterogeneity'. We used intention-to-treat data if possible in our analyses.
We first presented the effects of antimicrobial-impregnated CVCs versus non-impregnated CVCs as a whole in our meta-analysis. Since there might be differences in the effects of different types of antimicrobial impregnations, we also reported the effects of each specific type of impregnation with our subgroup analyses, as stated under the next heading.
For rate data such as CRBSI per 1000 catheter days, we followed the methods in Chapter 9.4.8 of the Cochrane Handbook. We first obtained the rate ratio by dividing the rate in the intervention group by the rate in the control group. We then derived the natural log (ln) of the rate ratios and entered these into RevMan using the generic inverse variance method. We obtained the standard error (SE) of the ln(rate ratio) by the following formula: SE of ln (rate ratio) = square root of ((1/rate of the intervention group) + (1/rate of the control group)). For the study in which we combined two intervention groups (Arvaniti 2012), we obtained the adjusted rate data of the combined group by apportioning the reported rates of the original individual group according to its total catheter days as a proportion of the combined total catheter days, as represented by the following formula: adjusted event rate (per 1000 catheter days) = ((E₁ x CD₁/CD₁+₂) + (E₂ x CD₂/CD₁+₂)), where E₁ = event rate (group one), E₂ = event rate (group two), CD₁ = total catheter days (group one), CD₂ = total catheter days (group two), CD₁+₂ = total combined catheter days.
Subgroup analysis and investigation of heterogeneity
We performed subgroup analyses if applicable for the following:
- Participants with CVCs intended for use over the short term (< 10 days) versus a long-term period (10 days or more).
- Studies using different types of catheter impregnation (e.g. C-SS or MR) in the experimental arm against unimpregnated catheters.
- Studies in different settings or with a certain type of patient as the predominant participants, e.g. those in intensive care units (ICUs), people receiving cancer treatments, those on long-term parenteral nutrition, those requiring CVCs for other purposes, and studies with a mixture of different types of participant.
- Studies in which the participants had higher or lower baseline risks, using the median event rates in the control group as cut-offs.
- Studies that adopted the definition of clinical sepsis developed from the 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference (Levy 2003) versus studies with other definitions.
- Studies with and without co-interventions (e.g. concurrent antiseptic device or procedures such as special dressing, hub, cutaneous antisepsis or the use of prophylactic antibiotics).
- For studies that examined cost effectiveness, those conducted in different countries with different currencies as the unit of measurement are analysed in different subgroups.
We performed sensitivity analyses on four major outcomes, i.e. our three primary outcomes of clinically diagnosed sepsis, CRBSI, and all-cause mortality, and the most frequently reported secondary outcome of catheter colonization. We conducted our sensitivity analyses based on two main criteria, namely, the risks of selection bias based on random sequence generation and allocation concealment, and the risk of attrition bias, as described under the headings of Assessment of risk of bias in included studies and Dealing with missing data respectively.
Description of studies
Results of the search
The initial search yielded 391 records from MEDLINE, 624 records from EMBASE, 157 records from CENTRAL and 95 records from CINAHL, giving a total of 1267 records. We performed additional searches in other Internet resources (as detailed in Appendix 8), and identified three further studies that appeared to be relevant, two of which have been completed (Antonelli 2011; Pachl 2010) and one still on-going (Steinberg 2009). We reran the search to May 2013 and identified further 403 records (MEDLINE: 221 records, EMBASE: 147 records, CENTRAL: 22 records and CINAHL: 13 records).
After removing duplicates of all records from our searches, there were 1125 records in total. We short listed 113 articles that appeared to be relevant after inspecting the titles, including the three articles mentioned above and three further articles that we have identified through our updated search to May 2013 (Antonelli 2012; Casey 2012; Hitz 2012). On further inspection of the abstracts, we excluded 41 articles and placed Steinberg 2009 under 'Ongoing studies' (see Characteristics of ongoing studies). From the remaining 71 articles, we obtained the full text of 65, and excluded a further nine based on the full text, leaving 56 studies to be included in our meta-analysis. We did not identify any additional titles that were considered relevant after checking the reference lists of all the full-text articles. Six studies (Antonelli 2011; Guggenbichler 1998; Pachl 2010; Casey 2012; Hitz 2012), including three that we have identified from our updated search, were not assessed in detail as we have not yet acquired full-text versions, and they were placed under 'Studies awaiting classification' (see Characteristics of studies awaiting classification). Among them, Antonelli 2012 appears to be a published full paper of Antonelli 2011, although we will confirm this when we next update the review.
The flow diagram of the studies from the initial search to the meta-analysis is shown in Figure 1. A description of all the included studies is displayed in the Characteristics of included studies table, and the excluded studies with the reasons for exclusion are given in the Characteristics of excluded studies table.After cross-searching MEDLINE, EMBASE, CENTRAL and CINAHL, we could not trace the two completed studies, suggesting that they have not been published.
|Figure 1. Study flow diagram.|
The 56 included studies are randomized controlled trials (RCTs) conducted in 17 countries, including the USA (20 studies), Germany (nine studies), UK (six studies), Spain (three studies), Australia, Austria, France, Sweden, Turkey (two studies each), Belgium, Brazil, Greece, Italy, Netherland, South Africa, Taiwan and Tunisia (one study each). Thirty-three trials are single centre RCTs and 23 are multicentre RCTs. We did not find any cluster-randomised trials among our included studies. The initial sample sizes of the studies range from 20 (Bach 1996b) to 960 participants (Walz 2010). Some studies only specified the number of catheters evaluated and not the number of participants (Darouiche 1999; Darouiche 2005; Fraenkel 2006; George 1997; Leon 2004; Maki 1988; Maki 1997; Mer 2009; Ostendorf 2005; Van Vliet 2001). In 24 studies, the minimum age for the participants was clearly stated. Among these, 21 studies included participants aged at least 18, and for the remaining three studies the minimum age for inclusion was 17 (Bennegard 1982), 12 (Collin 1999) and four years old (Abdelkefi 2007). Eleven studies did not provide the minimum age for inclusion but stated that their participants were "adults". For the remaining 21 studies, the minimum age for inclusion was not stated. In two studies (Bach 1996b; Bong 2003), participants were predominantly adult men, while all other studies included participants of both sexes in significant proportions.
Thirty-four studies were conducted in medical/surgical ICU settings, 10 studies in haematology/oncology units, eight studies enrolled a mixture of participants including patients from ICU, general medical or surgical units and those receiving total parenteral nutrition (TPN), three studies enrolled only participants receiving TPN, and one study (Bennegard 1982) had no description of the study setting or participant type.
There were three major categories of intervention:
- Two-arm comparison between antimicrobial impregnation and no impregnation (47 studies);
- Two-arm comparison between different catheter impregnations (five studies);
- Three-arm comparison between different impregnations with or without a non-impregnated group (four studies). A total of 11 antimicrobial impregnations were assessed, including chlorhexidine-silver sulphadiazine (C-SS), minocyclin-rifampicin (MR), miconazole-rifampicin, single antibiotics such as vancomycin, teicoplanin and cefazolin, silver-platinum-carbon, silver, silver-impregnated cuff, heparin and benzalkonium. One study (Rupp 2005) described the C-SS impregnation used as second generation impregnation. There was no evidence from any of the included studies that any participant had multiple study catheters concurrently, although there were participants who had multiple study catheters placed sequentially.
Catheter colonization was the most commonly evaluated outcome (49 studies), followed by catheter-related bloodstream infection (CRBSI) (45 studies). The major clinical outcomes of clinically diagnosed sepsis, mortality attributed to catheter-related infections, and all-cause mortality were assessed in 13, 5 and 11 studies respectively. Adverse effects were evaluated in 13 studies. There were wide ranges of baseline risks in the included studies, from 0.4% to 58% for clinically diagnosed sepsis, 0% to 40% for CRBSI, 8% to 59% for all-cause mortality and 12% to 80% for catheter colonization. Out of the 13 studies that assessed our primary outcome of clinically diagnosed sepsis, eight studies defined this outcome in accordance with the definition in this review. In another four studies, the authors did not provide sufficient information about the definition, and one study defined it in a way that was considered to be outside the scope of our definition for this review. Overall, the definition of CRBSI was consistent among the studies, which included suggestive clinical features, and a positive catheter culture with a positive blood culture growing the same organism. Most studies used previously validated laboratory methods to perform catheter and blood cultures, and adopted microbiological definitions for colonization and bloodstream infection that were consistent with the published literature in the evaluation of catheter-related infections. All studies reported catheter-related outcomes such as CRBSI and catheter colonization using the catheter as the unit, and none provided separate reports of these outcomes with participants as the unit. Thirty four studies provided the number of catheters as well as participants. The number of catheters matched the number of participants in 32 studies, and the number of catheters exceeded the number of participants by only one in two studies, suggesting that except for one participant who had two catheters evaluated, all participants had a single catheter. In terms of participant-level outcomes, 12 studies reported clinically diagnosed sepsis, nine reported all-cause mortality, five reported mortality attributed to CRBSI and 10 studies reported adverse effects. None of the included studies assessed quality of life in their participants.
We excluded 47 studies based on one or more of the following:
1. Study design (29 studies): the studies were either retrospective or prospective cohort studies, before-and-after intervention studies, prospective non-randomized intervention studies, meta-analyses, economic analyses with no original trial data, in-vitro experiments, or commentaries.
2. Population (18 studies): the participants in the studies were either children, people undergoing haemodialysis/extracorporeal detoxication or neurosurgical patients undergoing cerebral ventricular catheter placement.
3. Intervention (six studies): the studies either assessed antimicrobial-impregnated dressing or cerebral ventricular catheter.
A description of each study is available in the Characteristics of excluded studies table.
Risk of bias in included studies
The majority of the studies had either low or unclear risks of bias in most criteria, except for blinding, which did not appear possible for the participants and carers in most studies, due to the different appearances of the catheters evaluated. The risk of bias graph, which shows the overall degree of risks of bias in the studies included in this review by depicting the proportions of studies with low, high and unclear risks of bias according to each criterion, and the risk of bias summary, which details the risk of bias of each included study, are illustrated respectively in Figure 2 and Figure 3. A detailed description of the risk of bias in each study is provided in the Characteristics of included studies. Following are summaries of our risk of bias assessment under each major criterion.
|Figure 2. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
|Figure 3. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
For random sequence generation, 26 out of 56 included studies had low risk of bias. For allocation concealment, 15 studies had low risk of bias. Three out of 56 included studies had high risks of bias in both random sequence generation and allocation concealment. A large proportion of the studies had unclear risks of bias for these two criteria and for all other criteria, due mainly to a lack of information reported by the authors (Figure 2). In all studies with low risk of bias in allocation, the authors explicitly stated that some form of random number scheme was used, mostly by computers, to generate a random sequence. Explicit statements were also made on the independence between random sequence generation and allocation. All three studies with high risk of bias in allocation used some form of alternation based on participants' identifying number or catheter type.
The majority of the studies had unknown or high risks of bias for blinding. In 12 studies, the participants were described as blinded, and in 24 studies the participants were described as non-blinded. In the remaining studies, there was not enough information on blinding, although blinding appeared unlikely in most of them due to the different appearances between the catheters evaluated. For outcome assessment, 15 studies described the microbiological outcome assessors as blinded. In the remaining studies, the blinding status of the microbiological outcome assessors was unknown. For clinical outcome assessment, 15 studies described the assessors as blinded, 24 described them as non-blinded, and blinding status of the clinical outcome assessors in the remaining studies was unknown.
Incomplete outcome data
Judging from the completeness of the data across all the major outcomes including clinically diagnosed sepsis, CRBSI, mortality and catheter colonization, we considered 24 studies to have low risks of bias and 14 studies to have high risks of bias. We assessed a study as having a high risk of bias for either of the following two reasons, alone or in combination:
- High attrition rates, either in absolute terms (≥ 20% attrition) and/or in relation to the event rates in the control group;
- Marked imbalance in the attrition rates between the assigned groups.
Additionally, in five studies with high risks of bias, the reasons stated for withdrawals appeared dubious, e.g. catheter removal prior to day three or four (two studies), catheter change (two studies), transfer to another unit or death (three studies each), as these did not preclude the participant and/or catheters from being assessed for at least some of the outcomes, and they might indeed represent important and relevant outcomes, e.g. excluding those who died might be inappropriate as the deaths might be related to the interventions assessed.
Over half of the included studies (30) had low risks, and eight studies had high risks of reporting bias. In four of the eight studies, some outcomes were not reported in a format that would allow data extraction for meta-analysis. For example, the authors presented the results in graphs without data labels, or reported continuous outcomes as means without standard deviation. In another four studies,the authors failed to include any important clinical outcome such as CRBSI, sepsis or mortality in their study. We constructed a matrix that contains a more detailed description of these studies, along with other included studies in which there were discrepancies between the prespecified outcomes in the methods and reported outcomes in the results (Appendix 7).
Other potential sources of bias
Apart from a significant baseline imbalance in the major participant characteristics observed in one study (Smith 1995) (for details see Characteristics of included studies), we observed no additional major sources of bias in the included studies.
Effects of interventions
This review evaluates a total of 16,512 catheters in 56 studies. The total number of participants was unclear as some studies only specified the number of catheters and not the participants.
1. Comparison 1: Antimicrobial impregnation versus no impregnation
1.Clinically diagnosed sepsis: there is no difference between the impregnated group and the non-impregnated group (risk ratio (RR) 1.0, 95% confidence interval (CI) 0.88 to 1.13, 12 studies, 3686 catheters, I² = 19%) ( Analysis 1.1; Figure 4)). The funnel plot for this outcome (not shown) is asymmetrical, suggesting a possibility of publication bias as smaller studies with outcomes favouring non-impregnated catheters appear to be lacking.
|Figure 4. Forest plot of comparison: 1 Impregnated catheters versus non-impregnated catheters, outcome: 1.1 Clinically-diagnosed sepsis.|
2. a) Catheter-related bloodstream infection (CRBSI): there is a significant reduction in CRBSI in the impregnated group (absolute risk reduction (ARR) 2%, 95% CI 3% to 1%, number needed to treat for benefit (NNTB) 50, RR 0.61, 95% CI 0.51 to 0.73, 41 studies, 10133 catheters, I² = 22%) ( Analysis 1.2; Figure 5)). There is no evidence from the funnel plot of publication bias;
b) CRBSI per 1000 catheter days: there is no difference between the impregnated group and the non-impregnated group (RR 0.73, 95% CI 0.49 to 1.10, 14 studies, I² = 24%) ( Analysis 1.3)).
|Figure 5. Forest plot of comparison: 1 Impregnated catheters versus non-impregnated catheters, outcome: 1.2 Catheter related bloodstream infection (CRBSI).|
3. All-cause mortality: there is no difference between the impregnated group and the non-impregnated group (RR 0.88, 95% CI 0.75 to 1.05, nine studies, 2371 catheters, I² = 0%) ( Analysis 1.4)). There is no evidence from the funnel plot of publication bias .
1. Catheter-related local infection: there is no difference between the impregnated group and the non-impregnated group (RR 0.84, 95% CI 0.66 to 1.07,12 studies, 2688 catheters, I² = 1%) ( Analysis 1.5).
2. a) Catheter colonization: there is a significant reduction in the impregnated group (ARR 10%, 95% CI 13% to 7%, NNTB 10, RR 0.66, 95% CI 0.58 to 0.75, 42 studies, 9638 catheters, I² = 64%) ( Analysis 1.6; Figure 6);
b) Catheter colonization per 1000 catheter days: there is a significant reduction in the impregnated group (RR : 0.70, 95% CI: 0.51 to 0.96, 11 studies, I² = 50%) ( Analysis 1.7). There is no gross evidence from the funnel plot of publication bias. However, there is substantial heterogeneity among the included studies, as indicated by I² statistics of 64% and 50% respectively for both outcomes above. The degree of heterogeneity is not explained by the presence of many subgroups comprising different types of impregnation, as the I² statistic remains high within some of the subgroups, e.g. chlorhexidine-silver sulphadiazine (C-SS) impregnation versus no impregnation (20 studies, I² = 58%), minocycline-rifampicin (MR) impregnation versus no impregnation (four studies, I² = 71%) and silver impregnated cuff versus no impregnation (two studies, I² = 85%).
|Figure 6. Forest plot of comparison: 1 Impregnated catheters versus non-impregnated catheters, outcome: 1.6 Catheter colonization.|
Further exploration of the possible reasons for the heterogeneity within each of the subgroups above revealed that the differences in the settings of the study had probably contributed to the heterogeneity. There is a detailed description of the discordance in the pooled estimates between studies conducted in different settings under the heading of 'Subgroup analyses: 4 iii: Participant type' below. After our exploration of the causes of heterogeneity, we decided to use the random-effects model for the outcomes of catheter colonization and catheter colonization per 1000 catheter days.
3. Skin or insertion site colonization: there is a significant reduction in the impregnated group (RR 0.78, 95% CI 0.62 to 0.97, three studies, 366 catheters, I² = 55%) ( Analysis 1.8).
4. Mortality attributed to CRBSI: there is no difference between the impregnated group and the non-impregnated group (RR 0.24, 95% CI 0.03 to 2.20, five studies, 1098 catheters, I² = 0%) ( Analysis 1.9).
5. Adverse effects: there are no differences between the impregnated group and the non-impregnated group in any of the following outcomes:
a) Thrombosis/thrombophlebitis (RR 0.90, 95% CI 0.44 to 1.85, three studies, 829 catheters, I² = 0%);
b). Bleeding (RR 0.86, 95% CI 0.30 to 2.48, one study, 240 catheters);
c). Combined adverse effects of bleeding, pain, erythema and/or tenderness at the insertion site (RR 1.09, 95% CI 0.94 to 1.27, 10 studies, 3003 catheters, I² = 0%) ( Analysis 1.10). There is no evidence from the funnel plot of publication bias.
6. Number of catheters removed prematurely: there is no difference between the impregnated group and the non-impregnated group (RR 1.00, 95% CI 0.92 to 1.09, 15 studies, 3666 catheters, I² = 28%) ( Analysis 1.11).
7. Number of participants who were on systemic antibiotics: there is no difference between the impregnated group and the non-impregnated group (RR 0.95, 95% CI 0.87 to 1.04, two studies, 541 participants) ( Analysis 1.12).
8. Length of stay in ICU (days): there is no difference between the impregnated group and the non-impregnated group (mean difference (MD): -1.0 (95% CI : -4.81 to 2.81, one study, 275 participants) (Additional Table 1; outcome 1.13).
9. Cost: in the two studies (Maki 1988; Maki 1997) that reported this outcome, the authors derived the costs based on their data on CRBSI. Maki 1988 estimated that the use of a silver-impregnated cuff could save at least 8600 US Dollars (USD) for every 100 cuffs used, assuming that silver-impregnated cuff reduced CRBSI by at least three-fold. Maki 1997 estimated that the use of C-SS-impregnated catheters could save at least USD 55,000 in direct hospital costs for every 100 impregnated catheters used, assuming that C-SS-impregnated catheters reduced CRBSI by at least three-fold.
2. Comparison 2: One antimicrobial impregnation versus another
i) Minocycline-rifampicin (MR) versus chlorhexidine silver-sulphadiazine (C-SS) impregnation
MR impregnation was shown to significantly reduce CRBSI and catheter colonization compared to C-SS impregnation, but there was no difference between the two groups in mortality attributed to CRBSI and the rate of premature catheter removal.
1. a) CRBSI (RR 0.11, 95% CI 0.02 to 0.58, NNTB 33, two studies, 812 catheters, I² = 0%) ( Analysis 2.1);
b) CRBSI per 1000 catheter days (RR 0.07, 95% CI 0.01 to 0.43, one study) (Additional Table 1; outcome 2.3).
2. Catheter colonization (RR 0.36, 95% CI 0.25 to 0.53, NNTB 7, two studies, 812 catheters, I² = 0%) ( Analysis 2.2).
3. Mortality attributed to CRBSI (RR 0.21, 95% CI 0.01 to 4.39, one study, 720 catheters) (Additional Table 1; outcome 2.4).
4. Premature catheter removal/catheter failure (RR 1.06, 95% CI 0.86 to 1.31, one study, 738 catheters) (Additional Table 1; outcome 2.6).
ii) Silver impregnation versus chlorhexidine silver-sulphadiazine (C-SS) impregnation
There was only one study (Dunser 2005) with 155 participants and 325 catheters analysed under this comparison. The results for all outcomes assessed in this study are tabulated in Additional Table 1. The results favoured C-SS impregnation in two outcomes, namely catheter colonization (RR 2.32, 95% CI 1.22 to 4.42, NNTB 10) (outcome 3.3), and catheter colonization per 1000 catheter days (RR 2.44, 95% CI 1.05 to 5.66) (outcome 3.4). There were no differences between silver impregnation and C-SS impregnation in clinically diagnosed sepsis (RR 0.84, 95% CI 0.62 to 1.15) (outcome 3.1), all-cause mortality (RR 0.58, 95% CI 0.30 to 1.13) (outcome 3.2) and length of ICU stay (MD in days 0.00, 95% CI -5.06 to 5.06) (outcome 3.6).
iii) Heparin versus chlorhexidine silver-sulphadiazine (C-SS) impregnation
There was only one study (Carrasco 2004) with 260 catheters analysed under this comparison. The results for all outcomes assessed in this study are tabulated in Additional Table 1. The results favoured C-SS impregnation in the outcomes of catheter colonization (RR 2.16, 95% CI 1.18 to 3.97, NNTB 8) (outcome 4.3) and catheter colonization per 1000 catheter days (RR 2.08, 95% CI 1.02 to 4.20) (outcome 4.4). There were no differences between the groups in the other outcomes assessed, namely, CRBSI (RR 1.29, 95% CI 0.30 to 5.66) (outcome 4.1) and CRBSI per 1000 catheter days (RR 1.25, 95% CI 0.24 to 6.34) (outcome 4.2).
iv) Minocycline-rifampicin (MR) versus silver-platinum carbon (SPC) impregnation
There was only one study (Fraenkel 2006) with 646 participants and 574 evaluable catheters analysed under this comparison. The results for all outcomes assessed in this study are tabulated in Additional Table 1. In the only statistically significant result, MR impregnation was shown to reduce catheter colonization compared to SPC impregnation (RR 0.61, 95% CI 0.38 to 0.97, NNTB 17) (outcome 5.4). There were no differences between the two groups in five other outcomes assessed, namely, CRBSI (RR 0.84, 95% CI 0.23 to 3.10) (outcome 5.1), CRBSI per 1000 catheter days (RR 0.85, 95% CI 0.15 to 4.97) (outcome 5.2), all-cause mortality (RR 1.03, 95% CI 0.70 to 1.50) (outcome 5.3), catheter colonization per 1000 catheter days (RR 0.98, 95% CI 0.47 to 2.02) (outcome 5.5) and combined adverse effects (RR 1.49, 95% CI 0.86 to 2.57) (outcome 5.6).
v) Benzalkonium versus silver-platinum carbon (SPC) impregnation
There was only one study (Ranucci 2003) with 545 catheters analysed under this comparison. The results for all outcomes assessed in this study are tabulated in Additional Table 1. Of the two outcomes assessed, one favoured benzalkonium impregnation (catheter colonization (RR 0.63, 95% CI 0.46 to 0.86, NNTB 9) (outcome 6.2)), and the other showed no difference between the two groups (CRBSI (RR 0.78, 95% CI 0.33 to 1.81) (outcome 6.1)).
vi) 5-fluorouracil (5FU) versus chlorhexidine silver-sulphadiazine (C-SS) impregnation
There was only one study (Walz 2010) with 960 participants and 817 evaluable catheters under this comparison. The results for all outcomes assessed in this study are tabulated in Additional Table 1. Of the seven outcomes assessed, including clinically diagnosed sepsis (outcome 7.1), CRBSI (outcome 7.2), all-cause mortality (outcome 7.3), catheter colonization (outcome 7.4), catheter-related local infection (outcome 7.5), any adverse effects (outcome 7.6) and duration of antibiotics used (days) (outcome 7.7), none showed any significant difference between the two groups. However, the effects estimates of certain outcomes were too imprecise to derive any clear conclusion on the relative effectiveness of the two intervention assessed, as the 95% CIs were wide in the outcomes of CRBSI (outcome 7.2), catheter colonization (outcome 7.4), catheter-related local infection (outcome 7.5) and combined adverse effects (outcome 7.6).
3. Comparison 3: Antimicrobial impregnation versus other antimicrobial modifications
There was no eligible study that compared catheters with antimicrobial impregnation against catheters with other antimicrobial modifications, e.g. antiseptic dressings, hubs, tunnelling, needleless connectors or antiseptic lock solutions.
4. Subgroup analyses
We performed the following subgroup analyses, as specified in our Methods, to test for substantial difference in the results based on the type of impregnation, duration of catheter use, predominant participant type, baseline risk, study definition of clinically diagnosed sepsis, catheter impregnation with and without co-intervention, and cost effectiveness in different units of measurement. Specific data analyses and forest plots for these subgroup are not displayed separately.
i). Each specific type of impregnation versus no impregnation
From the overall comparison between impregnation and no impregnation as detailed under the heading of 'Comparison 1: Antimicrobial impregnation versus no impregnation' above, we divided the studies into subgroups comprising specific types of impregnation. Out of 11 types assessed, there were significant reductions in the rates of CRBSI in four types of impregnation, and significant reductions in catheter colonization rates in six types of impregnation. Following is a list of the catheter impregnation type according to the magnitude of reduction in CRBSI and catheter colonization respectively.
a) CRBSI: the greatest reduction was shown in studies assessing MR impregnation (RR 0.26, 95% CI 0.13 to 0.49, NNTB 20, four studies, 1335 catheters, I² = 9%), followed by heparin coating (RR 0.27, 95% CI 0.08 to 0.95, NNTB 14, one study, 240 catheters), silver impregnation (RR 0.53, 95% CI 0.35 to 0.79, NNTB 25, five studies, 1315 catheters, I² = 0%), and C-SS impregnation (RR 0.75, 95% CI 0.58 to 0.97, NNTB 100, 18 studies, 4653 catheters, I² = 21%), although there was no significant difference across the subgroups in the magnitudes of the effect as indicated by a P value of 0.06 in the test of subgroup differences (Chi² = 16.32, df = 9, I² = 44.8%) ( Analysis 1.2; Figure 5).
b) Catheter colonization: the greatest reduction was shown by studies assessing miconazole-rifampicin impregnation (RR 0.14, 95% CI 0.06 to 0.32, NNTB 3, one study, 223 catheters), followed by MR impregnation (RR 0.52, 95% CI 0.29 to 0.94, NNTB 8, four studies, 985 catheters, I² = 71%), benzalkonium impregnation (RR 0.56, 95% CI 0.39 to 0.83, NNTB 6, two studies, 254 catheters, I² = 0%), C-SS impregnation (RR 0.59, 95% CI 0.49 to 0.72, NNTB 9, 20 studies, 4449 catheters, I² = 58%), and vancomycin coating (RR 0.77, 95% CI 0.63 to 0.93, NNTB 6, one study, 176 catheters). The difference in the magnitudes of the effect was highly significant across the subgroup (P < 0.0001 in the test for subgroup differences (Chi² = 40.89, df = 10, I² = 75.5%)) ( Analysis 1.6; Figure 6). For the subgroup of silver-impregnated cuff versus no impregnation, the results differed markedly between the two studies conducted in two different settings, as detailed under the previous heading of 'Comparison 1: Antimicrobial impregnation versus no impregnation, Secondary outcome 2. Catheter colonization'. We therefore considered it inappropriate to refer to the pooled estimate for this subgroup.
c) All other outcomes: the pooled estimates from each subgroup of a specific impregnation type revealed no significant difference between the impregnated and non-impregnated groups, with the exception of two specific comparisons under two different outcomes. Firstly, for the outcome of CRBSI per 1000 catheter days, there was a significant reduction favouring silver impregnation over no impregnation (RR 0.36, 95% CI 0.15 to 0.85, two studies, 782 catheters, I² = 6%) ( Analysis 1.3.3); secondly, for the outcome of catheter-related local infection, there was a significant reduction favouring miconazole-rifampicin impregnation versus no impregnation (RR 0.25, 95% CI 0.10 to 0.64, one study, 223 catheters) ( Analysis 1.5.3). Notably, for the major outcomes of clinically diagnosed sepsis and mortality (all-cause and attributable to CRBSI), there were no differences between impregnated and non-impregnated groups in any types of impregnation assessed.
ii) Short-term and long-term catheter use
We were unable to perform subgroup analysis based on catheters used for the short term (< 10 days) or the long term (10 days or more) because there were no separate data for these two subgroups in the included studies. However, Darouiche 1999, comparing MR impregnation versus C-SS impregnation, provided separate data for catheters used for seven days or less against catheters used for more than seven days. The findings on catheter colonization were comparable between the two subgroups, in which MR impregnation was shown to reduce catheter colonization compared to C-SS impregnation (seven days or less: RR 0.28, 95% CI 0.16 to 0.50; more than seven days: RR 0.44, 95% CI 0.26 to 0.76). However, for the outcome of CRBSI, the result was markedly in favour of MR impregnation for catheters used for more than seven days (RR 0.11, 95% CI 0.01 to 0.86), as opposed to the subgroup of catheters used for seven days or less, in which there was no significant difference between the two impregnation (RR 0.48, 95% CI 0.04 to 5.30).
iii) Participant type
We evaluated two major outcomes of CRBSI and catheter colonization under the overall comparison of any antimicrobial impregnation versus no impregnation where there were a sufficient number of included studies. We grouped together studies that enrolled predominantly participants receiving intensive care, studies with a mixture of participants including ICU and non-ICU, studies that enrolled predominantly people in haematological or oncological units, and studies that enrolled predominantly participants requiring CVCs for long-term total parenteral nutrition (TPN).
The overall result showed that antimicrobial impregnation significantly reduced CRBSI compared to no impregnation (RR 0.61, 95% CI 0.51 to 0.73, 41 studies). Substantial reductions in CRBSI were shown in studies conducted in haematological and oncological units (RR 0.50, 95% CI 0.36 to 0.71, eight studies), in studies with a mixture of ICU and non-ICU patients (RR 0.47, 95% CI 0.27 to 0.81, seven studies) and in studies conducted only in ICUs (RR 0.69, 95% CI 0.54 to 0.89, 23 studies). However, there was no difference in the rate of CRBSI between impregnated and non-impregnated groups in studies conducted in participants in whom CVCs were inserted for TPN (RR 0.83, 95% CI 0.45 to 1.53, three studies). The overall difference in the magnitude of the results among the four subgroups was not significant, as indicated by a P value of 0.25 in the test for subgroup differences (Chi² = 4.15, df = 3, I² = 27.7%) ( Analysis 3.1).
b) Catheter colonization
The overall result showed that antimicrobial impregnation significantly reduced catheter colonization compared to no impregnation (RR 0.66, 95% CI 0.58 to 0.75, 42 studies). Studies conducted only in ICUs as well as studies with a mixture of ICU and non-ICU participants demonstrated substantial reductions in catheter colonization favouring antimicrobial impregnation (ICU: RR 0.68, 95% CI 0.59 to 0.78, 28 studies; mixed participants: RR 0.40, 95% CI 0.22 to 0.74, six studies). However, there were no differences in the rates of catheter colonization between impregnated and non-impregnated groups in studies conducted in haematological and oncological units, nor in studies conducted in participants in whom CVCs were inserted for TPN (haematological and oncological units: RR 0.75, 95% CI 0.51 to 1.11, six studies; TPN: RR 0.99, 95% CI 0.74 to 1.34, two studies).The overall difference in the magnitude of the results among the four subgroups was significant, as indicated by a P value of 0.03 in the test for subgroup differences (Chi² = 8.68, df = 3 (P = 0.03), I² = 65.4%) ( Analysis 3.2).
iv) Baseline risk
We screened for any major effect of baseline risks on the results by comparing the higher against the lower risk groups, using the median event rate of the control group across the included studies as the cut-off. We assigned studies with control event rates equal to or higher than the median event rate as higher risk, and the remaining studies as lower risk. We evaluated the two most frequently-assessed outcomes, namely, CRBSI and catheter colonization under the comparison of any antimicrobial impregnation versus no impregnation, where there were a sufficient number of included studies.
There was no difference in the magnitude of reduction in CRBSI between the higher and lower risk groups (Higher risk group: RR 0.59, 95% CI 0.47 to 0.74, 21 studies; Lower risk group: RR 0.66, 95% CI 0.48 to 0.90, 20 studies). The test for subgroup differences was non-significant: P = 0.57 (Chi² = 0.32, df = 1, I² = 0%) ( Analysis 4.1).
b) Catheter colonization
There was no difference in the magnitude of reduction in catheter colonization between the higher and lower risk groups (Higher risk group: RR 0.64, 95% CI 0.54 to 0.77, 22 studies; Lower risk group: RR 0.67, 95% CI 0.54 to 0.84, 20 studies). The test for subgroup differences was non-significant: P = 0.76 (Chi² = 0.10, df = 1, I² = 0%) ( Analysis 4.2).
v) Outcome of 'clinically diagnosed sepsis': definitions within and outside the scope of our definition
Eight out of 13 studies that evaluated clinically diagnosed sepsis defined this outcome in line with our prespecified definition in this review (see Appendix 1). Seven of the eight studies compared some form of antimicrobial impregnation against no impregnation, while the remaining study compared 5-fluorouracil (5-FU) impregnation against C-SS impregnation. The result of the seven studies that compared any impregnation against no impregnation was almost identical to the overall result, showing no difference between the impregnated and non-impregnated groups for this outcome (RR 0.97, 95% CI 0.84 to 1.13) (analysis not displayed).
Only one study provided a definition for this outcome that the review authors considered to be outside the scope of our definition (Maki 1988). This study compared catheters with silver impregnated cuff against non-impregnated catheters, and showed no difference in the rate of sepsis between the two groups.
In the remaining four studies, the definition of this outcome was unclear.
vi) Impregnation with and without co-intervention
While most included studies had some form of hygiene protocol to follow, no study evaluated catheter impregnation with another specific intervention.
vii) Cost effectiveness in different units of measurement
The two studies that included costs as an outcome reported their cost estimates in US dollars. Their findings are summarised under 'Comparison 1: Secondary outcomes: point 9' above.
5. Sensitivity analysis
We performed sensitivity analyses on four major outcomes, namely, clinically diagnosed sepsis, CRBSI, all-cause mortality and catheter colonization, to evaluate the impact of excluding some studies based on the risks of selection and attrition bias. We only evaluated the comparison of catheters with any antimicrobial impregnation versus non-impregnated catheters, as there were insufficient studies in the other comparisons to permit a meaningful analysis.
i) Clinically diagnosed sepsis
a) Selection bias: no study was excluded due to high risks of selection bias.
b) Attrition bias: excluding four studies with a high risk of attrition bias did not substantively alter the pooled estimates (RR 1.00, 95% CI 0.88 to 1.13, compared with RR 1.02, 95% CI 0.90 to 1.16).
a) Selection bias: the pooled estimates were identical whether or not the two studies with high risks of selection bias were excluded (RR 0.61, 95% CI 0.51 to 0.73).
b) Attrition bias: excluding 11 studies with a high risk of attrition bias did not substantively alter the final estimates (RR 0.61, 95% CI 0.51 to 0.73, compared with RR 0.61, 95% CI 0.50 to 0.75). However, in the subgroup of silver impregnation versus no impregnation, excluding two studies with high risks of attrition bias out of five included studies resulted in a substantial change favouring impregnation, from RR 0.53, 95% CI 0.35 to 0.79, to RR 0.61, 95% CI 0.37 to 1.01.
iii) All-cause mortality
a) Selection bias: excluding one study with a high risk of selection bias from nine studies in total did not substantively change the pooled estimates (RR 0.88, 95% CI 0.75 to 1.05, compared with RR 0.90, 95% CI 0.74 to 1.08).
b) Attrition bias: excluding two studies with a high risk of attrition bias from nine studies in total did not substantively change the pooled estimates (RR 0.88, 95% CI 0.75 to 1.05, compared with RR 0.88, 95% CI 0.74 to 1.06).
iv) Catheter colonization
a) Selection bias: excluding two studies with a high risk of selection bias from a total of 42 studies did not substantively change the pooled estimates (RR 0.66, 95% CI 0.58 to 0.75, compared with RR 0.66, 95% CI 0.57 to 0.75).
b) Attrition bias: excluding 11 studies with a high risk of attrition bias from a total of 42 studies did not substantively change the pooled estimates (RR 0.66, 95% CI 0.58 to 0.75, compared with RR 0.72, 95% CI 0.62 to 0.82).
Summary of main results
Antimicrobial impregnations for central venous catheters (CVCs) did not reduce clinically diagnosed sepsis and all-cause mortality despite reductions in catheter-related bloodstream infections (CRBSIs) and catheter colonization. There are no differences in the rates of adverse effects observed between participants with impregnated and non-impregnated catheters. The discrepancy in the results between the outcomes of CRBSI (significant reduction favouring impregnated catheters) and CRBSI per 1000 catheter days (no significant difference) was due to the difference in the number of studies included in the analysis for each outcome. While there were a large number of studies (41 studies, 10,133 catheters) that evaluated CRBSI, only 14 of these studies (4400 catheters) evaluated CRBSI per 1000 catheter days as well. Therefore, the findings on CRBSI in this review should be more representative than those on CRBSI per 1000 catheter days. Our findings on the seven major outcomes are displayed in the Summary of findings for the main comparison. Most studies that assessed catheter impregnation included catheter-specific outcomes, and fewer than a quarter assessed non-catheter-specific but critical outcomes of clinical sepsis and mortality. Limited evidence offered by studies that compared different types of impregnation head-to-head suggests that in terms of microbiological outcomes such as catheter colonization, minocycline-rifampicin (MR) impregnation appeared to be superior to chlorhexidine-silver sulphadiazine (C-SS) impregnation, which was in turn superior to silver impregnation and heparin coating. For the outcomes of catheter colonization, the magnitude of effects in the studies differed depending on the types of participants assessed, with significant benefits observed only in studies conducted for those in intensive care units (ICUs), and no significant benefits of antimicrobial impregnation in studies conducted in haematological or oncological unit or in studies for participants for whom CVCs were inserted for long-term total parenteral nutrition (TPN). There were wide ranges of baseline risks in major outcomes in the included studies, which were possibly explained by differences in factors such as study setting or participant type, study location and period. However, the variation in the degree of reduction in catheter colonization was not explained by the differences in baseline risks, as the degree of reduction in CRBSI and catheter colonization conferred by antimicrobial impregnation was similar between 'higher risk' and 'lower risk' participants. It was unclear whether such differences in the effects could be accounted for by other factors such as the underlying conditions of the participants and the existing infection control measures. In general, excluding studies with high risks of selection and/or attrition biases in our sensitivity analysis for the outcomes in which there are large number of included studies made no material difference to the results.
Overall completeness and applicability of evidence
We identified a large number of studies that contained the population, intervention, comparison and outcomes that matched our prespecified selection criteria. The studies were conducted in a range of settings in 17 different countries. The body of evidence that we have gathered is reflective of the current interest in the use of antimicrobial-impregnated central venous catheters as measures in reducing hospital-associated infections. Statements supporting the use of antimicrobial impregnated central venous catheters are found in practice guidelines from authoritative sources such as the Centers for Disease Control (CDC 2011).
Quality of the evidence
We included 56 studies with 16,512 catheters in our review. There was a sufficient number of studies to enable a meaningful meta-analysis for most of our prespecified outcomes, including the primary outcomes of clinically diagnosed sepsis, CRBSI and all-cause mortality, although fewer than a quarter of the included studies assessed the major outcomes of clinically diagnosed sepsis and all-cause mortality. The majority of the included studies had low or unclear risks of bias in most criteria, except for blinding, which was not possible for the participants and health care providers in most studies due to different appearances of the catheters. Further, there were issues with suspected publication bias for the outcomes of clinically diagnosed sepsis and heterogeneity for the outcome of catheter colonization. However, overall, the quality of evidence on the major outcomes was moderate to high to enable robust conclusions to be drawn (see Summary of findings for the main comparison).
Potential biases in the review process
The strengths of this review include a comprehensive search from multiple sources and extensive analyses incorporating non-catheter-specific but clinically important outcomes such as sepsis and mortality, as well as comprehensive subgroup analyses. A limitation of this review is that for catheter-specific outcomes such as CRBSI and catheter colonization, we reported the results in terms of catheters rather than participants, as all the included studies reported their results in the unit of catheters and none provided the number of participants for each outcome. Our failure to address the possible unit of analysis bias by adjusting the data for participants who had multiple catheters could have affected the results. Also, in the subgroup analysis on baseline risk, our decision to explore any possible effects of different baseline risks by simply comparing the 'higher risk' and 'lower risk' group instead of performing a meta-regression could have resulted in a loss of information of interest.
Agreements and disagreements with other studies or reviews
There have been several systematic reviews published since 1999 that assessed the effectiveness of CVC impregnations. Many reviews assessed chiefly C-SS and/or MR impregnation and found that the impregnated CVCs significantly reduced CRBSI and/or catheter colonization (Casey 2008; Falagas 2007; Hockenhull 2008; Hockenhull 2009; Niel-Weise 2007; Ramritu 2008; Veenstra 1999a) and were estimated to be cost-effective (Veenstra 1999b). However, other reviews found antimicrobial-impregnated CVCs to have no significant benefits (Gilbert 2008; McConnell 2003; Niel-Weise 2008). Notably, the authors in Niel-Weise 2007 found substantial benefits of antimicrobial-impregnated CVCs in a meta-analysis of 21 trials conducted either in ICUs or other acute care settings, but found no benefits in a separate meta-analysis of nine trials assessing CVCs for TPN and chemotherapy (Niel-Weise 2008). The authors postulated that the difference in the duration of catheter placement (mean of 6 to 12 days in the included studies in Niel-Weise 2007 and 11 to 20 days in the included studies in Niel-Weise 2008), the small sample sizes and the methodological limitations of the included studies in Niel-Weise 2008 were possible factors that could have influenced the findings. In Niel-Weise 2007, the authors included a study that assessed haemodialysis catheters and a study on children.
The systematic reviews cited above vary in scope, and most evaluated only catheter-specific outcomes such as CRBSI and catheter colonization. There is no systematic review that incorporates non-catheter-specific critical outcomes such as clinical sepsis and mortality for a direct comparison with our findings.
Implications for practice
While there is convincing evidence on the benefits of antimicrobial-impregnated CVCs in reducing CRBSI and catheter colonization, limited evidence to date shows no significant benefit of these catheters in reducing clinically diagnosed sepsis and mortality. Therefore, there remains uncertainty about the value of these modified catheters in improving overall patient mortality and morbidity. Furthermore, this review shows that there were significant benefits for the impregnated CVCs in catheter-related outcomes such as catheter colonization only in trials conducted in ICUs. Currently, while the overall body of evidence still allows recommendations in favour of their use in practice, there should be great caution in recommending the use of antimicrobial-impregnated CVCs across all settings without incorporating the current uncertainties on their overall benefits.
Implications for research
Despite strong evidence on the overall benefits of antimicrobial-impregnated CVCs in reducing CRBSI and catheter colonization, there is a need for ongoing research to evaluate their effects on these outcomes due to evolving patterns of hospital-associated infections, infection control measures and microbiological diagnostic techniques. However, following the findings of this review, the question of whether antimicrobial CVCs reduce overall rates of sepsis and mortality is now the critical research question. Future research should include these two key outcomes alongside catheter-specific outcomes such as CRBSI and catheter colonization. Any cost estimation should also include the figures based on the overall rates of sepsis due to the limitations of currently available microbiological diagnostic techniques. The setting, type of participants and the existing infection control measures should be clearly described in future studies to enable an evaluation of the possible reasons for any difference in the effects of these impregnated catheters among different settings.
As part of the prepublication editorial process, the protocol for the systematic review was commented on by a content editor and three peer reviewers (who are external to the editorial team), members of the Cochrane Consumer Network’s international panel of consumers and the Cochrane Anaesthesia Group’s Trials Search Co-ordinator.
We would like to thank Harald Herkner (content editor), Cathal Walsh (statistical editor), Djillali Annane, Nasia Safdar, Donna Gillies (peer reviewers) and Anne Lyddiat (consumer) for their help and editorial advice during the preparation of this systematic review.
We thank the staff at the International-Ann Arbor Safety Collaborative (I-A²SC) for providing invaluable help with checking risk of bias assessment and data entry. Specifically, Andy Hickner and Todd Greene provided assistance in cross-checking the risk of bias assessments; and Andy Hickner assisted in cross-checking data entry and proof-reading the draft review.
We gratefully acknowledge the assistance provided by Ms Jane Cracknell, Cochrane Anaesthesia Review Group Managing Editor and Dr Karen Hovhanisyen, Review Group Trials Search Co-ordinator, in our work.
Data and analyses
- Top of page
- Summary of findings [Explanations]
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Appendix 1. Diagnostic criteria for sepsis (2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference)
- Fever (core temperature > 38.3°C)
- Hypothermia (core temperature < 36°C)
- Heart rate > 90 bpm or > 2 SD above the normal value for age
- Tachypnea: > 30 bpm
- Altered mental status
- Significant oedema or positive fluid balance (> 20 ml/kg over 24 hrs)
- Hyperglycaemia (plasma glucose > 110 mg/dl or 7.7 mM/l) in the absence of diabetes
- Leukocytosis (white blood cell count > 12,000/µl)
- Leukopenia (white blood cell count < 4,000/µl)
- Normal white blood cell count with > 10% immature forms
- Plasma C reactive protein > 2 SD above the normal value
- Plasma procalcitonin > 2 SD above the normal value
- Arterial hypotension
b(systolic blood pressure < 90 mmHg, mean arterial pressure < 70, cValues of 3.5 – 5.5 are normal or a systolic blood pressure decrease > 40 mmHg in adults or < 2 SD below normal for age)
- Mixed venous oxygen saturation > 70%
- Cardiac index > 3.5 l min
Organ dysfunction parameters
- Arterial hypoxaemia (PaO
2/FIO2 < 300)
- Acute oliguria (urine output < 0.5 ml kg
-1h -1or 45 mM/l for at least 2 hrs)
- Creatinine increase ≥ 0.5 mg/dl
- Coagulation abnormalities (international normalized ratio > 1.5 or activated partial thromboplastin time > 60 s)
- Ileus (absent bowel sounds)
- Thrombocytopenia (platelet count < 100,000/µl)
- Hyperbilirubinemia (plasma total bilirubin > 4 mg/dl or 70 mmol/l)
Tissue perfusion parameters
- Hyperlactatemia ( > 3 mmol/l)
- Decreased capillary refill or mottling
Appendix 2. MEDLINE (OVID SP) search strategy
#1: exp Catheterization, Central Venous/
#2: (venous or vein) or catheter*.mp.
#3: #1 or #2
#4: (impregn* or bond* or coat*).mp.
#5: (anti?microbial or antisep* or antibiotic*).mp.
#6: (needleless and connector*).mp.
#7: exp Chlorhexidine/
#8: exp Silver-Sulfadiazine/
#9: exp Minocycline/
#10: exp rifampin/
#11: (Rifampi* or Minocyclin* or Silver?Sulfadiazine or Chlorhexidine or "Arrowgard" or "Cook Spectrum").mp.
#12: #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11
#13: #3 and #12
#14: Clinical trial.pt.
#15: exp clinical trial/
#18: #14 or #15 or #16 or #17
#19: #13 and #18
Appendix 3. CENTRAL search strategy
#1: MeSH descriptor Catheterization, Central Venous explode all trees
#2: (catheter near impregnat* ):ti,ab,kw
#3: (catheter* near coat*):ti,ab,kw
#4: (catheter* near bond*):ti,ab,kw
#5: #1 OR #2 OR #3 OR #4 (Restrict by product: Clinical Trials)
#6: (antimicrobial OR antisep* OR antibiotics):ti,ab,kw
#7: (needleless AND connector*):ti,ab,kw
#9: (silver near sulphadiazine):ti,ab,kw
#12: "Arrowgard" OR "Cook Spectrum":ti,ab,kw
#13: #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 (Restrict by product: Clinical Trials)
#14: #5 AND #13
Appendix 4. EMBASE search strategy
#1: Emtree: Explode: "Central Venous catheterization"/all subheadings
#2: ((venous OR vein) AND catheter):ab,ti
#3: #1 OR #2
#4: (impregn* OR bond* OR coat*):ab,ti
#5: (anti?microbial OR antiseptic OR antibiotic*):ab,ti
#6: (needleless AND connector*):ab,ti
#7: Emtree: Explode: "Chlorhexidine"/ all subheadings
#8: Emtree: Explode: "Sulfadiazine silver"/all subheadings
#9: Emtree: Explode: "Minocycline"/all subheadings
#10: Emtree: Explode: "Rifampicin"/all subheadings
#11: (Rifampi* OR Minocyclin* OR Silver?Sulfadiazine OR Chlorhexidine OR "Arrowgard" OR "Cook Spectrum"):ab,ti
#12: #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11
#13: #3 AND #12
#14: Emtree: Explode: "RANDOMIZED-CONTROLLED TRIAL"/ all subheadings
#15: Emtree: Explode: "RANDOMIZATION"/ all subheadings
#16: Emtree: Explode: "CONTROLLED STUDY"/ all subheadings
#17: Emtree: Explode:"MULTICENTER STUDY"/ all subheadings
#18: Emtree: Explode:"DOUBLE BLIND PROCEDURE"/ all subheadings
#19: Emtree: Explode:"SINGLE BLIND PROCEDURE"/ all subheadings
#20: #14 OR #15 OR #16 OR #17 OR #18 OR #19
#21: (RANDOM* OR CROSS?OVER* OR FACTORIAL* OR PLACEBO* OR VOLUNTEER*):ab,ti
#22: (SINGL* OR DOUBL* OR TREBL* OR TRIPL*) AND (BLIND* OR MASK*):ab,ti
#23: #20 OR #21 OR #22
#24: #13 AND #23
Appendix 5. CINAHL search strategy (via EBSCOHost)
#1 MH "Catheterization, Central Venous"/explode
#2 TI venous or AB venous
#3 TI vein or AB vein
#4 TI catheter* or AB catheter*
#5 #2 or #3
#6 #4 and #5
#7 #1 or #6
#8 TI impregn* or AB impregn*
#9 TI bond* or AB bond*
#10 TI coat* or AB coat*
#11 #8 or #10 or #9
#12 TI antimicrobial or AB antimicrobial
#13 TI antisep* or AB antisep*
#14 TI antibiotic$ or AB antibiotic$
#15 #12 or #13 or #14
#16 TI needleless or AB needleless
#17 TI connector$ or AB connector$
#18 #16 and #17
#19 MH "Chlorhexidine"
#20 MH "Silver Sulfadiazine"
#21 MH "Minocycline"
#22 MH "Rifampin"
#23 TI Rifampi* or AB Rifampi*
#24 TI Minocyclin* or AB Minocyclin*
#25 TI Silver sulphadiazine or AB Silver sulphadiazine
#26 TI Chlorhexidine or AB Chlorhexidine
#27 TI Arrowgard or AB Arrowgard
#28 TI Cook Spectrum or AB Cook Spectrum
#29 #23 or #24 or #25 or #26 or #27 or #28
#30 #29 or #11 or #15 or #18 or #19 or #20 or #21 or #22
#31 #7 and #30
#32 PT Clinical trial
#33 AB randomized or AB randomized
#34 AB random*
#35 TI trial
#36 MH "Clinical Trials"/explode
#37 #32 or #33 or #34 or #35 or #36
#38 #31 and #37
Appendix 6. Criteria for a judgment on the sources of bias in the included studies
1. Was the allocation sequence randomly generated?
- Yes, low risk of bias
A random (unpredictable) assignment sequence.
Examples of adequate methods of sequence generation are computer-generated random sequence, coin toss (for studies with two groups), rolling a dice (for studies with two or more groups), drawing of balls of different colours, dealing previously shuffled cards.
- No, high risk of bias
- Quasi-randomized approach: Examples of inadequate methods are: alternation, birth date, social insurance/security number, date in which they are invited to participate in the study, and hospital registration number.
- Non-random approaches: Allocation by judgement of the clinician; by preference of the participant; based on the results of a laboratory test or a series of tests; by availability of the intervention.
Insufficient information about the sequence generation process to permit judgement.
2. Was the treatment allocation adequately concealed?
- Yes, low risk of bias
Assignment must be generated independently by a person not responsible for determining the eligibility of the participants. This person has no information about the persons included in the trial and has no influence on the assignment sequence or on the decision about whether the person is eligible to enter the trial. Examples of adequate methods of allocation concealment are: Central allocation, including telephone, web-based, and pharmacy controlled, randomization; sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, sealed envelopes.
- No, high risk of bias
Examples of inadequate methods of allocation concealment are: alternate medical record numbers, unsealed envelopes; date of birth; case record number; alternation or rotation; an open list of random numbers any information in the study that indicated that investigators or participants could influence the intervention group.
Randomization stated but no information on method of allocation used is available.
3. Blinding was knowledge of the allocated interventions adequately prevented during the study?
Was the participant blinded to the intervention?
- Yes, low risk of bias
The treatment and control groups are indistinguishable for the participants or if the participant was described as blinded and the method of blinding was described.
- No, high risk of bias
Blinding of study participants attempted, but likely that the blinding could have been broken; participants were not blinded, and the non=blinding of others likely to introduce bias.
Was the care provider blinded to the intervention?
- Yes, low risk of bias
The treatment and control groups are indistinguishable for the care/treatment providers or if the care provider was described as blinded and the method of blinding was described.
- No, high risk of bias
Blinding of care/treatment providers attempted, but likely that the blinding could have been broken; care/treatment providers were not blinded, and the non blinding of others likely to introduce bias.
Was the outcome assessor blinded to the intervention?
- Yes, low risk of bias
Adequacy of blinding should be assessed for the primary outcomes. The outcome assessor was described as blinded and the method of blinding was described.
- No, high risk of bias
No blinding or incomplete blinding, and the outcome or outcome measurement is likely to be influenced by lack of blinding.
4. Were incomplete outcome data adequately addressed?
Was the drop-out rate described and acceptable?
The number of participants who were included in the study but did not complete the observation period or were not included in the analysis must be described and reasons given.
- Yes, low risk of bias
If the percentage of withdrawals and drop-outs does not exceed 20% for short-term follow-up and 30% for long-term follow-up and does not lead to substantial bias. (N.B. these percentages are arbitrary, not supported by literature);
No missing outcome data;
Reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias);
Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups;
Missing data have been imputed using appropriate methods.
- No, high risk of bias
Reason for missing outcome data likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups;
Were all randomizedparticipants analysed in the group to which they were allocated? (ITT analysis)
- Yes, low risk of bias
Specifically reported by authors that ITT was undertaken and this was confirmed on study assessment, or not stated but evident from study assessment that all randomized participants are reported/analysed in the group they were allocated to for the most important time point of outcome measurement (minus missing values) irrespective of non-compliance and co-interventions.
- No, high risk of bias
Lack of ITT confirmed on study assessment (patients who were randomized were not included in the analysis because they did not receive the study intervention, they withdrew from the study or were not included because of protocol violation) regardless of whether ITT reported or not.
‘As-treated’ analysis done with substantial departure of the intervention received from that assigned at randomization; potentially inappropriate application of simple imputation.
Described as ITT analysis, but unable to confirm on study assessment, or not reported and unable to confirm by study assessment.
5. Are reports of the study free of suggestion of selective outcome reporting?
- Yes, low risk of bias
If all the results from all prespecified outcomes have been adequately reported in the published report of the trial. This information is either obtained by comparing the protocol and the final trial report, or in the absence of the protocol, assessing that the published report includes enough information to make this judgment. Alternatively a judgement could be made if the trial report lists the outcomes of interest in the methods of the trial and then reports all these outcomes in the results section of the trial report.
- No, high risk of bias
Not all of the study’s prespecified primary outcomes have been reported;
One or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not prespecified;
One or more reported primary outcomes were not prespecified (unless clear justification for their reporting is provided, such as an unexpected adverse effect);
One or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta-analysis;
The study report fails to include results for a key outcome that would be expected to have been reported for such a study.
6. Other sources of potential bias:
PLEASE NOTE AUTHORS MUST DECIDE WHAT OTHER SOURCES OF POTENTIAL BIAS ARE APPROPRIATE TO THE REVIEW
THE DOMAINS BELOW ARE SUGGESTIONS:
Were the groups similar at baseline regarding the most important prognostic indicators?
Groups have to be similar at baseline regarding demographic factors, duration and severity of complaints, for example size and duration of ulcer. Alternatively if there were imbalances at baseline these have been accounted for in the analysis of the study.
Were co-interventions avoided or similar?
There were no co-interventions or there were co-interventions but they were similar between the treatment and control groups.
Was the compliance acceptable in all groups?
The review author determines if the compliance with the interventions is acceptable, based on the reported intensity, duration, number and frequency of sessions for both the treatment intervention and control intervention(s). For example, ultrasound treatment is usually administered over several sessions; therefore it is necessary to assess how many sessions each participant attended or if participants completed the course of an oral drug therapy. For single-session interventions (for example: surgery), this item is irrelevant.
Were the trials or trialists in receipt of financial support from agencies or organisations with a financial interest in the outcome of the trial?
Appendix 7. Studies with discrepancies between the prespecified outcomes in the methods and the outcomes reported in the results
Appendix 8. Additional searches in Internet resources
Last assessed as up-to-date: 15 March 2012.
Contributions of authors
Conceiving the review: Nai Ming Lai (NML)
Co-ordinating the review: NML
Undertaking manual searches: NML, Wilson Shu Cheng Pau (WP)
Screening search results: NML, Nathorn Chaiyakunapruk (NC)
Organizing retrieval of papers: Nai An Lai (NAL) , Elizabeth O'Riordan (EOR)
Screening retrieved papers against inclusion criteria: NAL, EOR
Appraising quality of papers: NML, NC
Abstracting data from papers: NML, NAL, NC, EOR, WP
Writing to authors of papers for additional information: NML, NAL, EOR, WP, NC
Providing additional data about papers: NML, NAL, NC, EOR
Obtaining and screening data on unpublished studies: NML, NAL, NC, EOR
Data management for the review: NML, NC, WP, EOR
Entering data into Review Manager 5 (Revman 5.2): NML
RevMan statistical data: NML, NC, EOR, NAL
Other statistical analysis not using RevMan: NML
Double entry of data: data entered by NML and cross-checked by a member of Sanjay Saint (SS)’s team (Andy Hickner), as acknowledged above.
Interpretation of data: NML, NC, NAL, EOR, SS
Statistical inferences: NML
Writing the review: NML, NC, NAL, WP, EOR, SS
Securing funding for the review: NML, SS
Performing previous work that was the foundation of the present study: N/A
Guarantor for the review (one author): NML
Person responsible for reading and checking review before submission: NAL, EOR, NC, WP
Declarations of interest
Nai Ming Lai: none known
Nathorn Chaiyakunapruk. none known
Nai An Lai: none known
Elizabeth O'Riordan: none known
Wilson Shu Cheng Pau: none known
Sanjay Saint has received numerous honoraria and speaking fees from academic medical centres, hospitals, group-purchasing organizations (e.g., Premier, VHA), specialty societies, state-based hospital associations, and non-profit foundations (e.g., Michigan Health and Hospital Association, Institute for Healthcare Improvement) for lectures about preventing healthcare-associated infection, implementing change, clinical problem-solving, and leadership. Dr Saint has provided expert testimony for legal cases focusing on medical malpractice. He has received grants/has grants pending from NIH; VA; AHRQ; Blue Cross Blue Shield of Michigan Foundation. Dr Saint has stock options with Doximity and is on the medical advisory board of Doximity, a new social networking site for physicians. However, Dr Saint has stated that these activities are not related to the topic of the review.
Sources of support
- No sources of support supplied
- SEA-ORCHID (South East Asian Optimising Reproductive and Child Health Outcomes in developing countries) Project, Not specified.Five-year project (2003 to 2008) aiming to promote synthesis and application of high level evidence in clinical practice especially on issues relevant to this region.
- International-Ann Arbor Safety Collaborative (I-A
Differences between protocol and review
- We changed the search platform from PubMed (National Library of Medicine) to MEDLINE (OVID SP), following a change of the contact author's affiliation where access to OVID database is possible. We translated the PubMed search strategy to MEDLINE (OVID SP) equivalent without any alteration of the content. We made corresponding revisions in our texts in the 'electronic searches' section.
- Types of studies: included cluster-randomized trials.
- Types of participants: amended with the addition of a statement accepting studies with multiple enrolments of the same participant.
- Types pf outcomes: we did not include the duration of catheterization as an outcome, as originally specified in our protocol, because this was listed as the characteristics of the population rather than an outcome in all the studies. Instead, we assessed catheter durability via the outcome of premature catheter removal, which was also listed in our protocol and reported as an outcome in the included studies.
- Types of outcomes: under the secondary outcomes, the unit of analysis has been amended from 'The number of participants' to 'The number of participants or catheters', to account for the effect of multiple enrolments.
- Types of outcomes: we removed the secondary outcome of 'duration of catheter use' as prespecified in our protocol, and chose to assess catheter durability via a single outcome of 'Number of participants or catheters with catheter failure or premature catheter removal'. This was because duration of catheter use was reported as a study characteristic rather than an outcome in all the included studies, in which there was a wide range of catheter indwelling time depending on the underlying conditions and the needs of the participants.
- Types of outcomes: in secondary outcome number eight, we added the number of participants who required systemic antibiotics during the course of the study alongside the total duration of antibiotics use.
- Assessment of risk of bias in included studies: rewritten in accordance with the new risk of bias assessment criteria in RevMan 5.2. A detailed description of the assessment criteria is included in the new Appendix 6.
- We added a paragraph under 'Data Synthesis' describing how we handled the rate data (such as CRBSI per 1000 catheter days) in individual groups as well as combined groups.
- Unit of analysis issues added: an extensive section was added to describe our approach in handling the unit of analysis issues that might arise from multiple enrolments of the same participants and from analysing cluster-randomized trials.
- We added a paragraph under the previously empty heading of 'Sensitivity analysis' detailing how we conducted our sensitivity analyses and reported our findings.
- In subgroup analysis point number three, we included the term 'predominant participant type' alongside 'study setting' as the criterion for separating studies into subgroups. We have also added baseline risk as a criterion for our subgroup analysis, and placed this as subgroup analysis number four. Additionally, we have revised our wording for subgroup analysis point number seven to indicate that we would separate studies conducted in different countries with different currencies used to measure costs in the outcome of cost effectiveness.
- References: we have replaced the CDC guideline for the prevention of intra-vascular related infections to an updated document that was published in 2011. The reference has been replaced accordingly from O'Grady 2002 to CDC 2011. We have also replaced cited reference for many of our outcomes from another Cochrane review on CVC (Webster 2011) to the primary source of reference, which is the CDC guideline (CDC 2011).
Medical Subject Headings (MeSH)
MeSH check words
* Indicates the major publication for the study