Retention versus replacement or removal of catheters for people with confirmed or suspected intravascular catheter‐related infections

Objectives

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To assess the effects (benefits and harms) of catheter retention (or delayed removal after an initial catheter retention strategy) compared to prompt catheter replacement or removal in people of all ages with a short‐ or long‐term central venous catheter or any arterial catheter, and either (i) catheter‐related or ‐associated infection, (ii) bloodstream infection from an uncertain source or (iii) sepsis from an uncertain source.

Background

Description of the condition

Short‐ and long‐term intravascular catheters are widely used in hospitalised people. In people with sepsis in whom no clear focus is identified, an intravascular catheter is often considered a potential source. Catheter‐associated bloodstream infections pose a significant medical burden on both critically and non‐critically ill people. In intensive care units (ICUs), when current prevention bundles for the insertion and maintenance of central venous catheters (CVCs) are applied, CVC‐associated bloodstream infections occur in 0.5% to 1.5% of persons [1]. The median incidence density ranges from 0.5 to 2.5 episodes per 1000 catheter‐days. The risk of arterial catheter‐related bloodstream infections appears similar to that observed with CVCs, which is approximately 1 episode per 1000 catheter‐days [1].

While 'catheter‐associated' bloodstream infection refers to those suspected to be linked to the presence of a catheter, 'catheter‐related' bloodstream infections generally require more specific evidence. For example, when a blood culture from the catheter yields the same organism as a peripheral blood culture, and no other source of bacteraemia is identified, the infection would be classified as catheter‐related. For the purpose of this review, we will consider these terms as one clinical entity.

Appropriate management is crucial for persons with a CVC or arterial catheter when a bloodstream infection is confirmed or suspected. Recent recommendations from the French Society of Intensive Care Medicine suggest that the catheter should be removed when a catheter‐related infection is confirmed or suspected [2]. However, their practice statement suggests retention if a catheter‐related infection is suspected but not confirmed or if the catheter cannot be replaced without major risk [2]. Previous guidance from the Infectious Diseases Society of America in 2009 recommended catheter removal for people with short‐term CVC‐related or arterial catheter‐related bloodstream infection [3]. However, for catheter‐related bloodstream infection amongst people undergoing haemodialysis with tunnelled catheters, retention was deemed reasonable for infections caused by Gram‐negative bacteria other than Pseudomonas species or coagulase‐negative staphylococci [3].

Some previous observational studies have examined people with bloodstream infections and shown an association between delayed CVC removal and increased mortality [4, 5]. However, one observational study found that delayed removal for persons with suspected CVC‐related infections was not associated with increased mortality [6]. Furthermore, another study showed that immediate CVC replacement for persons with suspected catheter‐related bloodstream infection was associated with increased mortality [7]. While arterial catheters are recognised as a source of catheter‐related bloodstream infections [8], there is less information available regarding their management in comparison to CVCs.

Given the guideline statements and the mixed results of previous studies, the appropriate practice for intravascular catheter‐related infections remains uncertain and requires further investigation.

In this review, 'intravascular catheters' include a wide range of short‐ and long‐term central venous catheters (e.g. non‐tunnelled and tunnelled catheters, peripherally inserted central catheters, haemodialysis catheters, pulmonary artery catheters, extracorporeal life support catheters) and arterial catheters (peripheral and central).

Description of the intervention and how it might work

Use of a catheter retention strategy (i.e. leaving the catheter in place), alongside antibiotic treatment, has fewer risks of mechanical complications than a removal strategy, as well as no extra cost for additional catheters, and no additional workload related to new catheter insertion. However, the retention strategy could lead to incomplete infection source control and additional infection‐related clinical morbidity or mortality.

In contrast, although catheter removal provides source control for catheter‐related bacteraemia, catheter replacement is necessary for persons who need to maintain vascular access for infusions or invasive monitoring. Catheter replacement may be performed either by removal and insertion of a new catheter at a different site, or by exchange over a guidewire, depending on clinical context and trial protocol. Catheter replacement carries the possibility of mechanical complications such as arterial puncture, non‐target venous puncture, bleeding; and, for upper body central catheters, pneumothorax and haemothorax. Furthermore, replacement may result in an unnecessary procedure for the patient, which may be especially relevant if the diagnosis is eventually confirmed as not being a catheter‐related infection. In theory, replacement may not result in adequate infection source control, particularly if the blood has not been sterilised.

Why it is important to do this review

Overall, it seems there is little high‐quality evidence to inform the optimal strategy—retention versus replacement or removal—when individuals have a catheter in place and have a documented or suspected bloodstream infection; however, to our knowledge, there have been no systematic reviews addressing suspected intravascular catheter‐associated infections. As the choice between catheter retention and prompt replacement or removal may influence timely infection source control and patient outcomes, we are conducting this Cochrane review to provide a reliable and up‐to‐date summary and evaluation of the clinical trial evidence.

Objectives

To assess the effects (benefits and harms) of catheter retention (or delayed removal after an initial catheter retention strategy) compared to prompt catheter replacement or removal in people of all ages with a short‐ or long‐term central venous catheter or any arterial catheter, and either (i) catheter‐related or ‐associated infection, (ii) bloodstream infection from an uncertain source or (iii) sepsis from an uncertain source.

Methods

Criteria for considering studies for this review

Types of studies

We will consider all individually randomised controlled trials (RCTs) and quasi‐randomised studies (defined as studies that evaluate the effects of an intervention without using truly random assignment of participants to groups, e.g. alternate allocation). We will also consider cluster‐randomised trials (and variants such as step‐wedge and cluster‐cross‐over) for which the analysis appropriately accounts for clustered data, or that report the intracluster correlation coefficient (ICC) [9], or for which an estimate of the ICC is available from an external source.

We will exclude cross‐over studies since it is not feasible for participants in this review to experience the same conditions twice.

Types of participants

We will include people of any age with confirmed or suspected central venous catheters (CVC), or any arterial catheter‐related or ‐associated infections.

CVCs will include short‐term central catheters, peripherally inserted central catheters (PICCs), pulmonary artery catheters ("Swan‐Ganz" catheters), tunnelled central catheters, central catheters for haemodialysis, intravascular temperature control device central catheters, and extracorporeal life support catheters. Arterial catheters will include peripheral arterial catheters (e.g. radial, brachial, axillary and dorsalis pedis arterial catheters) and central arterial catheters or cannulas (e.g. femoral arterial catheters, intra‐aortic balloon pumping catheters and extracorporeal life support catheters).

When data are available and the catheter‐associated infections are due to bloodstream infections, we will record whether the bloodstream infection is deemed related to the catheter or whether it is associated with another non‐catheter source of infection.

We will not exclude any participants based on demographic factors.

When only a subset of participants in a study meet our eligibility criteria (e.g. in a study including both central venous catheters and peripheral intravenous catheters), we will handle them as follows.

• If the study reports results separately for the eligible subgroup (e.g. CVCs), we will include the subgroup data as reported.

• If no subgroup‐specific data are available, we will contact the study authors to request the details. If the author is not able to provide the necessary data, we will include all participants in our primary analysis, but we will perform sensitivity analysis excluding such studies to assess the robustness of the findings.

Types of interventions

Treatment intervention: intravascular catheter retention (leaving the catheter in place as the initial strategy)

Comparator: intravascular catheter replacement (either by exchange over a guidewire or removal and replacement with an entirely new catheter), or removal without replacement (i.e. no longer needed)

We will classify participants according to the strategy to which they were initially assigned. If a trial protocol allows catheter replacement/removal after predefined clinical deterioration following initial retention, we will still consider this as part of the 'retention strategy' (i.e. delayed removal after initial retention), consistent with the trial’s intention‐to‐treat comparison.

Outcome measures

Critical outcomes

• All‐cause mortality (death from any cause)

Where reported and where appropriate for the population, we plan to use the longest follow‐up data from each trial, up to 180 days, regardless of the duration of follow‐up (e.g. studies may measure mortality at ICU discharge, 28‐ or 30‐day mortality, mortality at hospital discharge, 60‐day mortality, 90‐day mortality, etc).

Important outcomes

• Complications that may be related to the catheter retention strategy

  • Persistently positive blood cultures for the bloodstream infection‐causing organism if bloodstream infections have been confirmed at study enrolment

  • Metastatic foci of infection (including endocarditis, abscess, etc)

  • Progression of illness (e.g. development of septic shock as defined by initiation of intravenous vasoactive medications or development of increased lactate or organ dysfunction)

  • Other adverse effects defined by the study authors

• Complications that may be related to intravascular catheter replacement (as defined by study authors)

  • Arterial puncture by CVC replacement

  • Pseudoaneurysm by CVC or arterial catheter replacement

  • Clinically important bleeding

  • Pneumothorax

  • Haemothorax

  • Significant haematoma

  • Blood transfusion of at least 1 unit of red blood cells

  • Discomfort or pain

  • More than one placement attempt

  • Other adverse effects defined by the study authors

• Quality of life measured on any validated scale, such as the 36‐Item Short Form (SF‐36) Survey [10], at the longest time point reported

• Number of catheters replaced and the time from randomisation to the first catheter replacement/removal (and subsequent replacements, if reported)

• Antibiotic use (by antibiotic‐days or antibiotic‐free days)

• Length of hospital stay during the index hospitalisation (i.e. the hospital admission during which the index bloodstream infection or suspected catheter‐related infection occurred)

Search methods for identification of studies

Electronic searches

We will identify RCTs through systematic searches of the following bibliographic databases.

• Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library

• MEDLINE Ovid

• Embase Ovid (if available)

• Web of Science Core Collection (Clarivate Analytics)

We will modify the initial search strategy for MEDLINE (Ovid) to use across other databases (Supplementary material 1). Additionally, a search will be conducted on ClinicalTrials.gov (https://www.ClinicalTrials.gov) and the WHO International Clinical Trials Registry Platform (ICTRP) Search Portal (https://apps.who.int/trialsearch) for any ongoing or unpublished trials. All databases will be searched from their inception to the present, with no restrictions on the language of publication or publication status. We will not search separately for adverse events; instead, we will consider only those described in the included studies.

Searching other resources

We will check reference lists of all included studies and relevant systematic reviews for additional trial references.

We will investigate any pertinent retraction statements and errata for the studies included in the review by:

• checking retraction/erratum notices linked within bibliographic databases (e.g. MEDLINE and Embase);

• checking the publisher’s webpage for each included study (including Crossmark status where available); and

• searching the Retraction Watch Database for included study citations/DOIs (https://retractionwatch.com/).

We will exclude and document any retracted studies.

Data collection and analysis

Selection of studies

Two authors (JI and NKJA) will independently screen titles and abstracts to determine the inclusion of potential studies identified through our search. WE will retrieve the full text of any citation deemed potentially eligible by either review author. JI and NKJA will further screen these full texts to select studies for inclusion, documenting the reasons for exclusion of those deemed ineligible. We will resolve any disagreements through discussion or, if necessary, by consulting a third review author (ND or RF). We will identify and remove duplicates, consolidating reports of the same study while focusing on individual studies as the units of interest. We will document the selection process in detail to construct a PRISMA flow diagram. We will compile our reasons for exclusions in a 'Characteristics of excluded studies' table.

Data extraction and management

We will use a standardised data collection form to extract study characteristics and outcome data. Before applying it to the rest of the included studies, we will conduct a pilot test of this data extraction sheet on at least one study. Two review authors (JI and NKJA) will carry out data extraction, collecting the following study characteristics.

• Methods: total duration of study, details of any 'run‐in' period, number of study centres and location, study setting, and date of study

• Participants: number randomised, number lost to follow‐up/ withdrawn, number analysed, mean age, sex, inclusion and exclusion criteria

• Interventions: detailed description of the assigned strategy (retention/delayed removal versus replacement/removal), including timing windows, criteria for replacement/removal, method of replacement (guidewire exchange versus new catheter at a different site) and key co‐interventions (e.g. antibiotic strategy, lock therapy, planned catheter salvage)

• Outcomes: outcome definitions, measurement instruments/scales used (where applicable) and assessment time points

• Notes: funding for the trial, and notable conflicts of interest of the trial authors

• Supplementary material: we will make the data extraction form available on reasonable request or provide it as part of the supplementary material with the completed review.

We will resolve any disagreements by consensus or through involvement of a third review author (ND or RF). One review author (JI) will enter the data into Cochrane's Review Manager (RevMan) software [11].

Risk of bias assessment in included studies

The same two review authors (JI and NKJA) will independently evaluate the risk of bias for each study. We will use Cochrane's risk of bias tool RoB 2 as the primary tool [12], and we will use the Risk Of Bias instrument for Use in SysTematic reviews‐for Randomised Controlled Trials (ROBUST‐RCT) as a secondary tool [13]. Any disagreements will be addressed through discussion or with the assistance of another review author (ND or RF). For the domain related to outcome measures, our assessment will focus on the following main outcomes, which we will include in our summary of findings table.

• All‐cause mortality

• Complications that may be related to the catheter retention strategy

• Complications that may be related to intravascular catheter replacement

• Quality of life measured on any validated scale

• Number of catheters replaced, and timing relative to randomisation

• Antibiotic use (by antibiotic‐days or antibiotic‐free days)

• Length of hospital stay during the index hospitalisation

Our time points of interest for each outcome measurement are specified in the Outcome measures and Certainty of the evidence assessment sections. We will evaluate bias risk based on the following domains.

RoB 2 domains

1. Bias arising from the randomisation process

2. Bias due to deviations from intended interventions

3. Bias due to missing outcome data

4. Bias in measurement of the outcome

5. Bias in selection of the reported result

For RoB 2 domains, we will evaluate and classify each potential source of bias as the following response options: yes, probably yes, probably no, no, or no information. Once the signalling questions are answered, the next step is to reach a risk‐of‐bias judgement and assign one of the following three levels to each domain.

• Low risk of bias

• Some concerns

• High risk of bias

We will add the domain 'Bias arising from the timing of identification and recruitment of participants' for cluster‐randomised trials.

ROBUST‐RCT domains

1. Random sequence generation

2. Allocation concealment

3. Blinding of participants

4. Blinding of healthcare providers

5. Blinding of outcome assessors

6. Outcome data not included in analysis

For ROBUST‐RCT domains, we will evaluate and classify each potential source of bias as low risk of bias, high risk of bias or unclear risk of bias.

We will provide a rationale for our assessment in a risk of bias table and summarise the risk of bias judgements across various studies for each listed domain. If the risk of bias relates to unpublished data or communications with trial authors, we will indicate this in the risk of bias table.

Measures of treatment effect

We will analyse dichotomous variables (i.e. variables in two mutually exclusive categories such as all‐cause mortality, complications that may be related to the catheter retention strategy, complications that may be related to intravascular catheter replacement) as risk ratios (RRs) with 95% confidence intervals (CIs). For continuous outcomes (such as length of hospital stay), we will present mean differences (MDs) with 95% CIs. When all trials employ the same quality of life scale, we will provide MDs along with 95% CIs and define the minimally important change on the scale as the minimally important difference in quality of life. In cases where trials use different quality of life scales, we will present standardised mean differences (SMDs) with 95% CIs.

We will analyse the outcome of the number of catheters replaced in either or both of two ways: rate ratios if the count data of the outcome and the follow‐up period are available, or hazard ratios if data are presented using Kaplan‐Meier curves or if hazard ratios are presented. When the information about follow‐up periods or time‐to‐event is unavailable, we will present it as a continuous outcome, using the total number of catheter replacements during the study follow‐up period. We will enter data into RevMan [11] in a way that ensures a consistent direction of effect is maintained.

Unit of analysis issues

When analysing multi‐armed trials, we will merge all relevant experimental intervention groups of the study into a single group and all relevant control intervention groups into a single control group. We will exclude from the analysis any arm that cannot be categorised as either experimental or comparator.

To incorporate the results of cluster‐RCTs with ICCs alongside individual RCTs, we will seek the effective sample size for cluster‐RCTs using the design effect derived from the number of clusters and the ICC. Once the cluster‐RCTs have been reduced to the effective sample size, we will merge the results of both individual and cluster‐RCTs in a meta‐analysis. For cluster‐RCTs that report odds ratios from mixed models, we will convert these to RRs [14].

For cluster‐cross‐over trials, where the same clusters (e.g. wards or ICUs) are exposed to different interventions at different time periods, we will account for the correlation between observations within clusters across periods. When available, we will extract effect estimates that appropriately adjust for both clustering and period effects (e.g. from mixed‐effects models or generalised estimating equations). If such adjusted estimates are not reported, we will contact the study authors for additional data or use approximate methods (e.g, using design effects) based on available information about the number of clusters, cluster sizes, and ICCs. These approaches will also be applied, where appropriate, to other variants of cluster‐randomised designs.

Dealing with missing data

We will reach out to investigators or study sponsors to confirm key study features and gather any missing numerical outcome data, especially for studies represented only as abstracts. When feasible, we will use the RevMan Web calculator to derive missing standard deviations from other trial data, such as confidence intervals. If this is not achievable and the lack of data is expected to introduce significant bias, we will assess the influence of including these studies on the overall findings through sensitivity analysis. We will conduct analysis on an intention‐to‐treat basis for all outcomes whenever possible.

Reporting bias assessment

To evaluate reporting biases, we will use funnel plots if 10 or more studies are included in a meta‐analysis, and we will assess asymmetry in the funnel plot through visual inspection. Along with funnel plots, we will conduct Egger’s test to examine small‐study effects [15].

Synthesis methods

We will conduct meta‐analyses for the outcomes only when it is appropriate, meaning that the treatments, participants, and clinical questions are sufficiently similar to allow for sensible pooling. We will use the random‐effects model since we expect there will be clinical diversity. The Restricted Maximum Likelihood (REML) estimator will be used to estimate between‐trial variance. The Hartung‐Knapp‐Sidik‐Jonkman method will be used to calculate a confidence interval for the meta‐analysis effect estimate when there are at least three studies and the estimate of heterogeneity is greater than zero. In other scenarios (i.e. in pooled analyses of two studies, or where the estimate of heterogeneity is equal to zero), we will use the Wald‐type method [12].

Investigation of heterogeneity and subgroup analysis

We will assess clinical heterogeneity by comparing the characteristics of participants (e.g. age, clinical setting, catheter type, or infection type), interventions (e.g. retention, delayed removal, replacement, or removal), and outcome definitions across studies. We will assess methodological heterogeneity by reviewing study design features, such as randomisation methods, blinding, and risk of bias domains. Where important clinical or methodological heterogeneity is identified, we will explore potential reasons through prespecified subgroup analyses and sensitivity analyses.

For statistical heterogeneity, we plan to carry out the following subgroup analyses.

• Confirmed versus suspected infections, hypothesising that the retention strategy is more harmful for confirmed infections.

• Participants younger than one year versus those aged one year or older, hypothesising that the retention strategy is more harmful for those younger than one year.

• CVCs versus any arterial catheters, hypothesising that the retention strategy is more harmful for CVCs.

• Critically‐ill versus non‐critically ill persons, hypothesising that the retention strategy is more harmful for critically‐ill persons.

We will conduct subgroup analyses for the critical outcome (all‐cause mortality) only.

We will first visually inspect forest plots to assess the direction and size of effects, as well as the extent of overlap in confidence intervals. We will use the I² statistic to measure heterogeneity across the trials in each analysis. However, we recognise the considerable uncertainty in the I² value when the number of studies is limited.

We will follow the recommendations for the threshold of the Cochrane Handbook for Systematic Reviews of Interventions [12].

• 0% to 40% might not be important.

• 30% to 60% may represent moderate heterogeneity.

• 50% to 90% may represent substantial heterogeneity.

• 75% to 100% may represent considerable (high) heterogeneity.

If we observe significant and considerable heterogeneity, we will document it and investigate potential reasons using our prespecified subgroup analysis.

For any subgroup effects examined statistically, we will use the formal test for subgroup differences in RevMan, which applies a Chi² test for interaction, and we will assess for the credibility of subgroup findings if the P value is less than 0.1 using the Instrument to assess the Credibility of Effect Modification Analyses (ICEMAN) criteria for meta‐analyses [16, 17].

Equity‐related assessment

Equity‐related factors such as age, sex, and other PROGRESS‐Plus characteristics (Place of residence, Race/ethnicity, Occupation, Gender, Religion, Education, Socioeconomic status, and Social capital, plus other context‐specific factors such as comorbidity burden) may influence both the management and outcomes of catheter‐related infections. We will extract and report these characteristics where available in the included studies. No specific subgroup analyses by equity‐relevant factors are planned, but any relevant findings will be described narratively in the review.

Sensitivity analysis

We plan to perform sensitivity analyses restricted to studies at overall low risk of bias, and sensitivity analyses that remove any studies where only a subset of participants meet the eligibility criteria. We may perform additional sensitivity analyses if, during the conduct of the review, we identify key methodological factors or decisions that could influence the effect sizes.

Certainty of the evidence assessment

Summary of review findings and assessment of the certainty of the evidence

Before writing the results and conclusions of our review, we intend to extract study data and organise our comparisons into data tables. Following the methods and recommendations provided in the Cochrane Handbook for Systematic Reviews of Interventions [12], and using the GRADEpro GDT software [18], we will produce a single comprehensive summary of findings (SoF) table comparing catheter retention with replacement or removal strategies for confirmed or suspected infections associated with CVCs or any arterial catheters. We will present all of our prespecified critical and important outcomes in the SoF table, whether data are available or not. The template for our SoF table is shown in Table 1. For each outcome, we will prioritise data from the time points and using the measurement methods specified in the Outcome measures section in the Methods of this review.

1. Template of 'Summary of findings' table.

Outcomes	Anticipated absolute effects (95% CI)		Relative effect (95% CI)	Number of participants (number of studies)	Certainty of the evidence (GRADE)	Comments
	Risk with catheter retention strategy	Risk with catheter replacement/removal strategy
All‐cause mortality
Complications related to catheter retention strategy
Complications related to catheter replacement
Quality of life
Number of catheters replaced and timing
Antibiotic days or antibiotic‐free days
Length of hospital stay

• All‐cause mortality (mortality of any cause): to minimise multiplicity, for each outcome we will include one prespecified time point per study. We will prioritise the longest follow‐up period reported (up to 180 days), following the hierarchy defined in the Outcome measures section. If both a fixed‐day time point and an event‐based time point are reported (e.g. ICU discharge), we will prioritise the fixed‐day time point.

• Complications that may be related to the catheter retention strategy: we will include events reported during the index hospitalisation or the predefined study follow‐up period, as defined by trial authors.

• Complications that may be related to intravascular catheter replacement: we will include events reported during the index hospitalisation or the predefined study follow‐up period, as defined by trial authors.

• Quality of life measured on any validated scale (e.g. SF‐36), prioritising the longest follow‐up reported, regardless of the instrument used

• Number of catheters replaced and timing: we will assess this outcome based on the total number of catheter replacements during the study follow‐up period. When sufficient data are available, we will consider rate ratios (for count data with known follow‐up duration) or hazard ratios (for time‐to‐event data). If such information is not available, we will use the total number of replacements as a continuous measure.

• Antibiotic use: we will extract the total number of antibiotic days (or antibiotic‐free days).

• Length of hospital stay: we will use total hospital stay during the index admission when available.

To evaluate the certainty of the evidence from studies contributing data to the meta‐analyses for these outcomes, we will assess the five GRADE criteria [12]: risk of bias, inconsistency, imprecision, indirectness, and publication bias. Two review authors (JI and NKJA) will independently assess evidence certainty, resolving any disagreements through discussion between themselves or with a third author (ND or RF). We will justify, document and integrate these judgements into our description of the results for each outcome. We will also provide our GRADE levels of certainty in the SoF table and provide justifications for downgrading the certainty of the evidence in footnotes, and we will include comments to enhance the reader's understanding of the results, where appropriate.

Consumer involvement

We acknowledge the importance of involving consumers, including patients, carers and members of the public, in Cochrane reviews to enhance the relevance and applicability of evidence. However, due to limited resources and time constraints, we will not directly involve consumers in this review. We recognise the potential value of incorporating consumer perspectives and will consider seeking input, for example, through consultation or feedback, during dissemination of the review findings or in future projects related to the review.

Supporting Information

Supplementary materials are available with the online version of this article: 10.1002/14651858.CD016329.

Supplementary materials are published alongside the article and contain additional data and information that support or enhance the article. Supplementary materials may not be subject to the same editorial scrutiny as the content of the article and Cochrane has not copyedited, typeset or proofread these materials. The material in these sections has been supplied by the author(s) for publication under a Licence for Publication and the author(s) are solely responsible for the material. Cochrane accordingly gives no representations or warranties of any kind in relation to, and accepts no liability for any reliance on or use of, such material.

Supplementary material 1 Search strategies

New

Additional information

Contributions of authors

All authors contributed to the drafting of the protocol (Junichi Izawa (JI), Nick Daneman (ND), Neill KJ Adhikari (NKJA), Robert Fowler (RF)).

For the full review, the contributions of the review authors will be as follows.

Selecting studies: JI, NKJA

Extracting data from studies: JI, NKJA

Carrying out the analysis: JI, NKJA

Interpreting the analysis: JI, NKJA

Drafting and commenting on the final review: JI, ND, NKJA, RF

Resolving disagreements: ND, NKJA, RF

Coordinating the review: RF

Declarations of interest

Junichi Izawa has no conflicts of interest.

Nick Daneman has no conflicts of interest.

Neill KJ Adhikari has no conflicts of interest.

Robert Fowler has no conflicts of interest.

Sources of support

Internal sources

• No sources of support provided

External sources

• No sources of support provided

Registration and protocol

This topic has been approved for a Cochrane review.

Data, code and other materials

Data sharing is not applicable to this article as it is a protocol, so no datasets were generated or analysed.