Criteria for considering studies for this review
Types of studies
We will include all published and unpublished randomised controlled trials (RCTs). We considered whether to include other types of study design but decided not to do so. The reasons for this decision were as follows.
1. Several RCTs are known to be available.
2. Designs other than RCTs would almost certainly be at higher risk of bias than RCTs. This is likely to be the case even for possible harms of some cardioplegia methods (see below, 'Quality of evidence on adverse effects'). Also, one type of cardioplegia method may be preferred by a surgical team on the basis of a patient’s risk of postoperative morbidity (that is indicated), putting a non-randomised study at high risk of confounding by indication.
3. We do not suspect that relevant outcomes are more likely to have been studied in non-RCTs than RCTs.
Types of participants
Men and women (18 years or older) undergoing isolated CABG.
Types of interventions
a. Comparison of blood cardioplegia (any temperature and composition) versus crystalloid cardioplegia (any temperature and composition).
b. Comparison of warm (21 °C to 37 °C) blood cardioplegia versus cold (≤ 20 °C) blood cardioplegia.
Although most investigators have defined ‘warm’ cardioplegia as 32 °C to 37 °C and ‘cold’ cardioplegia as < 10 °C, a variety of temperatures have been reported in the literature (warm, 28 °C to 37 °C; cold, 4 °C to 20 °C). Some investigators have split the ‘warm’ range into two groups: warm (37 °C); and lukewarm (28 °C to 30 °C). For the purpose of this review, we have defined two mutually exclusive ranges for warm (21 °C to 37 °C) and cold (≤ 20 °C) to allow (theoretically) every participant to be characterised. We have taken room temperature (20 °C for numerical convenience) as the cut-off point since anything colder than this must represent some active form of cooling and anything warmer than this must represent some active form of warming.
Types of outcome measures
All-cause mortality (30 days)
Perioperative myocardial infarction (MI)
Need for renal dialysis (when not on dialysis preoperatively)
Length of hospital stay*
Intra-aortic balloon pump usage
Biomarkers of myocardial injury (e.g. cardiac troponin and CK-MB)
Serious complications are rare and we will explore the feasibility of constructing a composite dichotomous outcome from MI, stroke, and need for renal dialysis. We will consider including other serious complications if consistently reported by primary studies. We will also investigate the feasibility of combining different outcome measures that are related from different studies (for example low output syndrome and inotropic support).
*We are aware of the risk of bias issues associated with this type of outcome. We have explained our strategy of analysis in the 'Measures of the treatment effect' section.
Search methods for identification of studies
We will not restrict the searches by language or publication status.
The Cochrane Injuries Group Trials Search Co-ordinator will search the following electronic databases:
Cochrane Injuries Group Specialised Register (latest version);
CENTRAL, DARE, HTA and NHS EED (The Cochrane Library, latest Issue);
MEDLINE (OvidSP) (1946 to most recent date available);
EMBASE Classic + EMBASE (OvidSP) (1947 to most recent date available);
ISI Web of Science: Science Citation Index Expanded (SCI-EXPANDED) (1970 to most recent date available);
ISI Web of Science: Conference Proceedings Citation Index-Science (CPCI-S) (1990 to most recent date available).
Searching other resources
The authors will search the reference lists of all included studies and relevant published reviews. We will also contact experts in the field and trial authors for further information or unpublished data.
We will search the following trials registers to identify ongoing and unpublished studies: World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (http://apps.who.int/trialsearch/) and Current Controlled Trials (http://www.controlled-trials.com).
Data collection and analysis
The Cochrane Injuries Group Trials Search Co-ordinator will design the search strategies, run the searches, and collate the results, which will then be screened by two of the review authors (MP and JH).
Selection of studies
Two review authors (MP and JH) will independently assess for inclusion the full text papers for all the studies that were identified as potentially eligible by screening. Any disagreements will be resolved by discussion with a third review author (BR). Duplicate publications will be excluded. The study selection process will be presented in a PRISMA flow diagram.
Data extraction and management
Two review authors (MP and JH) will independently extract the data from all included studies. Disagreements will be resolved through discussion and consensus with a third review author (BR). We plan to extract the following information and data from each study.
1. Publication details (authors, title, date of publication, language, country of origin, funding source).
2. Study characteristics (setting, study design, method of randomisation (sequence generation, concealment of allocation), blinding of outcomes, number of patients randomised to each group, sample size calculation if performed).
3. Participant characteristics (inclusion and exclusion criteria, demographics, clinical characteristics (e.g. extent of disease, presence of co-morbidities), any baseline differences between treatment groups).
4. Procedure characteristics (e.g. urgent or elective surgery, cardiopulmonary bypass time, and aorta cross-clamp time).
5. Cardioplegia characteristics (temperature in each group, timing of cardioplegia delivery (intermittent or continuous), route of delivery (antegrade, retrograde, or mixed)).
6. Outcomes (as detailed in previous section). We will record the number of participants assessed for each outcome, mean values and standard deviations (if available), or medians and interquartile ranges for continuous data, number of events in each group for dichotomous data, and any reported summary statistics (e.g. effect estimates, confidence intervals (CI), standard errors (SE), ranges).
We will contact the trial authors for information if any of the above data items are missing. MP will enter the data into Review Manager software, and JH will be responsible for checking the data that is entered.
Assessment of risk of bias in included studies
Two review authors (MP and JH) will independently assess the risk of bias in each included study using the domain-based evaluation tool described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Disagreements will be resolved by discussion with a third review author (BR).
We will assess the following domains as low risk of bias, unclear or high risk of bias.
1. Generation of allocation sequence (selection bias).
2. Allocation concealment (selection bias).
3. Blinding of participants, personnel, and outcome assessors (performance bias).
4. Incomplete outcome data (attrition bias).
5. Selective reporting (reporting bias).
6. Other sources (e.g. presentation data bias, sampling bias, sponsorship bias).
Quality of evidence on adverse effects
Different methods of cardioplegia have been introduced based on the assumption that they reduce the risk of perioperative morbid events. However, possible harms of a cardioplegia method are likely to be manifested through the same morbid events. Therefore, most harms of a cardioplegia method will be captured by the outcomes already identified. Nevertheless, we will extract any additional information from included studies about adverse effects that researchers suspect were related to the method of cardioplegia studied and assess the quality of evidence on adverse events as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Measures of treatment effect
We will calculate risk ratios (RRs) and 95% confidence intervals (CIs) for the following dichotomous outcomes: all-cause mortality at 30 days; perioperative myocardial infarction (MI); permanent stroke; need for renal dialysis; atrial fibrillation; inotrope usage; intra-aortic balloon pump usage; adverse effects.
For continuous outcomes (biomarkers of cardiac injury) we will calculate mean differences (MD) and 95% CIs when the same scale is used to measure an outcome across separate studies, or standardised mean differences (SMD) and 95% CIs if results are reported using different scales or different versions of the same scale.
Length of hospital stay is strictly time-to-event data. However, we suspect that this will be reported as continuous data with treatment effects described as MDs with 95% CIs, despite the high likelihood that the raw data were not normally distributed and the risk of early postoperative deaths introducing bias. For example, the mean hospital stay for a method of cardioplegia for which there were several early deaths may be biased downwards, that is too short duration. For group sample sizes > 20, MDs can be pooled with reasonable confidence (on the basis of the central limit theorem). If authors report medians and interquartile ranges for length of hospital stay, we will calculate hazard ratios (HR) and 95% CIs (allowing for censoring). Care will be taken to assess the risk of bias from selective attrition. If no appropriate data are available, then length of stay will be reported narratively.
Unit of analysis issues
We do not anticipate that there will be any unusual study designs (for example cluster-randomised trials). We will avoid double-counting of participants where there are multiple interventions in the same trial (for example warm blood, cold blood, and cold crystalloid). We will also take into account multiple observations for the same outcome (for example serial troponin, or other biomarker measurements) on the basis of the level of data available.
Dealing with missing data
We will contact trial authors for important variables that were not reported in the papers or missing statistics. However, given that many of the trials relevant to this systematic review were conducted over 15 years ago, we do not anticipate a high response rate. Therefore our main analyses will be based on participants who completed the trial, but we will perform sensitivity analysis for worst and best case scenarios and discuss the potential impact of these on the findings of the review in the 'Discussion' section.
Assessment of heterogeneity
We will assess clinical heterogeneity across studies by examining the details of participants, baseline data, interventions, and outcomes to determine whether studies are similar. We will quantify statistical heterogeneity using the I2 statistic; we will consider the statistical heterogeneity to be high if I2 > 50% (Higgins 2011). We will attempt to explain any observed clinical or statistical heterogeneity in the results of the review.
Assessment of reporting biases
If 10 or more studies are included in an analysis, we will use funnel plots to assess reporting bias at the study level (publication bias). Selective outcome reporting will be assessed as part of the domain-based assessment of the risk of bias in the included studies.
We will summarise trial findings using a random-effects model for both comparisons (blood cardioplegia versus crystalloid cardioplegia and warm blood cardioplegia versus cold blood cardioplegia) given that there is substantial variability in cardioplegia delivery (a variety of cardioplegia temperatures included in the ‘warm’ and ‘cold’ groups, intermittent versus continuous delivery, antegrade versus retrograde delivery, variation in composition of blood and crystalloid cardioplegia, variation in volume of delivery, etc.). We will summarise the included studies in narrative form if we do not find trials that are sufficiently similar to justify a meta-analysis.
Subgroup analysis and investigation of heterogeneity
We will perform subgroup analyses for the primary outcome and secondary outcomes (if data are available from sufficient trials) for the following.
1. Mode of delivery of cardioplegia (antegrade versus retrograde).
2. Type of delivery for cardioplegia (intermittent versus continuous).
3. Temperature (warm, 32 °C to 37 °C, versus lukewarm, 28 °C to 31 °C).
Some of these factors may be investigated in meta-regression depending on how findings are reported.
We plan to conduct sensitivity analyses on trials classified as having good allocation concealment versus trials classified as having poor allocation concealment. We will also conduct a sensitivity analysis restricted to published trials. We will assess the impact of any study that has a large effect size on the results of the meta-analysis.