Inotropes for the prevention of low cardiac output syndrome and mortality for paediatric patients undergoing surgery for congenital heart disease: a network meta‐analysis
Abstract
Objectives
This is a protocol for a Cochrane Review (intervention). The objectives are as follows:
-
To compare in a network meta‐analysis the efficacy and safety of different prophylactic medications either against no treatment, or placebo, or against each other (as mono‐ or combination prophylaxis) for the prevention of low cardiac output syndrome and mortality for paediatric patients undergoing surgery for congenital heart disease.
-
To generate a clinically useful ranking of prophylactic medications (mono‐ and combination prophylaxis) according to efficacy and safety.
Background
Glossary in Appendix 1.
Description of the condition
Cardiac surgery for congenital heart disease (CHD) is often performed at a young patient age (Warnes 2001). Depending on the type of surgical intervention and patient age, postoperative low cardiac output syndrome (LCOS), also termed postoperative myocardial dysfunction, postoperative cardiocirculatory dysfunction, postsurgery heart failure, or postcardiotomy shock, is a frequent postoperative complication in children with CHD. Approximately 25% of children may be affected postoperatively (Wernovsky 1995).
LCOS leads to increased postoperative morbidity and mortality (Baysal 2010); a higher risk of cardiopulmonary resuscitation or extracorporeal life support (Delmo 2010); and prolonged mechanical ventilation and intensive care stay (Shi 2008). Prevention or early detection and treatment of LCOS is therefore important in order to improve patient outcomes. The diagnosis of LCOS is, however, not based on a uniform definition. While the pathophysiology consists of inadequate oxygen delivery to the organs due to cardiocirculatory insufficiency (Stocker 2006), invasive measurement of cardiac output is preferred in adults but is often not possible in children (Teng 2011). Otherwise, a combination of signs of inadequate peripheral oxygen delivery may be used to detect LCOS, for example tachycardia, low systolic blood pressure/high inotrope requirement (Shore 2001), or poor peripheral perfusion with an increased temperature difference between peripheral skin and body core.
Description of the intervention
Pharmacological LCOS treatment consists mainly of vasoactive drugs and of drugs intended to increase cardiac output, such as inotropes, inodilators, or inovasopressors. Examples include dobutamine, dopamine, epinephrine, norepinephrine (catecholamines), milrinone (phosphodiesterase III inhibitor), or levosimendan (calcium sensitiser).
Drugs from other classes, such as steroids, tri‐iodothyronine, nitric oxide, or sildenafil have also been studied as possible treatments for LCOS, but will not be part of this review; nor will pure vasodilators such as nitroprusside or nitroglycerin.
Inotropes act by increasing myocardial contractility, vasodilators relax vascular smooth muscle cells leading to dilation of blood vessels and a reduction of vascular resistance, and vasopressors increase afterload by constricting blood vessels by contraction of vascular smooth muscle cells. Inodilators or inovasopressors provide a combination of inotrope effects and either dilation or constriction of blood vessels, respectively.
Drugs used for treatment of LCOS may also be used prophylactically in order to prevent LCOS after cardiac surgery. Intravenous medications are primarily given by continuous infusion (or, with certain drugs, as a loading dose followed by continuous infusion). Drug administration is begun before, during, or immediately after separation from cardiopulmonary bypass, or within several hours, and can be continued for several days.
How the intervention might work
The context of the use of inotropic medications in postoperative LCOS involves different contributing factors to the condition: a) blood exposure to foreign surfaces of the cardiopulmonary bypass (CPB), which leads to systemic inflammation and capillary leak, oedema, myocardial systolic and diastolic dysfunction; b) myocardial reperfusion injury after cardioplegia and cardiac arrest; c) pulmonary reperfusion injury with impaired oxygen supply; d) severe increase of systemic vascular resistance after CPB, which cannot be overcome by the weakened myocardium. Therapeutic strategies to optimise cardiac output and minimise oxygen demand are therefore not restricted to drugs acting on the heart itself or to drugs as such. It is important to optimise mechanical ventilation and oxygen supply, volume load for the right and the left ventricles, and body temperature; and, equally important, to prevent dysrhythmia. In addition to all these measures, pharmacological treatment may be needed to increase myocardial contractility (inotropic medications), to decrease afterload (vasodilators without inotropic effect), or a combination of these mechanisms (inodilators).
Catecholamines such as dobutamine, dopamine, epinephrine, and norepinephrine act on dopamine receptors or on alpha or beta (or both) adrenoceptors. These are located in the myocardium and in blood vessels. Some of the substances are synthesised by the body physiologically, for example as stress hormones. The combination of its alpha and beta activity, which may vary based on drug dosing, determines the net effect of a given substance. In the myocardium, catecholamines have positive inotropic, chronotropic, and dromotropic effects and increase myocardial oxygen consumption. In the vasculature, they can have vasopressor or vasodilatory effects.
Phosphodiesterase type III inhibitors, for example amrinone, milrinone, olprinone, or enoximone (inodilators), reduce the degradation of cyclic adenosine monophosphate by the enzyme phosphodiesterase type III, thereby increasing phosphorylation of protein kinases that activate cardiac calcium channels, which has positive inotropic and lusitropic effects on the myocardium. In the vasculature, they act as dilators. In children undergoing congenital heart surgery, the prophylactic use of milrinone has so far not been sufficiently proven to prevent mortality or LCOS in clinical studies (Burkhardt 2015).
Calcium sensitisers (inodilatory effect) such as levosimendan bind to troponin C in the myocardium and increase its responsiveness to calcium, which increases inotropy. They also open adenosine‐triphosphate‐sensitive potassium channels in the smooth muscle cells of systemic, pulmonary, and coronary vessels, which leads to vasodilation (Turanlahti 2004). Levosimendan is a long‐acting drug, as its active metabolite OR‐1896 has a half‐life of approximately 80 hours. Due to low‐quality evidence in clinical studies, it is currently not clear whether levosimendan prevents mortality or LCOS in paediatric patients undergoing congenital heart surgery (Hummel 2017).
Why it is important to do this review
So far, there are no national or international guidelines on the safe and effective use of drugs for the prevention of LCOS in children undergoing congenital heart surgery. We have previously reviewed the use of milrinone and levosimendan for prevention of LCOS in this population, but did not find enough evidence for each drug separately among existing clinical trials (Burkhardt 2015; Hummel 2017). This is due to a small number of available studies and small numbers of paediatric patients, even in multicentre trials. New studies have been conducted or reported in the meantime, sometimes using both of these drugs in separate treatment arms. To synthesise all existing evidence on these two drugs and the potential of inotropic medications in general, a network meta‐analysis seems appropriate. This review is an essential step to provide further information on how to prevent LCOS and mortality in paediatric patients undergoing surgery for congenital heart disease.
Objectives
-
To compare in a network meta‐analysis the efficacy and safety of different prophylactic medications either against no treatment, or placebo, or against each other (as mono‐ or combination prophylaxis) for the prevention of low cardiac output syndrome and mortality for paediatric patients undergoing surgery for congenital heart disease.
-
To generate a clinically useful ranking of prophylactic medications (mono‐ and combination prophylaxis) according to efficacy and safety.
Methods
Criteria for considering studies for this review
Types of studies
We will include parallel‐arm randomised controlled trials (RCTs). We will not include cluster‐randomised trials, because studies in this area focus on the individual patient level; nor trials with cross‐over design, which would not be suitable for the prophylactic approach. We will include studies reported as full text, and those published as abstract only, asking the authors for further information in the latter case.
Types of participants
We will include paediatric patients from birth to 18 years of age who have undergone corrective or palliative heart surgery for congenital heart disease. We will first use pooled data for analysis and, in a second step, stratify the study population according to age and to circulatory physiology. This benefits the transitivity needed for network meta‐analysis, where all interventions are considered legitimate alternatives and therefore jointly randomisable.
There are no specific exclusion criteria for participants. If a study includes both eligible and ineligible participants, we will contact the study authors and ask them to provide data for the eligible patients only. If data for the eligible patient subset are not available, we will include a trial if at least 75% of patients fulfil our inclusion criteria and assess this inclusion with a sensitivity analysis.
Types of interventions
We will include studies that compare any pharmacological intervention(s) belonging to one of the following drug classes versus each other or versus no pharmacological intervention or placebo.
-
Catecholamines
-
Phosphodiesterase type III inhibitors
-
Calcium sensitisers
If combinations of drugs from classes 2 and 3 are used prophylactically, each combination will represent a separate intervention of interest. We will examine the geometry of the network meta‐analysis (NMA) for all interventions (mono‐ and combination prophylaxis) versus mono‐prophylaxis alone. In a case when we cannot analyse all combinations, we will split the network and examine the single networks separately. Combination treatments can be split into their components, and the effects of the components can be analysed separately, under certain assumptions even for disconnected networks (Rücker 2019). A preview of nodes is displayed in Figure 1. We are not planning to consider unspecified interventions for post hoc inclusion in the network within the context of jointly randomisable interventions. We expect co‐interventions, as explained above (How the intervention might work), but unless otherwise reported we will assume that all other measures besides inotropic medications are exhausted for patients in all study arms of the included trials.
We will define prophylactic administration as starting during surgery for congenital heart disease or up to four hours after disconnecting from cardiopulmonary bypass, and it shall be continued for at least four hours. The lower dosing limits for a medication to count as an intervention of interest shall be 0.2 μg/kg/min for milrinone, 0.05 μg/kg/min for levosimendan, 0.01 μg/kg/min for epinephrine and norepinephrine, and 5 μg/kg/min for dopamine and dobutamine.
Types of outcome measures
Unless otherwise stated, we will assess the outcomes at the longest available follow‐up.
Reporting one or more of the outcomes listed here in the trial is not an inclusion criterion for the review. Where a published report does not appear to report one of these outcomes, we will access the trial protocol and contact the trial authors to ascertain whether the outcomes were measured but not reported. We will include in the review, as part of the narrative, relevant trials which measured these outcomes but did not report the data at all, or not in a usable format.
Primary outcomes
We will calculate the relative effects of the competing interventions using the following primary outcomes.
-
All‐cause mortality within 30 days
-
Time to death (censored after three months)
-
Low cardiac output syndrome (at any time) defined as two or more of the following:
-
blood lactate > 3 mmol/l (27 mg/dl) or increase in blood lactate of at least 2 mmol/l (18 mg/dl) from baseline (Charpie 2000);
-
central venous oxygen saturation < 50% in biventricular physiology without shunts (Stocker 2006);
-
increase in arterial to central venous oxygen saturation difference by at least 20% from baseline prior to administration of the intervention of interest;
-
urine output < 0.5 ml/kg/h;
-
peripheral skin temperature to core body temperature difference of > 7 °C;
-
cardiac index as determined by Doppler echocardiography of < 2.2 l/min/m² (Rao 1996).
-
Secondary outcomes
-
Length of intensive care stay (days)
-
Length of hospital stay (days)
-
Duration of mechanical ventilation (hours)
-
Inotrope score (as defined in Wernovsky 1995, if available)
-
Number of patients requiring mechanical circulatory support (e.g. extracorporeal membrane oxygenation (ECMO), pulsatile assist devices)
-
Number of patients requiring cardiac transplantation
-
Number/proportion of patients experiencing adverse effects (adverse effects include: arrhythmia; hypotension defined as blood pressures below blood pressure appropriate for age or body surface area; headache; intraventricular haemorrhage; hypocalcaemia; hypokalaemia, bronchospasm; thrombocytopaenia defined as a platelet count < 50/nl or drop in platelet count of > 100% from baseline prior to administration of the intervention of interest; anaemia defined as a haemoglobin value below the age‐appropriate normal value; elevated serum levels of liver enzymes defined as serum enzymatic activities more than two‐fold the age‐appropriate normal values; left ventricular ejection fraction < 50% or left ventricular fraction of shortening < 28% as assessed by biplane or M‐mode echocardiography).
Regarding the number/proportion of adverse effects, we will analyse these separately as the individual components mentioned above. If there is insufficient information available, we will report the various adverse effects narratively, for example in a table.
Search methods for identification of studies
Electronic searches
We will identify trials through systematic searches of the following bibliographic databases.
-
Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library
-
MEDLINE (Ovid, from 1946 onwards)
-
Embase (Ovid, from 1980 onwards)
-
Conference Proceedings Citation Index ‐ Science (CPCI‐S) on the Web of Science
We will adapt the preliminary search strategy for MEDLINE (Ovid) (Appendix 2) for use in the other databases. We will apply the Cochrane precision and sensitivity‐maximising RCT filter to MEDLINE (Ovid) and adaptations of it to the other databases, except CENTRAL (Lefebvre 2011).
We will also conduct a search of www.ClinicalTrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) Search Portal (apps.who.int/trialsearch) for ongoing or unpublished trials.
We will also search regulatory data from EMA (www.ema.europa.eu/ema) and FDA (www.fda.gov/Drugs/InformationOnDrugs).
We will search all databases from their inception to the present, and we will impose no restriction on language of publication or publication status.
We will not perform a separate search for adverse effects of interventions used for the prevention of LCOS and mortality in paediatric patients undergoing surgery for congenital heart disease. We will consider adverse effects described in included studies only.
Searching other resources
We will check reference lists of all included studies and any relevant systematic reviews identified for additional references to trials. We will also examine any relevant retraction statements and errata for included studies. We will contact study authors for missing data and to obtain any available data from ongoing trials.
Data collection and analysis
Selection of studies
Two review authors (BB, JH) will independently screen titles and abstracts for inclusion of all the potential studies we identify as a result of the search and code them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. If there are any disagreements, we will ask a third author (BS) to arbitrate. We will retrieve the full‐text study reports/publication and two review authors (BB, JH) will independently screen the full text and identify studies for inclusion, and identify and record reasons for exclusion of the ineligible studies. We will resolve any disagreement through discussion or, if required, we will consult a third author (BS). We will identify and exclude duplicates and collate multiple reports of the same study so that each study rather than each report is the unit of interest in the review. We will record the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table (Liberati 2009).
Data extraction and management
We will use a data collection form, which we will have piloted on at least one study in the review, for study characteristics and outcome data. Two review authors (BB, JH) will extract study characteristics from included studies. We will extract the following study characteristics.
-
Methods: study design, total duration of study, number of study centres and location, study setting, and date of study.
-
Participants: N randomised, N lost to follow‐up/withdrawn, N analysed, mean age, age range, age subgroups (if available), gender, type and severity of CHD (including univentricular versus biventricular CHD), diagnostic criteria, inclusion criteria, and exclusion criteria.
-
Interventions: intervention, comparison, concomitant medications, and excluded medications.
-
Outcomes: primary and secondary outcomes specified and collected, and time points reported. If additional data are needed to derive hazard ratios, we will contact the individual study authors and ask them to provide additional data.
-
Notes: funding for trial, and notable conflicts of interest of trial authors.
Two review authors (BB, JH) will independently extract outcome data from included studies. We will resolve disagreements by consensus or by involving a third person (BS). One review author (BB) will transfer data into the Review Manager 5 (RevMan 5) file (Review Manager 2014). We will double‐check that data are entered correctly by comparing the data presented in the systematic review with the data extraction form. A second review author (JH) will spot‐check study characteristics for accuracy against the trial report.
Assessment of risk of bias in included studies
Two review authors (BB, JH) will independently assess risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We will resolve any disagreements by discussion or by involving another author (BS). We will assess the risk of bias according to the following domains.
-
Random sequence generation.
-
Allocation concealment.
-
Blinding of participants and personnel.
-
Blinding of outcome assessment.
-
Incomplete outcome data.
-
Selective outcome reporting.
-
Other bias.
We will grade each potential source of bias as high, low or unclear and provide a quote from the study report together with a justification for our judgment in the 'Risk of bias' table. We will summarise the risk of bias judgements across different studies for each of the domains listed. Where information on risk of bias relates to unpublished data or correspondence with a trialist, we will note this in the 'Risk of bias' table.
When considering treatment effects, we will take into account the risk of bias for the studies that contribute to that outcome.
Assessment of bias in conducting the systematic review
We will conduct the review according to this published protocol and report any deviations from it in the 'Differences between protocol and review' section of the systematic review.
Measures of treatment effect
We will assess dichotomous data (LCOS, 30‐day mortality) with relative risk as an effect measure. For time‐to‐event variables (time to death, duration of mechanical ventilation), if available, we will calculate hazard ratios. If only summary statistics are available, we will use methods described by Tierney 2007. For continuous variables, we will use the mean difference (MD) with 95% confidence intervals. For length of intensive care stay and length of hospital stay, which can assume a wide range of values, we will use a ratio of means (ROM). We will narratively describe skewed data reported as medians and interquartile ranges.
Unit of analysis issues
We will include RCTs with two or more parallel treatment arms. In case of three or more treatment arms, we will include these in the data set as a series of two‐arm comparisons, and will adjust the standard error of each of these comparisons for correlation between the arms. For this, we will use backcalculated standard errors in the weighted least‐square estimator to reflect the within‐study correlation (Rücker 2012; Rücker 2014).
Dealing with missing data
We will contact investigators or study sponsors in order to verify key study characteristics and obtain missing numerical outcome data where possible (e.g. when a study is identified as abstract only). Where possible, we will use the RevMan 5 calculator to calculate missing standard deviations using other data from the trial, such as confidence intervals, based on methods outlined in the Handbook (Higgins 2019). Where this is not possible, and the missing data are thought to introduce serious bias, we will explore the impact of excluding such studies in the overall assessment of results by a sensitivity analysis. We will perform analyses for efficacy outcomes, as far as possible, on an intention‐to‐treat (ITT) basis. In case of missing data, we will exclude all participants from the analysis for whom no outcome is available.
Assessment of heterogeneity
Pairwise comparisons
We will assess clinical and methodological heterogeneity within each pairwise comparison by descriptive statistics regarding study and population baseline characteristics. In case of excessive clinical heterogeneity, we will not pool the findings of the included studies. We will assess clinical heterogeneity (significantly different baseline characteristics) wherever these data are available for the study populations: a) patient age, b) severity of disease (RACHS/Aristotle/STS‐EACTS score). We will assess statistical heterogeneity by both Cochran’s Q test and the I² statistic. We will regard I² between 0% and 50% as a sign of low or moderate heterogeneity; and I² above 50% as representing substantial heterogeneity.
Network meta‐analysis
We will assess the assumption of transitivity across treatment comparisons by the distribution of effect modifiers in the different pairwise comparisons. We will assess patient age (according to the subgroups mentioned above) and circulatory physiology (univentricular versus biventricular) as possible effect modifiers. If this distribution is substantially imbalanced (intransitivity), we will not include those studies for indirect comparisons.
We will check consistency by statistically comparing direct and indirect evidence. We will also decompose existing heterogeneity/inconsistency into within‐design heterogeneity and inconsistency between designs.
We will investigate local incoherence using the function 'netsplit' from R package netmeta, which implements the SIDE (Separating Indirect from Direct Evidence) splitting approach. We will check global incoherence based on the full treatment‐design interaction model.
Assessment of reporting biases
If we are able to pool more than 10 trials, we will create and examine a funnel plot to explore possible small‐study biases for the primary outcomes or missing/unpublished studies leading to reporting bias (Higgins 2011). We will examine funnel plot asymmetry according to Chaimani 2013 with a comparison‐adjusted funnel plot applied to the network meta‐analysis.
Data synthesis
We will undertake meta‐analyses only where this is meaningful ‒ that is, if the treatments, participants and the underlying clinical question are similar enough for pooling to make sense.
Pairwise comparisons
We will perform standard pairwise meta‐analyses using a random‐effects model (inverse variance weighting) for each treatment comparison with at least two studies, because we expect methodological and clinical heterogeneity across the included studies resulting in varying effect sizes between studies of pairwise comparisons.
Network meta‐analysis
We will also perform random‐effects NMAs for the primary outcomes 'LCOS' and 'mortality', based on a frequentist framework in R as described by Rücker 2014. For analysis of drug combinations, we will use these as different nodes in the network like single‐drug interventions. We will separately analyse the data using component network meta‐analysis (CNMA) (Mills 2012; Rücker 2019; Welton 2009). The additive CNMA model assumes that the effects of combination interventions are the sum of their components and thus allows the disentanglement of the effects of single components. More general interaction models can also be applied, and the models can be compared using likelihood ratio tests. We will analyse ranking of treatments using P scores (Rücker 2015).
'Summary of findings' table
We will create a 'Summary of findings' table using the following outcomes: all‐cause mortality within 30 days; time to death (censored after three months); low cardiac output syndrome defined as described above; length of intensive care stay; length of hospital stay; duration of mechanical ventilation; and adverse events.
We will use the five GRADE considerations (study limitations; consistency of effect; imprecision; indirectness; and publication bias) to assess the quality of a body of evidence as it relates to the studies which contribute data to the meta‐analyses for the prespecified outcomes. Grading will be facilitated by the CINeMA (Confidence in Network Meta‐Analysis) framework (as in Huhn 2019). We will create 'Summary of findings' tables for all primary outcomes and their time points as mentioned above, according to Yepes‐Nuñez 2019. We will compare substance classes as above (catecholamines, phosphodiesterase type III inhibitors, calcium sensitisers) to each other and to placebo. We will make comments to aid readers' understanding of the review where necessary.
Two review authors (BB, JH), working independently, will make judgments about evidence quality and will resolve disagreements by discussion or by involving a third author (BS). We will justify, document and incorporate judgments into reporting of results for each outcome.
We plan to extract study data, format our comparisons in data tables and prepare 'Summary of findings' tables before writing the results and conclusions of our review. We plan that reporting will follow the PRISMA checklist for network meta‐analysis (Hutton 2016).
Subgroup analysis and investigation of heterogeneity
We plan to carry out analyses with the following subgroup stratifications.
-
Age groups: neonates < 1 month, infants 1 month to 1 year, and paediatric patients 1 year of age or older. In the case of insufficient patient numbers with three age subgroups that preclude drawing meaningful conclusions, subgroup analysis will involve neonates < 1 month and paediatric patients 1 month of age or older.
-
Cardiovascular physiology: patients with biventricular hearts, patients with univentricular hearts.
We will use the following outcomes in analysing the subgroups.
-
All‐cause mortality within 30 days
-
Time to death (censored after three months)
-
Low cardiac output syndrome defined as two or more of the following: a) blood lactate > 3 mmol/l (27 mg/dl) or increase in blood lactate of at least 2 mmol/l (18 mg/dl) from baseline prior to administration of the intervention of interest; b) central venous oxygen saturation < 50% in biventricular physiology without shunts; c) increase in arterial to central venous oxygen saturation difference by at least 20% from baseline prior to administration of the intervention of interest; d) urine output < 1 ml/kg/h; e) peripheral skin temperature to core body temperature difference of > 7 °C; (f) cardiac index as determined by Doppler echocardiography of < 2.2 l/min/m².
We will use the formal test for subgroup differences in the pairwise analysis in Review Manager 5 (Review Manager 2014), and base our interpretation on this.
Sensitivity analysis
The available studies may have different levels of risk of bias. We will perform a sensitivity analysis by re‐running the meta‐analysis after excluding studies with a high risk of bias (in any of the four domains 'random sequence', 'allocation concealment', 'incomplete outcome data' and 'selective reporting'). Where we find it impossible to calculate missing numerical data, and we think the missing data introduce serious bias, we will explore the impact of excluding such studies in the overall assessment of results by a sensitivity analysis. We will assess the effect of including any studies with a mixed population in sensitivity analysis. We will restrict sensitivity analyses to the primary outcomes.
Reaching conclusions
We will base our conclusions only on findings from the quantitative or narrative synthesis of included studies for this review. We will avoid making recommendations for practice, and our implications for research will suggest priorities for future research and outline what the remaining uncertainties are in the area.
