Prophylactic use of inotropic agents for the prevention of low cardiac output syndrome and mortality in adults undergoing cardiac surgery

Abstract

Background

As the burden of cardiovascular disease grows, so does the number of cardiac surgeries. Surgery is increasingly performed on older people with comorbidities who are at higher risk of developing perioperative complications such as low cardiac output state (LCOS). Surgery‐associated LCOS represents a serious pathology responsible for substantial morbidity and mortality. Prevention of LCOS is a critical and worthwhile aim to further improve the outcome and effectiveness of cardiac surgery. However, guidelines consistently report a lack of evidence for pharmacological LCOS prophylaxis.

Objectives

To assess the benefits and harms of the prophylactic use of any inotropic agent to prevent low cardiac output and associated morbidity and mortality in adults undergoing cardiac surgery.

Search methods

We identified trials (without language restrictions) via systematic searches of CENTRAL, MEDLINE, Embase, and CPCI‐S Web of Science in October 2022. We checked reference lists from primary studies and review articles for additional references. We also searched two registers of ongoing trials.

Selection criteria

We included randomised controlled trials (RCTs) enrolling adults who underwent cardiac surgery and were prophylactically treated with one or multiple inotropic agent(s) in comparison to any type of control (i.e. standard cardiac care, placebo, other inotropic agents).

Data collection and analysis

We used established methodological procedures according to Cochrane standards. Two review authors independently extracted data and assessed risk of bias according to a pre‐defined protocol. On request, we obtained a reply and additional information from only one of the included study authors. We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of evidence from the studies that contributed data to the meta‐analyses for the pre‐specified outcomes. Based on the identified studies, there were seven comparison groups: amrinone versus placebo, dopamine versus placebo, milrinone versus placebo, levosimendan versus dobutamine, levosimendan versus milrinone, levosimendan versus standard cardiac care, and levosimendan versus placebo.

Main results

We identified 29 eligible studies, including 3307 individuals, and four ongoing studies. In general, confidence in the results of the analysed studies was reduced due to relevant study limitations, imprecision, or inconsistency. Domains of concern encompassed inadequate methods of sequence generation and lack of blinding. The majority of trials were small, with only a few included participants, and investigated the prophylactic use of levosimendan.

Our meta‐analyses showed that levosimendan as compared to placebo may reduce the risk of LCOS (risk ratio (RR) 0.43, 95% confidence interval (CI) 0.25 to 0.74; I² = 66%; 1724 participants, 6 studies; GRADE: low) and probably reduces all‐cause mortality (RR 0.65, 95% CI 0.43 to 0.97; I² = 11%; 2347 participants, 14 studies; GRADE: moderate). This translates into a number needed to treat for an additional beneficial outcome (NNTB) of 8 to prevent one event of LCOS post surgery and of 44 to prevent one death at 30 days. Subgroup analyses revealed that the beneficial effects of levosimendan were predominantly observed in preoperative drug administration. Our meta‐analyses further indicated that levosimendan as compared to placebo may shorten the length of intensive care unit (ICU) stay (mean difference ‐1.00 days, 95% CI ‐1.63 to ‐0.37; 572 participants, 7 studies; GRADE: very low) and the duration of mechanical ventilation (mean difference ‐8.03 hours, 95% CI ‐13.17 to ‐2.90; 572 participants, 7 studies; GRADE: very low) but the evidence is very uncertain. The risk of adverse events did not clearly differ between levosimendan and placebo groups (cardiogenic shock: RR 0.65, 95% CI 0.40 to 1.05; I² = 0%; 1212 participants, 3 studies; GRADE: high; atrial fibrillation: RR 1.02, 95% CI 0.82 to 1.27; I² = 60%; 1934 participants, 11 studies; GRADE: very low; perioperative myocardial infarction: RR 0.89, 95% CI 0.61 to 1.31; I² = 13%; 1838 participants, 8 studies; GRADE: moderate; non‐embolic stroke or transient ischaemic attack: RR 0.89, 95% CI 0.58 to 1.38; I² = 0%; 1786 participants, 8 studies; GRADE: moderate). However, levosimendan as compared to placebo might reduce the number of participants requiring mechanical circulatory support (RR 0.47, 95% CI 0.24 to 0.91; I² = 74%; 1881 participants, 10 studies; GRADE: low).

There was no conclusive evidence on the effect of levosimendan compared to standard cardiac care on LCOS (RR 0.49, 95% CI 0.14 to 1.73; I² = 59%; 208 participants, 3 studies; GRADE: very low), all‐cause mortality (RR 0.37, 95% CI 0.13 to 1.04; I² = 0%; 208 participants, 3 studies; GRADE: low), adverse events (cardiogenic shock: RR 0.62, 95% CI 0.22 to 1.81; 128 participants, 1 study; GRADE: very low; atrial fibrillation: RR 0.40, 95% CI 0.11 to 1.41; I² = 60%; 188 participants, 2 studies; GRADE: very low; perioperative myocardial infarction: RR 0.62, 95% CI 0.22 to 1.81; 128 participants, 1 study; GRADE: very low; non‐embolic stroke or transient ischaemic attack: RR 0.56, 95% CI 0.27 to 1.18; 128 participants, 1 study; GRADE: very low), length of ICU stay (mean difference 0.33 days, 95% CI ‐1.16 to 1.83; 80 participants, 2 studies; GRADE: very low), the duration of mechanical ventilation (mean difference ‐3.40 hours, 95% CI ‐11.50 to 4.70; 128 participants, 1 study; GRADE: very low), and the number of participants requiring mechanical circulatory support (RR 0.88, 95% CI 0.50 to 1.55; I² = 0%; 208 participants, 3 studies; GRADE: low).

Authors' conclusions

Prophylactic treatment with levosimendan may reduce the incidence of LCOS and probably reduces associated mortality in adult patients undergoing cardiac surgery when compared to placebo only. Conclusions on the benefits and harms of other inotropic agents cannot be drawn due to limited study data. Given the limited evidence available, there is an unmet need for large‐scale, well‐designed randomised trials. Future studies of levosimendan ought to be designed to derive potential benefit in specific patient groups and surgery types, and the optimal administration protocol.

Plain language summary

Do heart surgery patients benefit from preventive treatment with inotropic drugs?

Key messages

• Treatment of heart surgery patients with the drug ‘levosimendan’ before surgery may be beneficial.

• However, the current evidence is limited, and more studies are needed before definitive recommendations can be made.

Rationale

Reduced heart function is a potentially fatal complication after heart surgery. A possible approach to prevent this complication is treatment with drugs that stimulate the contraction of the heart (so‐called inotropic agents).

What did we want to find out?

We assessed whether the use of inotropic agents before or during the surgical procedure prevented reduced heart function and death in adults after heart surgery.

What did we do?

We searched different medical literature databases and trial registers that collect information about planned, ongoing, and completed clinical studies. We considered studies in which one group had received an inotropic agent and a second group had received another inotropic drug or a sham medication or standard care. Two review authors independently screened and collected the data.

What did we find?

Study characteristics

We identified 29 studies that had a total of 3307 adult patients of both sexes who had heart surgery. The studies were conducted in different hospitals in Europe, Israel, Japan, Korea, Turkey, Canada, and the USA. Nine studies were funded by the manufacturer of the investigated drug. In 10 studies, the relationship to the pharmaceutical industry was not determined. When we asked all of the study authors for additional information about their studies, only one author responded.

The majority of the studies (24 in total) dealt with the inotropic drug levosimendan. So, the available data did not allow us to judge agents other than levosimendan. The participants were given the drug before, during, or immediately after heart surgery. They were monitored for up to 30 days.

Main results

Patients treated with levosimendan prior to heart surgery possibly have a lower risk of reduced heart function and death and may spend less time in intensive care compared to patients receiving a sham medication. However, the available data revealed no clear difference between levosimendan and standard care or treatment with another inotropic drug in preventing reduced heart function and death and reducing time spent in intensive care after heart surgery. The available data also showed no clear differences in the prevention of adverse events between levosimendan and comparator treatment.

What are the limitations of the evidence?

We have little confidence in the evidence, mainly because of the small number of included participants (the results are very imprecise). Thus, most results of the review must be viewed with caution.

How up‐to‐date is this evidence?

This evidence is up‐to‐date to October 2022.

Summary of findings

Summary of findings 1. Levosimendan compared to standard cardiac care.

Patients/population: adults (> 18 years) undergoing cardiac surgery Intervention: levosimendan (started for preventive reasons) Comparison: standard cardiac care Setting of interest: hospital‐based randomised controlled trials conducted in Turkey or India
Outcomes	Anticipated absolute effects (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)
	Risk with standard cardiac care	Risk with levosimendan
All‐cause mortality (number of deaths within 30 days)	115 per 1000a	42 per 1000 (15 to 120)	RR 0.37 (0.13 to 1.04)	208 (3 studies)	lowc
Incidence of LCOS (number of events within 48 hours post surgery)	192 per 1000a	94 per 1000 (27 to 332)	RR 0.49 (0.14 to 1.73)	208 (3 studies)	very lowd
Adverse events: cardiogenic shock (number of events within hospital stay)	125 per 1000b	79 per 1000 (27 to 226)	RR 0.62 (0.22 to 1.81)	128 (1 study)	very lowe
Length of ICU stay (in days within ICU care)	The mean length of ICU stay across control groups was from 1.40 to 3.67 days	The mean length of ICU stay in the intervention group was 0.33 higher (1.16 lower to 1.83 higher)	‐	80 (2 studies)	very lowd
Duration of mechanical ventilation (in hours within ICU care)	The mean duration of mechanical ventilation in the control group was 83.2 hours	The mean duration of mechanical ventilation in the intervention group was 3.40 hours lower (11.50 lower to 4.70 higher)	‐	128 (1 study)	very lowe
Number of participants requiring mechanical circulatory support (number of events within hospital stay)	192 per 1000a	169 per 1000 (96 to 298)	RR 0.88 (0.50 to 1.55)	208 (3 studies)	lowc
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; LCOS: low cardiac output syndrome; ICU: intensive care unit; RR: risk ratio
GRADE working group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^aControl group risk estimate comes from the control group risk in included studies.
^bControl group risk estimate comes from the control group risk in a small included study.
^cDowngraded 1 level for risk of bias due to inadequate methods of sequence generation and lack of blinding, and 1 level for imprecision due to the optimal information size criterion not being met.
^dDowngraded 1 level for risk of bias due to inadequate methods of sequence generation and lack of blinding, 1 level for imprecision due to the optimal information size criterion not being met, and 1 level for inconsistency due to heterogeneity according to the I² statistic.
^eDowngraded 1 level for risk of bias due to inadequate methods of sequence generation and lack of blinding, and 2 levels for imprecision due to the optimal information size criterion not being met (data based on a single small study).

Summary of findings 2. Levosimendan compared to placebo.

Patients/population: adults (> 18 years) undergoing cardiac surgery Intervention: levosimendan (started for preventive reasons) Comparison: placebo Setting of interest: hospital‐based randomised controlled trials conducted in Europe, China, or India
Outcomes	Anticipated absolute effects (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)
	Risk with placebo	Risk with levosimendan
All‐cause mortality (number of deaths within 30 days)	66 per 1000a	43 per 1000 (28 to 64)	RR 0.65 (0.43 to 0.97)	2347 (14 studies)	moderateb
Incidence of LCOS (number of events within 48 hours post surgery)	238 per 1000a	102 per 1000 (60 to 176)	RR 0.43 (0.25 to 0.74)	1724 (6 studies)	lowc
Adverse events: cardiogenic shock (number of events within hospital stay)	63 per 1000a	41 per 1000 (25 to 66)	RR 0.65 (0.40 to 1.05)	1212 (3 studies)	high
Length of ICU stay (in days within ICU care)	The mean length of ICU stay across control groups was from 1.07 to 8.15 days	The mean length of ICU stay in the intervention group was 1.00 lower (1.63 lower to 0.37 lower)	‐	572 (7 studies)	very lowd
Duration of mechanical ventilation (in hours within ICU care)	The mean duration of mechanical ventilation across control groups was between 5.67 and 157.44 hours	The mean duration of mechanical ventilation in the intervention group was 8.03 lower (13.17 lower to 2.90 lower)	‐	572 (7 studies)	very lowd
Number of participants requiring mechanical circulatory support (number of events within hospital stay)	145 per 1000a	68 per 1000 (35 to 132)	RR 0.47 (0.24 to 0.91)	1881 (10 studies)	lowc
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; LCOS: low cardiac output syndrome; ICU: intensive care unit; RR: risk ratio
GRADE working group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^aControl group risk estimate comes from the control group risk in included studies.
^bDowngraded 1 level for risk of bias due to inadequate methods of sequence generation and lack of blinding.
^cDowngraded 1 level for risk of bias due to inadequate methods of sequence generation and 1 level for inconsistency due to heterogeneity according to the I² statistic.
^dDowngraded 1 level for risk of bias due to inadequate methods of sequence generation and lack of blinding, and 2 levels for inconsistency due to heterogeneity according to the I² statistic.

Background

Description of the condition

The worldwide increasing number of individuals with cardiovascular disease results in a steadily rising number of patients undergoing heart surgery. Globally, more than 1.5 million individuals undergo heart surgery in about 4000 centres annually; the vast majority of surgeries are performed in high‐income countries (Vervoort 2020).

The state of low cardiac output (LCOS) represents a potential complication of heart surgery, and is associated with a marked increase in morbidity and mortality. It is characterised mainly by a spontaneous drop of systolic cardiac function and a reduction of circulatory capacity, resulting in cardio‐circulatory insufficiency; perfusion of the microcirculation and oxygen supply of end organ tissue cannot be maintained. However, there is no standardised definition of LCOS. For the purpose of this review, we will consider the most widely used definition of cardio‐circulatory insufficiency (cardiac index (CI) < 2.2 L/min/m²; mean arterial pressure (MAP) < 60 mmHg; mixed venous oxygen saturation (SvO2) < 60 %) despite optimal therapy including euvolaemia within two hours after completion of the index procedure (Pérez 2012). This haemodynamic state must be associated with clinical signs and symptoms of malperfusion, or end‐organ hypoperfusion (e.g. cumulating lactate levels > 2.0 mmol/L; cool extremities, oliguria or anuria, with a diuresis < 0.5 mL/kg/h; Pérez 2012). The state of LCOS can further destabilise into cardiogenic shock (CS), when systolic blood pressure (SBP) falls below 90 mmHg, or catecholamines are required for a duration of at least 15 minutes, or both (Pérez 2012; van Diepen 2017). Following the adverse spiral of shock, subsequent peripheral malperfusion can culminate in multi‐organ dysfunction (van Diepen 2017). LCOS related to cardiac surgery can have a variety of causes. Both worsening of pre‐existing heart disease and the adverse impact of cardiac surgery may contribute to the development of LCOS (Carl 2010; Habicher 2018).

Description of the intervention

To maintain a physiological haemodynamic state during cardiac surgery can be challenging, and frequently requires pharmacological interventions or mechanical circulatory support. This might be even more challenging in light of the growing population of the elderly and people with multiple comorbidities who are currently undergoing heart surgery. The drugs predominantly used for haemodynamic stabilisation in LCOS are inotropes. Inotropes restore adequate perfusion by increasing systolic cardiac function.

More traditional inotropes, such as catecholamines (e.g. dopamine, epinephrine, norepinephrine, dobutamine, dopexamine) and phosphodiesterase (PDE)‐III‐inhibitors (e.g. milrinone, enoximone, amrinone), act by increasing the calcium influx of cardiomyocytes in different ways, thereby enhancing contractility (Habicher 2018). To some degree, this comes with a risk of arrhythmias (Carl 2010; Habicher 2018).

As an established alternative, myocardial contractility can also be increased by triggering myocardial coupling to a given calcium concentration. This can be addressed pharmacologically with calcium sensitisers (Carl 2010; Habicher 2018). Out of the drug class of calcium sensitisers, the broadest body of evidence exists for levosimendan. Levosimendan acts by enhancing the responsiveness of myofilaments to calcium and, thereby, increasing cardiac output without elevating myocardial oxygen consumption (Habicher 2018; Pathak 2013). Therefore, it is considered to be fairly safe and effective in the prevention or treatment of surgery‐related LCOS (Carl 2010; Habicher 2018). Similarly to other inotropes, levosimendan is administered intravenously, with an option to start with a bolus prior to continuous infusion. For prophylactic use, levosimendan is applied before, during, or after the person is weaned from cardiopulmonary bypass. Since its active metabolite, OR‐1896, has a half‐life of 70 to 80 hours, it is thought that levosimendan provides prolonged haemodynamic benefit (Pathak 2013). Levosimendan is a widely studied calcium sensitiser. In addition to many small‐scale studies, several randomised controlled trials (RCTs) and meta‐analyses generating heterogenous data have recently been published (Chen 2017; Chen 2018; Cholley 2017; Conte 2019; Desai 2018; Elbadawi 2018; Landoni 2017; Lee 2017; Mehta 2017; Ng 2019; Putzu 2018; Qiang 2018; Sanfilippo 2017; Schumann 2018; Tena 2018; van Diepen 2020; Wang 2018; Wang 2019; Wang 2019a; Weber 2020; Yan 2020; Zhou 2018; Zhu 2019).

How the intervention might work

The outcome of people undergoing cardiac surgery is critically dependent on minimising or avoiding functional or morphological cell damage of the myocardium which, in turn, would alter contractility. People are even more at risk for a negative haemodynamic cascade if left ventricular function was impaired beforehand (Carl 2010; Habicher 2018). The main treatment strategies in LCOS aim to prevent or to improve impaired macro‐ and microcirculatory perfusion, in order to maintain oxygen supply at the cellular level and to modulate local and systemic inflammatory responses (van Diepen 2017). These mechanistic leverages are intended to prevent or minimise secondary multi‐organ dysfunction, or failure.

Classic inotropic agents (catecholamines, PDE‐III‐inhibitors) increase cardiac contractility and reduce systemic vascular resistance (SVR), thereby attenuating left ventricular afterload via activation of the cAMP pathway (Habicher 2018). The resulting increase in myocardial performance and haemodynamic improvement is accompanied by intensified myocardial oxygen consumption, a relevant consequence of inotropic support (Carl 2010; Habicher 2018). Therefore, if clinically required, the use of inotropes should be restricted to a limited duration.

Calcium sensitisers have been extensively studied in an attempt to prevent LCOS. The effect of calcium sensitisers, such as levosimendan, is based on a dual mechanism. First, the pharmacological agent binds to troponin C, increasing the stability of the tropomyosin molecule (Habicher 2018; Pathak 2013). This results in enhanced myocardial contractility. The presence of calcium is a prerequisite for the binding of levosimendan to troponin C. As levosimendan does not negatively affect unbound intracellular calcium levels, it does not influence the relaxation of the myocardium (Habicher 2018; Pathak 2013). Second, levosimendan causes relaxation of smooth muscles by opening ATP‐sensitive potassium channels in the vasculature (Habicher 2018; Pathak 2013). This facilitates coronary blood flow and myocardial perfusion. Similarly, levosimendan elicits vasodilatation of peripheral arteries which, in turn, reduces left ventricular afterload (Habicher 2018; Pathak 2013). So, levosimendan exerts both inotropic and vasodilatory effects.

This review will focus on all inotropic agents that are being used to prophylactically minimise the incidence of LCOS in adults undergoing cardiac surgery.

Why it is important to do this review

As the burden of cardiovascular disease grows in western society, so does the number of cardiac surgeries. Due to medical progress and refinements of care at all healthcare levels, i.e. perioperative myocardial protection, specialised cardio‐anaesthesia, haemodynamic monitoring, surgical techniques, intraoperative perfusion, and postoperative management, the morbidity and mortality of people undergoing cardiovascular surgery have decreased overall. Against this background, surgical procedures are performed in increasingly elderly and comorbid people, who are already at a higher risk of developing perioperative complications, such as LCOS. LCOS represents a serious pathology that accounts for substantial morbidity and mortality following cardiac surgery. Beyond its prognostic impact, LCOS can increase the length of stay in the intensive care unit (ICU) and in the hospital overall. As a result of a complicated postoperative course, LCOS increases the use of medical resources (e.g. ventilators, extracorporeal membrane oxygenation (ECMO) circuits, dialysis) and ultimately increases medical costs. Hence, the prevention of LCOS is a critical and worthwhile aim to further improve the outcome and effectiveness of cardiac surgery.

Notwithstanding the relevance described above, research on the prophylactic use of inotropic agents in cardiac surgery has not resulted in high‐quality evidence in support of the concept (Elbadawi 2018; Ng 2019; Wang 2019a). Based on heterogenous original data, national and international guidelines consistently state a lack of evidence for pharmacological treatment of LCOS, and do not recommend it (Cholley 2017; Desai 2018; Mehta 2017; Wang 2019). This is especially true for its prophylactic use (Carl 2010; Habicher 2018; Ponikowski 2016; Pérez 2012; van Diepen 2017). It remains uncertain whether people with specific disorders do better, or if certain surgical methods, drugs or administration protocols provide better outcomes.

In light of the growing number of people undergoing heart surgery, and the critical impact on clinical practice, this review will assess the efficacy and safety of a precautionary administration of inotropes, by analysing available randomised controlled trials. This review aims to generate a basis for evidence‐based decision‐making in future clinical practice, improving perioperative morbidity and mortality. The evaluation of current evidence and the identification of persistent knowledge gaps may contribute to the conception and design of future clinical trials.

Objectives

To assess the benefits and harms of the prophylactic use of any inotropic agent to prevent low cardiac output and associated morbidity and mortality in adults undergoing cardiac surgery.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) of parallel or multi‐arm design with randomisation at the level of the participant, regardless of the study publication status. We excluded studies that aimed to evaluate interventions but did not use randomisation for participant allocation to different study arms (quasi‐randomised trials), due to concerns regarding internal validity, as well as cross‐over trials, due to the investigation of all‐cause mortality as the primary outcome. We excluded RCTs whose unit of allocation was not the individual participant (cluster‐RCTs), due to our interest in the individual effect of an intervention.

Types of participants

We included RCTs that investigated the prophylactic administration of inotropic agents, enrolling adults (≥ 18 years of age) undergoing cardiac surgery, without signs of low cardiac output (LCOS) at the time of inclusion, with a follow‐up period spanning at least the hospitalisation phase. We considered any type of cardiac surgery, with the exception of complex heart defects, which require several operations in succession. We considered procedures with and without cardiopulmonary bypass (CPB). Participants were included regardless of clinical heart failure state or pre‐existing left ventricular (LV) dysfunction. We assessed the aspects of 'type of cardiac surgery' and 'pre‐existing LV dysfunction' by subgroup analyses. Studies including only a subset of relevant participants were also considered.

Types of interventions

We included trials investigating the use of inotropic agents (i.e. dopamine, epinephrine, norepinephrine, dobutamine, dopexamine, milrinone, enoximone, amrinone, levosimendan), either alone or in combination, started for preventive reasons before, during, or after the surgical intervention. There were no restrictions in terms of delivery or dosage. We considered the comparison of (i) inotropic agent(s) to treatments without specific experimental single drugs (standard cardiac care), or (ii) inotropic agent(s) to placebo or (iii) inotropic agents(s) to other inotropic agent(s).

Types of outcome measures

Reporting in the trial of one or more of the outcomes listed here was not an inclusion criterion for the review. When a published report did not appear to report one of these outcomes, we accessed the trial protocol and contacted the trial authors to ascertain whether the outcomes were measured but not reported. We used quantitative data provided by the study authors. We included relevant trials, which measured these outcomes but did not report the data at all or not in a usable format, as part of the narrative in the review.

Given the clinical relevance of LCOS during the acute perioperative and immediate postoperative period, we focused on the short‐term course (intensive care unit (ICU) and hospital stay). We also examined data at the longest follow‐up given in each study.

Primary outcomes

1. All‐cause mortality (within 30 days)

2. Incidence of LCOS, as defined in the background section (perioperatively until 48 hours after the completion of the procedure)

3. Adverse events (number or proportion of participants)

   1. Cardiogenic shock (mean arterial pressure (MAP) < 60 mmHg or requirement of catecholamines for more than 15 minutes; cardiac index (CI) < 2.2 L/min/m²; pulmonary capillary wedge pressure (PCWP) > 18 mmHg; lactate level > 2.0 mmol/L; venous oxygen saturation (SvO2) < 60%; clinical signs of malperfusion)

   2. Arrhythmia (atrial fibrillation, supraventricular tachycardia, ventricular arrhythmia of new onset)

   3. Perioperative myocardial infarction (increasing troponin and creatine kinase‐MB (CK‐MB) levels of more than 50% relative to baseline; electrocardiogram (ECG) changes, or regional wall motion abnormalities, or angiographic changes, or a combination)

   4. Non‐embolic stroke or transient ischaemic attack

Secondary outcomes

1. Length of hospital stay (days)

2. Length of intensive care unit (ICU) stay (days)

3. CI (L/min/m²) in ICU

4. MAP pressure (mmHg) in ICU

5. PCWP (mmHg) in ICU

6. Duration of mechanical ventilation (hours)

7. Proportion of weaning failure within 48 hours of extubation (objective (e.g. tachypnoea, hypertension, hypoxaemia, acidosis) or subjective (e.g. agitation, distress, diaphoresis, increased respiratory efforts), or both), signs of respiratory insufficiency; unsuccessful permanent discontinuation of mechanical ventilation

8. Number of participants requiring mechanical circulatory support

9. Number of participants requiring additional inotropic drugs

10. Number of participants requiring cardiac transplantation or ventricular assist device implantation of any bridging strategy

11. Number of participants requiring cardiopulmonary resuscitation

12. Proportion of renal failure (increase of serum creatinine of at least 0.3 mg/dL, oliguria or anuria with diuresis < 0.5 mL/kg/h for at least six hours, or requirement of renal replacement therapy, or a combination)

We included any data on participants' quality of life or medical cost in the analysed studies as part of the narrative in the review.

Search methods for identification of studies

Electronic searches

We identified trials through systematic searches of the following bibliographic databases on 17 October 2022:

• Cochrane Central Register of Controlled Trials (CENTRAL; 2022, Issue 10) in the Cochrane Library;

• MEDLINE All (Ovid, 1946 to 14 October 2022);

• Embase (Ovid, 1980 to 2022 week 41);

• Conference Proceedings Citation Index ‐ Science (CPCI‐S) on the Web of Science (Clarivate Analytics, 1990 to 17 October 2022).

We adapted the search strategy for MEDLINE Ovid for use in the other databases (Appendix 1). We applied the Cochrane sensitivity and precision‐maximising RCT filter to MEDLINE Ovid, and adapted it to the other databases, except CENTRAL (Lefebvre 2019).

We also conducted a search of ClinicalTrials.gov (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 on 17 October 2022.

We searched all databases from their inception to the present, with no restriction on language or status of publication.

Searching other resources

We checked the reference lists of all included studies and relevant systematic reviews identified for additional references of interest. We also examined any relevant retraction statements and errata for included studies.

Data collection and analysis

Selection of studies

Two review authors (DG and JS) independently screened titles and abstracts of all the potential studies we identified in our search, and coded them as 'retrieve' (eligible, potentially eligible, or unclear) or 'do not retrieve'. Disagreements were resolved by consensus. We retrieved the full‐text study reports or publication, and two review authors (DG and JS) independently screened the full text, identified studies for inclusion, and identified and recorded reasons for excluding the ineligible studies. Any disagreement was resolved by discussion. We identified and excluded duplicates and collated multiple reports of the same study, so that each study was the unit of interest in the review process. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Figure 1) and 'Characteristics of excluded studies' table (Liberati 2009).

1.

PRISMA study flow diagram

Data extraction and management

We used a data collection form for study characteristics and outcome data, which had been piloted on at least one study in the review process. Two review authors (DG and JS) extracted study characteristics from the included studies. We extracted the following study characteristics.

1. Methods: study design, total duration of enrolment and follow‐up, number and location of study centres, study setting, and period of study.

2. Participants: N randomised, N lost to follow‐up or withdrawn, N analysed, mean age, age range, gender, participant with or without pre‐existing LV dysfunction, type of cardiac surgery, inclusion criteria, and exclusion criteria.

3. Interventions: intervention, comparison, concomitant medications, excluded medications, timing of the administration, and in the case of levosimendan, the administration protocol (with or without bolus).

4. Outcomes: primary and secondary outcomes specified and collected, and time points reported.

5. Notes: funding for the trial and notable conflicts of interest of trial authors.

Two review authors (DG and JS) independently extracted outcome data from the included studies. Disagreements were resolved by consensus. One review author (JS) transferred data into the Review Manager 5 file (Review Manager 2020). We double‐checked whether data were entered correctly by comparing the data presented in the review with the data extraction form. A second review author (DG) spot‐checked study characteristics for accuracy in comparison to the trial reports.

No transformations of reported data were made before presentation in the review article and no numerical data were extracted from graphs.

Assessment of risk of bias in included studies

Two review authors (DG and JS) independently assessed risk of bias for each study using Cochrane RoB 1, as outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Disagreements were resolved by consensus. We assessed the risk of bias in trial results according to the following domains:

1. random sequence generation;

2. allocation concealment;

3. blinding of participants, personnel, and outcome assessment;

4. incomplete outcome data;

5. selective reporting;

6. other sources of bias (cross‐over design, baseline differences, conduct of the study affected by interim results, deviation from the study protocol not reflecting clinical practice, inappropriate administration of an intervention, contra‐active or similar pre‐randomisation intervention).

We assessed the risk of bias for the outcomes of the included studies that are included in our summary of findings tables. The effect of interest was that of assignment to intervention, and assessment for risk of bias was the result corresponding to a full intention‐to‐treat (ITT) analysis.

We rated each domain as low, unclear, or high risk of bias. We summarised the risk of bias judgements across different studies for each of the domains listed, and for each outcome. The overall risk of bias is the least favourable assessment across the domains of bias.

When considering treatment effects, we took the risk of bias for the contributing outcome data into account.

Measures of treatment effect

We analysed dichotomous data as risk ratios (RR) with 95% confidence intervals (CI), and continuous data as mean difference (MD) with 95% CIs. We entered data presented as a scale with a consistent direction of effect.

We narratively described skewed data reported as medians and interquartile ranges.

Unit of analysis issues

We included RCTs with a parallel design, as well as multi‐arm studies. For outcomes measured at multiple time points, we performed separate analyses, with time frames that reflected short‐term and longest follow‐up. In the case where a study contributed multiple, correlated comparisons, we combined groups to create a single pair‐wise comparison as outlined in the CochraneHandbook (Higgins 2019). We analysed groups in an appropriate way, which avoided arbitrary omission or double counting of participants.

Dealing with missing data

We contacted investigators or study sponsors to verify key study characteristics, and obtain missing numerical outcome data, when possible (e.g. when a study was identified as an abstract only). Wherever possible, we used the Review Manager 5 calculator to calculate missing standard deviations (SD), using other data from the trial, such as confidence intervals, based on methods outlined in the CochraneHandbook (Higgins 2019). Whenever this was not possible, and we thought missing data may introduce serious bias, we explored the impact of including such studies in the overall assessment of results with a sensitivity analysis.

Assessment of heterogeneity

We inspected forest plots visually to consider the direction and magnitude of effects and the degree of overlap between confidence intervals. We used the I² statistic to measure heterogeneity among the trials in each analysis, but acknowledge that there is substantial uncertainty in the value of I² if there are only a small number of studies available. We also considered the P value from the Chi² test. If we identified substantial heterogeneity (I² ≥ 50%), or we suspected heterogeneity from the confidence intervals in the forest plots, we reported it, and explored possible causes by prespecified subgroup analyses.

Assessment of reporting biases

If we were able to pool more than 10 trials, we created and examined a funnel plot, to explore possible small study biases for the primary outcomes (Sterne 2011). To assess asymmetry, we performed the Habord test, as implemented in the statistical software Stata 15 (Stata).

Data synthesis

We performed a meta‐analysis only where it was meaningful, i.e. if treatments, participants, and the underlying clinical question were similar enough for pooling to make sense.

In the primary analysis, we included all studies. We assessed the potential effects of studies at high risk, or unclear risk of bias in sensitivity analyses.

We used a random‐effects model, because we expected clinical heterogeneity to arise from differences in study characteristics, and the associated assumption that the effects being estimated in the different studies are not identical, but follow a similar distribution.

Subgroup analysis and investigation of heterogeneity

We carried out the following subgroup analyses for any outcomes with substantial heterogeneity.

1. Regimen of the control group (placebo‐controlled versus active control).

2. Timing of the administration (preoperative versus perioperative versus postoperative).

3. Administration protocol for levosimendan (with versus without bolus).

4. Type of cardiac surgery (single versus combined structural procedures, procedures with versus without bypass).

5. Participants with versus without pre‐existing LV dysfunction, i.e. heart failure with preserved ejection fraction (HFpEF; LVEF > 50%), heart failure with mid‐range ejection fraction (HFmEF; LVEF 41% to 49%), and heart failure with reduced ejection fraction (HFrEF; LVEF < 40%).

We used the formal test for subgroup differences in Review Manager 5, and based our interpretation on this.

Sensitivity analysis

Depending on the number of included studies, we planned to carry out the following sensitivity analyses, to test whether key methodological factors or decisions had affected the main result.

• Primary analyses including all studies and sensitivity analysis including only studies judged to be at an overall low risk of bias.

• Primary analysis using the random‐effects model, due to expected heterogeneity between studies; we also explored a fixed‐effect model if the I² statistic value was < 50%.

Summary of findings and assessment of the certainty of the evidence

We created summary of findings tables using the following outcomes:

• All‐cause mortality

• Incidence of LCOS

• Adverse events

• Length of ICU stay

• Duration of mechanical ventilation

• Number of participants requiring mechanical circulatory support

In clinical practice, the debate on prophylactic medication mainly relates to the use of the calcium sensitiser levosimendan. Most RCTs on this topic therefore focus on the efficacy and safety of this specific drug. For this reason, the summary of findings tables were restricted to the comparisons of levosimendan with placebo or standard care.

Two review authors (DG and JS) independently judged the certainty of the evidence, with disagreements resolved by discussion. We justified judgements by documenting decisions and incorporating them into the results for each outcome. We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of the body of evidence as it relates to the studies that contribute data to the meta‐analyses for the prespecified outcomes. We used the overall risk of bias judgement for the GRADE assessment, as described in Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2019). We justified all decisions to downgrade the certainty of studies using footnotes, and we made comments to aid the reader's understanding of the review, where necessary.

Results

Description of studies

Randomised controlled trials (RCTs) in individuals undergoing cardiac surgery and treated prophylactically with inotropic drugs.

Results of the search

After removing duplicates, we identified a total of 6416 references. By screening the titles and abstracts, we ascertained 49 references of potential relevance, whose full texts we assessed by applying our inclusion and exclusion criteria. This resulted in the inclusion of 29 studies (see Characteristics of included studies). Furthermore, we identified four ongoing studies (see Characteristics of ongoing studies). Details of the remaining studies can be found in Characteristics of excluded studies. The overall process is recorded in the PRISMA flowchart (Figure 1).

Included studies

The inclusion criteria were met by 29 randomised controlled studies. Each study characteristic is presented briefly in the table Characteristics of included studies. The qualitative and quantitative analysis of the included studies is based on published information. Contacting the authors of the included studies resulted in feedback and additional information only from the corresponding author of the study Baysal 2014, which was included in the analysis.

Study size and location

Twenty‐four studies were conducted as single‐centre trials in Belgium (De Hert 2007), China (Hu 2020), Denmark (Juhl‐Olsen 2015), Egypt (Amin 2019), Finland (Jävelä 2008; Lahtinen 2011; Leppikangas 2011), Germany (Erb 2014), Greece (Anastasiadis 2016), India (Desai 2018; Gandham 2013; Kandasamy 2017; Kodalli 2013; Mishra 2016; Shah 2014; Sharma 2014), Iran (Hadadzadeh 2013), Israel (Gatot 2004), Italy (Tritapepe 2006; Tritapepe 2009), Japan (Kikura 2002), Korea (Jo 2010), or Turkey (Baysal 2014; Ersoy 2013). In these studies, the number of enrolled study participants varied from 20 to 207.

Five studies were conducted as multicentre trials in Canada (125 participants enrolled; Denault 2016), Finland (60 participants enrolled; Eriksson 2009), France (336 participants enrolled; Cholley 2017), the USA (252 participants enrolled; Levin 2012), or the USA plus Canada (882 participants enrolled; Mehta 2017).

Examined surgical interventions

The types of surgery assessed included coronary artery bypass graft (CABG), off‐pump coronary artery bypass graft (OPCABG), and valve surgery. In one study, there was no detailed description of the cardiac surgery performed (De Hert 2007). Only CABG (Anastasiadis 2016; Eriksson 2009; Gatot 2004; Kikura 2002; Levin 2012; Tritapepe 2006; Tritapepe 2009) and OPCABG (Amin 2019; Desai 2018; Hadadzadeh 2013; Jo 2010; Kandasamy 2017; Kodalli 2013; Shah 2014) participants were included in seven studies each; five trials were limited to participants with valve surgery (Baysal 2014; Ersoy 2013; Gandham 2013; Juhl‐Olsen 2015; Mishra 2016). In one study, both patients with CABG and patients with valve surgery were included (Hu 2020). Also, eight studies included participants with combined procedures, i.e. CABG with valve surgery or valve surgery with CABG (Cholley 2017; Denault 2016; Erb 2014; Jävelä 2008; Lahtinen 2011; Leppikangas 2011; Mehta 2017; Sharma 2014).

Emergency surgery was an exclusion criterion in 13 studies (Anastasiadis 2016; Baysal 2014; Cholley 2017; Denault 2016; Desai 2018; Hadadzadeh 2013; Jo 2010; Kandasamy 2017; Kikura 2002; Levin 2012; Mishra 2016; Shah 2014; Sharma 2014). A total of 12 studies excluded participants with a history of heart surgery (Anastasiadis 2016; Baysal 2014; Desai 2018; Eriksson 2009; Gandham 2013; Gatot 2004; Kandasamy 2017; Kodalli 2013; Shah 2014; Sharma 2014; Tritapepe 2006; Tritapepe 2009).

Examined prophylactic medication

The majority of trials investigated the prophylactic usage of levosimendan compared to placebo (Anastasiadis 2016; Cholley 2017; Erb 2014; Eriksson 2009; Hu 2020; Jävelä 2008; Juhl‐Olsen 2015; Kodalli 2013; Lahtinen 2011; Leppikangas 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014; Tritapepe 2006; Tritapepe 2009), standard cardiac care (Baysal 2014; Desai 2018; Ersoy 2013), or another inotropic agent (either milrinone (Amin 2019; De Hert 2007; Mishra 2016) or dobutamine (Gandham 2013; Kandasamy 2017)). Altogether, 2290 participants were enrolled in the studies on levosimendan compared to placebo, 208 participants in the studies on levosimendan compared to standard cardiac care, 100 participants in the studies on levosimendan compared to milrinone, and 140 participants in the studies on levosimendan compared to dobutamine. There were only four trials investigating milrinone versus placebo (Denault 2016; Hadadzadeh 2013; Jo 2010; Kikura 2002), one trial investigating dopamine versus placebo (Gatot 2004), and one trial investigating amrinone versus milrinone or placebo (Kikura 2002). The number of enrolled participants was 275 for the comparison of milrinone versus placebo, 89 for the comparison of dopamine versus placebo, and 45 for the comparison of amrinone versus milrinone or placebo.

Levosimendan was administered as a continuous infusion without loading dose in 15 studies (Amin 2019; Anastasiadis 2016; Cholley 2017; De Hert 2007; Desai 2018; Erb 2014; Gandham 2013; Hu 2020; Jävelä 2008; Juhl‐Olsen 2015; Kandasamy 2017; Kodalli 2013; Mehta 2017; Shah 2014; Sharma 2014). In seven studies, the continuous administration was preceded by a bolus (Baysal 2014; Eriksson 2009; Ersoy 2013; Lahtinen 2011; Leppikangas 2011; Levin 2012; Mishra 2016). In two studies, levosimendan was administered by bolus only (Tritapepe 2006; Tritapepe 2009). The majority of studies used a dose of 0.1 µg/kg/min for continuous administration (Amin 2019; Anastasiadis 2016; Baysal 2014; Cholley 2017; De Hert 2007; Desai 2018; Erb 2014; Ersoy 2013; Gandham 2013; Juhl‐Olsen 2015; Kandasamy 2017; Kodalli 2013; Levin 2012; Mishra 2016). In four studies, a dose of 0.1 to 0.2 µg/kg/min was used (Hu 2020; Mehta 2017; Shah 2014; Sharma 2014) and in another four studies a dose of 0.2 µg/kg/min was administered (Eriksson 2009; Jävelä 2008; Lahtinen 2011; Leppikangas 2011).

In 15 studies, continuous levosimendan administration was maintained for 24 hours. In Amin 2019 and Juhl‐Olsen 2015, the administration was limited to the entire duration of surgery and in Gandham 2013 for the duration of weaning. Erb 2014 and Hu 2020 limited the administration to a total dose of 12.5 µg. In De Hert 2007 and Kodalli 2013, levosimendan administration ended in the ICU according to the physician's decision to wean from inotropics.

Differences were also seen in the timing of levosimendan administration. In seven studies, levosimendan was administered preoperatively (on the day before surgery (Leppikangas 2011), 24 hours before surgery (Anastasiadis 2016; Levin 2012; Shah 2014; Sharma 2014), 12 hours before surgery (Desai 2018), or 4 hours before surgery (Juhl‐Olsen 2015)). In 12 studies, administration occurred at the start of surgery, i.e. with induction of anaesthesia (Amin 2019; Cholley 2017; Erb 2014; Eriksson 2009; Ersoy 2013; Jävelä 2008; Kandasamy 2017; Kodalli 2013; Lahtinen 2011), before skin incision (Mehta 2017), or before cardiopulmonary bypass (CPB) (Tritapepe 2006; Tritapepe 2009). In four studies, levosimendan administration started at weaning from CPB (Baysal 2014; De Hert 2007; Gandham 2013; Mishra 2016). In one study, levosimendan was administered only after the surgical procedure (Hu 2020).

The studies investigating milrinone were also characterised by a wide variety of timing, dosage, duration, and route of drug administration. Denault 2016 used a single dose as an inhalant at the time of induction of anaesthesia. Jo 2010 administered milrinone perioperatively at a dose of 0.5 µg/kg/min. Kikura 2002 initiated milrinone at weaning from CPB and continued for 10 hours. Hadadzadeh 2013 started milrinone for a period of 24 hours after completion of the surgical procedure. Both Kikura 2002 and Hadadzadeh 2013 initially gave a bolus of 50 µg/kg followed by a continuous dose of 0.5 µg/kg/min.

Participants

The mean age varied between 52.5 and 76 years, with the exception of three studies that were characterised by a more juvenile age structure (Ersoy 2013; Gandham 2013; Mishra 2016). In six studies, inclusion was restricted to adult patients (at least 18 years) (Cholley 2017; Denault 2016; Erb 2014; Lahtinen 2011; Mehta 2017; Tritapepe 2009). Other age restrictions described were: 25 to 70 years (Hu 2020), 30 to 65 years (Kandasamy 2017), 35 to 75 years (Desai 2018), 40 to 70 years (Amin 2019), < 60 years (Gandham 2013), and < 80 years (Anastasiadis 2016). No age restriction was documented in the remaining 17 studies.

The proportion of male participants ranged from 40% (Ersoy 2013) to 93.75% (Anastasiadis 2016) between studies.

A total of 17 studies focused on the examination of patients with left ventricular (LV) dysfunction (Amin 2019; Anastasiadis 2016; Baysal 2014; Cholley 2017; De Hert 2007; Desai 2018; Erb 2014; Eriksson 2009; Ersoy 2013; Hadadzadeh 2013; Hu 2020; Kandasamy 2017; Levin 2012; Mehta 2017; Mishra 2016; Shah 2014; Sharma 2014). In contrast, five studies exclusively included participants without LV dysfunction (Gandham 2013; Jävelä 2008; Juhl‐Olsen 2015; Kikura 2002; Kodalli 2013). In seven studies, both participants with and without LV dysfunction were enrolled (Denault 2016; Gatot 2004; Jo 2010; Lahtinen 2011; Leppikangas 2011; Tritapepe 2006; Tritapepe 2009).

Funding by pharmaceutical industry

Trials that acknowledged funding by the pharmaceutical industry were: Anastasiadis 2016 (supported by Orion Pharma), Cholley 2017 (supported by Orion Pharma), Eriksson 2009 (supported by Orion Pharma), Juhl‐Olsen 2015 (supported by Orion Pharma; author associated with BK medical and GE Healthcare), Lahtinen 2011 (supported by and authors associated with Orion Pharma), Leppikangas 2011 (supported by and authors associated with Orion Pharma), Mehta 2017 (supported by Tenax Therapeutics; authors associated with Orion Pharma and Tenax Therapeutics), and Tritapepe 2006; Tritapepe 2009 (authors associated with Orion Pharma and Abbott). No information was given by Baysal 2014, De Hert 2007, Ersoy 2013, Gatot 2004, Hadadzadeh 2013, Hu 2020, Jävelä 2008, Jo 2010, Levin 2012, or Mishra 2016.

Excluded studies

We excluded 20 trials for at least one of the following reasons: limited to perioperative evaluation, treatment not prophylaxis of LCOS, no randomisation, no assessment of LCOS, goal‐directed dosing of study drugs, exclusion of participants with postoperative LCOS, inclusion of participants with complex heart defects, cross‐over design, post hoc analysis of an included study. Details for each study can be found in the table Characteristics of excluded studies.

Risk of bias in included studies

A summary of all investigated sources of bias in the 29 eligible studies is presented in Figure 2 and Figure 3. The risk of bias tables for the individual trials are given in Characteristics of included studies.

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

3.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

All included studies used intention‐to‐treat analysis and were published in peer‐reviewed journals. The studies were small, with the exception of Mehta 2017, which enrolled 882 participants. Due to the small number of participants included (high imprecision of results), the evidence is considered limited. Other issues identified in the risk of bias review include inappropriate sequencing methods (selection bias) and lack of blinding (performance and detection bias). Funding from the pharmaceutical industry was acknowledged or conflicts of interest were not disclosed in 19 of the 29 studies.

Allocation

We rated the risk of bias for random sequence generation as well as allocation concealment as low for 14 studies (Cholley 2017; De Hert 2007; Denault 2016; Erb 2014; Eriksson 2009; Gandham 2013; Jävelä 2008; Juhl‐Olsen 2015; Kandasamy 2017; Lahtinen 2011; Mehta 2017; Mishra 2016; Tritapepe 2006; Tritapepe 2009). For Anastasiadis 2016, Desai 2018, and Hu 2020, we also assessed the risk of random sequence generation to be low, but there was no information regarding allocation concealment. For Gatot 2004, Hadadzadeh 2013, Kikura 2002, and Leppikangas 2011, we rated the risk of allocation concealment as low, but no information was available regarding random sequence generation. We judged a high risk of bias for random sequence generation in five studies, with a low risk of bias for allocation concealment in Baysal 2014, Levin 2012, Shah 2014, and Sharma 2014, whereas no information regarding allocation concealment was available for Ersoy 2013. Entirely no information regarding allocation was given in Amin 2019, Jo 2010, or Kodalli 2013.

Blinding

We rated the risk of bias for blinding of participants and personnel (performance bias), as well as blinding of outcome assessment (detection bias), as low for 22 studies (Anastasiadis 2016; Cholley 2017; Denault 2016; Erb 2014; Eriksson 2009; Gandham 2013; Gatot 2004; Hadadzadeh 2013; Jävelä 2008; Jo 2010; Juhl‐Olsen 2015; Kandasamy 2017; Kikura 2002; Kodalli 2013; Lahtinen 2011; Leppikangas 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014; Tritapepe 2006; Tritapepe 2009). We judged a high risk of performance bias due to missing blinding of caregivers, yet low detection bias, for Baysal 2014, De Hert 2007, and Mishra 2016. In three studies, blinding was not possible due to different administration of the study drugs (Desai 2018; Ersoy 2013; Hu 2020). No information regarding blinding was given in Amin 2019.

Incomplete outcome data

We rated the risk of bias for incomplete outcome data as low for all studies.

Selective reporting

We rated the risk of bias for selective reporting as low for 12 studies (Anastasiadis 2016; Baysal 2014; Cholley 2017; Denault 2016; Erb 2014; Eriksson 2009; Gandham 2013; Juhl‐Olsen 2015; Lahtinen 2011; Levin 2012; Mehta 2017; Tritapepe 2009). In De Hert 2007, prespecified secondary endpoints were missing, and neither primary nor secondary endpoints were defined in the remaining studies. However, in these studies, the measures listed in the methods section were reported in full.

Other potential sources of bias

We rated the risk of bias for other potential sources of bias as low for 18 studies (Anastasiadis 2016; Cholley 2017; De Hert 2007; Desai 2018; Erb 2014; Eriksson 2009; Gandham 2013; Gatot 2004; Hu 2020; Jävelä 2008; Jo 2010; Kandasamy 2017; Kikura 2002; Lahtinen 2011; Levin 2012; Mehta 2017; Tritapepe 2006; Tritapepe 2009) and unclear for 10 studies due to baseline differences in gender (Amin 2019; Baysal 2014; Denault 2016; Ersoy 2013; Hadadzadeh 2013; Juhl‐Olsen 2015; Kodalli 2013; Leppikangas 2011; Sharma 2014) or baseline differences in age and missing information on gender (Mishra 2016). Although the title and inclusion criteria of the study Shah 2014 implied that the enrolled participants underwent OPCABG, surgery was completed with the help of CPB for 1 of 25 participants from the intervention group and 8 of 25 participants from the comparison group.

None of the included trials reported any cross‐over or were affected by interim results.

Effects of interventions

See: Table 1; Table 2

Amrinone versus placebo

There was only one single‐centre study with 30 participants investigating amrinone compared to placebo (Kikura 2002). All that the authors reported were haemodynamic parameters and the need for additional inotropics.

Primary outcomes

All‐cause mortality (within 30 days)

This outcome was not reported in the included study.

Incidence of LCOS

This outcome was not reported in the included study.

Adverse events

This outcome was not reported in the included study.

Secondary outcomes

Length of in‐hospital and intensive care unit stay

These outcomes were not reported in the included study.

Haemodynamics (cardiac index, MAP, PCWP)

Cardiac index was increased in the amrinone‐treated group compared to the placebo group during this period (2.5 ± 0.2 versus 1.9 ± 0.5 l/min/m²) (Analysis 1.1). With regard to MAP, the study authors reportedly observed no differences between the study groups (observation period: 7 to 10 hours after CPB; no quantitative data available). Information on PCWP was not reported.

1.1. Analysis.

Comparison 1: Comparison 1: Amrinone versus placebo, Outcome 1: Secondary outcome: cardiac index (postoperative nadir)

Duration of mechanical ventilation

This outcome was not reported in the included study.

Proportion of weaning failure within 48 hours of extubation

This outcome was not reported in the included study.

Number of patients requiring mechanical circulatory support (IABP)

This outcome was not reported in the included study.

Number of patients requiring additional inotropic drugs

Dopamine was administered when needed. The authors claimed that the mean total dose of dopamine administered in the first 24 hours after surgery was lower in the amrinone group than in the placebo group. This was said to be due to a slower rate of dopamine infusion in the period 7 to 10 hours after CPB. Quantitative data were not available. No information was given on the number of patients who received additional dopamine.

Number of patients requiring cardiac transplantation, ventricular assist device (VAD) implantation or cardiopulmonary resuscitation

These outcomes were not reported in the included study.

Proportion of renal failure

This outcome was not reported in the included study.

Quality of life

This outcome was not reported in the included study.

Dopamine versus placebo

There was only one single‐centre study with 81 participants investigating dopamine compared to placebo (Gatot 2004).

Primary outcomes

All‐cause mortality (within 30 days)

This outcome was not reported in the included study.

Incidence of LCOS

No RR and resulting estimations of absolute effect were possible since no events were observed in either group (Analysis 2.1).

2.1. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 1: Primary outcome: incidence of LCOS

Adverse events

Information on cardiogenic shock or non‐embolic stroke/transient ischaemic attack was not reported. There was no evidence of a difference in atrial fibrillation with 15 out of 41 participants (36.6%) suffering atrial fibrillation in the dopamine intervention group compared with 20 out of 40 participants (50.0%) in the placebo control group (RR 0.73, 95% CI 0.44 to 1.22; 81 participants, 1 study; very low‐certainty evidence; heterogeneity was not applicable) (Analysis 2.2). No RR and resulting estimations of absolute effect on perioperative myocardial infarction were possible since no events were observed in either group (Analysis 2.3).

2.2. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 2: Primary outcome: adverse events ‐ atrial fibrillation

2.3. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 3: Primary outcome: adverse events ‐ perioperative myocardial infarction

Secondary outcomes

Length of in‐hospital and intensive care unit stay

The mean length of in‐hospital stay was similar between study groups, with 10 ± 3 days in the dopamine intervention group compared with 10 ± 4 days in the placebo control group (Analysis 2.4). The length of intensive care unit stay was not reported in the included study.

2.4. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 4: Secondary outcome: length of hospital stay (days)

Haemodynamics (cardiac index, MAP, PCWP)

While information on cardiac index and PCWP was not reported in the included study, MAP showed little to no difference between groups (87 ± 10 versus 84 ± 10 mmHg at 24 hours after surgery) (Analysis 2.5).

2.5. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 5: Secondary outcome: mean arterial pressure (24 hours after surgery)

Duration of mechanical ventilation

The mean duration of mechanical ventilation in the dopamine intervention group (10 hours) was similar to the placebo control group (mean difference 0.00, 95% CI ‐1.11 to 1.11; 81 participants, 1 study; very low‐certainty evidence; heterogeneity was not applicable) (Analysis 2.6).

2.6. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 6: Secondary outcome: duration of mechanical ventilation (hours)

Proportion of weaning failure within 48 hours of extubation

This outcome was not reported in the included study.

Number of patients requiring mechanical circulatory support (IABP)

This outcome was not reported in the included study.

Number of patients requiring additional inotropic drugs

While no participants in the dopamine intervention group (n = 41) did require additional inotropic drugs, 4 out of 40 participants in the placebo control group required additional inotropic drugs (dopamine). However, the risk ratio showed no evidence of a difference between groups (RR 0.11, 95% CI 0.01 to 1.95) (Analysis 2.7).

2.7. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 7: Secondary outcome: number of participants requiring additional inotropic drugs ‐ dopamine

Number of patients requiring cardiac transplantation, ventricular assist device (VAD) implantation or cardiopulmonary resuscitation

These outcomes were not reported in the included study.

Proportion of renal failure

No participant suffered from renal failure in the dopamine intervention group (n = 41) compared to 4 out of 40 participants in the placebo control group. However, the risk ratio demonstrated no evidence of a difference between groups (RR 0.11, 95% CI 0.01 to 1.95) (Analysis 2.8).

2.8. Analysis.

Comparison 2: Comparison 2: Dopamine versus placebo, Outcome 8: Secondary outcome: proportion of renal failure

Quality of life

This outcome was not reported in the included study.

Milrinone versus placebo

Three small, single‐centre trials covering 150 participants (Hadadzadeh 2013; Jo 2010; Kikura 2002) and one multicentre trial with 124 participants (Denault 2016) investigated milrinone compared to placebo.

Primary outcomes

All‐cause mortality (within 30 days)

Information on mortality within 30 days was restricted to two studies (Hadadzadeh 2013; Jo 2010). However, the relative effect was only generated from one trial (Hadadzadeh 2013), while the relative effect from Jo 2010 was not estimable since no events were observed in either group. When milrinone was compared to the placebo group, there was no evidence of a difference in all‐cause mortality within 30 days (RR 1.00, 95% CI 0.06 to 15.44; 40 participants, 1 study; low‐certainty evidence; heterogeneity was not applicable) (Analysis 3.1).

3.1. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 1: Primary outcome: all‐cause mortality

Incidence of LCOS

This outcome was not reported in any of the included studies.

Adverse events

For adverse events, there was only one study with limited quantitative data for each outcome: non‐embolic stroke/transient ischaemic attack data were provided by Jo 2010, and cardiogenic shock, atrial fibrillation, and perioperative myocardial infarction data were contributed by Hadadzadeh 2013.

All four adverse event outcomes showed little to no difference between groups. For example, 1 out of 40 participants experienced a cardiogenic shock in both the milrinone (2.5%) and placebo groups (2.5%) with a risk ratio of 1.00 (95% CI 0.06 to 15.44; 80 participants, 1 study; low‐certainty evidence; heterogeneity was not applicable) (Analysis 3.2). Moreover, 2 out of 40 participants (5%) suffered atrial fibrillation in the milrinone group compared with 3 out of 40 participants (7.5%) in the placebo group (RR 0.67, 95% CI 0.12 to 3.78; 80 participants, 1 study; low‐certainty evidence; heterogeneity was not applicable) (Analysis 3.3). For perioperative myocardial infarction, 4 out of 40 patients (10%) had this event in the milrinone intervention group compared with 9 out of 40 patients (22.5%) in the placebo control group (RR 0.44, 95% CI 0.15 to 1.33; 80 participants, 1 study; low‐certainty evidence; heterogeneity was not applicable) (Analysis 3.4). Similarly, Jo 2010 narratively reported similar rates of perioperative myocardial infarction between the study groups. Lastly, there was no participant who suffered from non‐embolic stroke/transient ischaemic attack in the milrinone group compared with 1 out of 20 participants (5%) in the placebo group (RR 0.33, 95% CI 0.01 to 7.72; 40 participants, 1 study; low‐certainty evidence; heterogeneity was not applicable) (Analysis 3.5).

3.2. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 2: Primary outcome: adverse events ‐ cardiogenic shock

3.3. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 3: Primary outcome: adverse events ‐ atrial fibrillation

3.4. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 4: Primary outcome: adverse events ‐ perioperative myocardial infarction

3.5. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 5: Primary outcome: adverse events ‐ non‐embolic stroke or transient ischaemic attack

Secondary outcomes

Length of in‐hospital and intensive care unit stay

Information on the length of in‐hospital stay was restricted to Denault 2016 (8 (IQR 6 to 11) versus 7 (IQR 5 to 10) days). Information on the length of intensive care unit stay was restricted to Hadadzadeh 2013 and Denault 2016. Hadadzadeh 2013 reported that the mean length of stay in intensive care in the placebo control group was 2.5 ± 0.59 days. In the milrinone intervention group, the mean length of intensive care stay was 0.15 days shorter (mean difference ‐0.15, 95% CI ‐0.40 to 0.10) (Analysis 3.6). Denault 2016 reported similar median durations of intensive care in both study groups (1.79 (IQR 0.89 to 3.15) versus 1.09 (IQR 0.90 to 2.95) days).

3.6. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 6: Secondary outcome: length of ICU stay (days)

Haemodynamics (cardiac index, MAP, PCWP)

Information on cardiac index was restricted to two studies (Jo 2010; Kikura 2002). There was little to no difference in mean cardiac index at end of surgery with a mean difference of 0.10 (95% CI ‐0.22 to 0.42; 40 participants, 1 study; heterogeneity was not applicable) (Analysis 3.7). However, at 6 hours after surgery and 12 hours after surgery, the mean cardiac index differed between groups with mean differences of 0.50 (95% CI 0.29 to 0.71; 70 participants, 2 studies; low heterogeneity) and 0.50 (95% CI 0.25 to 0.75; 40 participants, 1 study; heterogeneity was not applicable), respectively (Analysis 3.7).

3.7. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 7: Secondary outcome: cardiac index

For MAP, there were two included studies that reported quantitative data (Denault 2016; Jo 2010). Denault 2016 focussed on the perioperative period until the end of surgery. Jo 2010 covered the postoperative period from the end of surgery until 12 hours after surgery. There was little to no difference in mean MAP between groups at baseline (mean difference 2.00, 95% CI ‐2.23 to 6.23; 124 participants, 1 study; heterogeneity was not applicable), at end of surgery (mean difference ‐1.66, 95% CI ‐5.43 to 2.12; 164 participants, 2 studies; low heterogeneity), 6 hours after surgery (mean difference ‐3.00, 95% CI ‐9.82 to 3.82; 40 participants, 1 study; heterogeneity was not applicable), and 12 hours after surgery (mean difference ‐1.00, 95% CI ‐8.19 to 6.19; 40 participants, 1 study; heterogeneity was not applicable) (Analysis 3.8). Kikura 2002 reported no quantitative data but claimed no differences between the study groups in the observation period 7 to 10 hours after CPB.

3.8. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 8: Secondary outcome: mean arterial pressure

There was only one included study providing means of PCWP at three time points (at end of surgery, 6 and 12 hours after surgery) (Jo 2010). There was little to no difference in mean PCWP between groups at the end of surgery and at 12 hours after surgery with mean differences of ‐3.00 (95% CI ‐6.74 to 0.74) and ‐3.00 (95% CI ‐8.00 to 2.00), respectively (Analysis 3.9). However, at 6 hours after surgery, the mean PCWP differed between groups (mean differences ‐4.00, 95% CI ‐7.16 to ‐0.84; 40 participants, 1 study; heterogeneity was not applicable) (Analysis 3.9).

3.9. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 9: Secondary outcome: PCWP

Duration of mechanical ventilation

Information on this outcome was provided by Hadadzadeh 2013 and Denault 2016. Hadadzadeh 2013 showed that the mean duration of mechanical ventilation in the placebo control group was 14.3 hours. The mean duration of mechanical ventilation in the milrinone intervention group was 3.98 hours lower (95% CI ‐7.28 to ‐0.68) (Analysis 3.10). Denault 2016 reported a median duration of mechanical ventilation of 8.5 (IQR 5.3 to 18.2) days in the milrinone intervention group compared with 6.8 (IQR 4.4 to 17) days in the placebo control group.

3.10. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 10: Secondary outcome: duration of mechanical ventilation (hours)

Proportion of weaning failure within 48 hours of extubation

Information on the proportion of weaning failure was reported by Denault 2016 and Jo 2010, covering 164 participants. The risk ratio demonstrated no evidence of a difference between the milrinone intervention group and the placebo control group (RR 1.17, 95% CI 0.37 to 3.69) (Analysis 3.11).

3.11. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 11: Secondary outcome: proportion of weaning failure within 48 hours of extubation

Number of patients requiring mechanical circulatory support (IABP)

Information on the number of patients requiring mechanical circulatory support was provided by two studies (Denault 2016; Hadadzadeh 2013). In the milrinone group, 3 out of 103 participants required mechanical circulatory support compared to 8 out of 101 participants in the control group. However, no evidence of a difference was found in the risk ratio between groups (RR 0.41, 95% CI 0.13 to 1.35; 204 participants, 2 studies; moderate‐certainty evidence; low heterogeneity) (Analysis 3.12).

3.12. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 12: Secondary outcome: number of participants requiring mechanical circulatory support

Number of patients requiring additional inotropic drugs

For overall additional inotropic drugs (without reporting a specific drug type or name), two studies reported that 36 out of 103 patients required additional inotropic drugs in the milrinone intervention group compared to 38 out of 101 patients in the placebo control group (Denault 2016; Hadadzadeh 2013). Little to no difference was found between groups (RR 0.93, 95% CI 0.65 to 1.31; 204 participants, 2 studies; low heterogeneity) (Analysis 3.13).

3.13. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 13: Secondary outcome: number of participants requiring additional inotropic drugs

Within this outcome category, Denault 2016 also reported on specific inotropic drugs, i.e. epinephrine, norepinephrine, dobutamine, phenylephrine, and additional milrinone. In detail, 9 out of 63 participants in the milrinone group required additional epinephrine compared to 7 out of 61 participants in the control group (RR 1.24, 95% CI 0.49 to 3.13) (Analysis 3.14), and 8 out of 63 participants in the milrinone group required additional norepinephrine compared to 8 out of 61 participants in the placebo control group (RR 0.97, 95% CI 0.39 to 2.42) (Analysis 3.15). In addition, Denault 2016 stated that 2 out of 63 participants required additional dobutamine in the milrinone group compared to 1 out of 61 participants in the control group and little to no difference was found between groups (RR 1.94, 95% CI 0.18 to 20.81) (Analysis 3.16). Denault 2016 also provided information on the number of patients that required additional phenylephrine. While no participants required additional phenylephrine in the milrinone group (n = 63), 1 out of 61 participants was in need of phenylephrine in the control group (RR 0.32, 95% CI 0.01 to 7.78) (Analysis 3.17). Notably, in each case, 8 participants in the intervention group (n = 63) and the control group (n = 61) received additional milrinone.

3.14. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 14: Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine

3.15. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 15: Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine

3.16. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 16: Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine

3.17. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 17: Secondary outcome: number of participants requiring additional inotropic drugs ‐ phenylephrine

Kikura 2002 reported that dopamine was administered when needed with the mean total dose of dopamine given in the first 24 hours after surgery being significantly lower in the milrinone intervention group compared to the placebo control group. No information was given on the number of patients who received additional dopamine.

Number of patients requiring cardiac transplantation, ventricular assist device (VAD) implantation or cardiopulmonary resuscitation

These outcomes were not reported in any of the included studies.

Proportion of renal failure

Information on the proportion of renal failure was restricted to two included studies with 120 participants (Hadadzadeh 2013; Jo 2010). There were 2 out of 60 patients suffering from renal failure in the milrinone group compared to 5 out of 60 patients in the control group. However, no evidence of a difference in risk ratio was found between groups (RR 0.40, 95% CI 0.08 to 2.01) (Analysis 3.18).

3.18. Analysis.

Comparison 3: Comparison 3: Milrinone versus placebo, Outcome 18: Secondary outcome: proportion of renal failure

Quality of life

This outcome was not reported in any of the included studies.

Levosimendan versus dobutamine

Two small, single‐centre trials with 140 participants investigated levosimendan compared to dobutamine (Gandham 2013; Kandasamy 2017).

Primary outcomes

All‐cause mortality (within 30 days)

This outcome was not reported in any of the included studies.

Incidence of LCOS

This outcome was not reported in any of the included studies.

Adverse events

Information in both included studies was restricted to atrial fibrillation (Gandham 2013; Kandasamy 2017). The study authors did not report on cardiogenic shock, non‐embolic stroke/transient ischaemic attack or perioperative myocardial infarction. There were 8 out of 70 patients suffering from atrial fibrillation in the levosimendan intervention group compared with 15 out of 70 patients in the dobutamine control group and little to no difference in risk ratio was found between groups (RR 0.53, 95% CI 0.16 to 1.75; 140 participants, 2 studies; low‐certainty evidence; moderate heterogeneity) (Analysis 4.1).

4.1. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 1: Primary outcome: adverse events ‐ atrial fibrillation

Secondary outcomes

Length of in‐hospital and intensive care unit stay

Information on the length of in‐hospital stay was restricted to Kandasamy 2017. The mean stay in hospital in the control group with dobutamine was 6.85 days. The mean length of in‐hospital stay in the levosimendan group was 0.97 days shorter (95% CI ‐1.28 to ‐0.66) (Analysis 4.2).

4.2. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 2: Secondary outcome: length of hospital stay (days)

Information on the length of intensive care stay was reported by both included studies. The mean length of stay in intensive care in the control group with dobutamine was 2.8 to 3.2 days. The mean length of intensive care stay in the levosimendan group was 0.29 days shorter (95% CI ‐0.51 to ‐0.08; 140 participants, 2 studies; moderate‐certainty evidence; low heterogeneity) (Analysis 4.3).

4.3. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 3: Secondary outcome: length of ICU stay (days)

Haemodynamics (cardiac index, MAP, PCWP)

Information on cardiac index was restricted to Gandham 2013, which showed little to no difference at baseline, 6 hours and 12 hours after surgery (Analysis 4.4). However, there was a difference at 24 hours after surgery between groups and this outcome was favourable to the levosimendan group compared to the dobutamine group (mean difference 0.46, 95% CI 0.35 to 0.57; 60 participants, 1 study; heterogeneity was not applicable) (Analysis 4.4).

4.4. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 4: Secondary analysis: cardiac index

MAP was reported by both included studies (Gandham 2013; Kandasamy 2017). No evidence of a difference in mean MAP between the levosimendan intervention group and the dobutamine control group was observed at baseline with ‐0.06 (95% CI ‐2.11 to 2.00) and at 24 hours after surgery with ‐5.42 (95% CI ‐12.52 to 1.68). Instead, at 6 hours and 12 hours after surgery, the mean MAP was lower in the levosimendan intervention group compared with the dopamine control group (mean difference at 6 hours post‐surgery ‐9.27, 95% CI ‐14.18 to ‐4.36; mean difference at 12 hours post‐surgery ‐9.17, 95% CI ‐10.85 to ‐7.50) (Analysis 4.5).

4.5. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 5: Secondary outcome: mean arterial pressure

Information on PCWP was restricted to Kandasamy 2017. There was little to no difference in mean PCWP between groups at baseline (mean difference 0.63, 95% CI ‐0.07 to 1.33). At 6 hours, 12 hours and 24 hours after surgery, however, the mean PCWP was reduced in the levosimendan group compared with control with ‐2.47 (95% CI ‐3.04 to ‐1.90), ‐2.42 (95% CI ‐2.90 to ‐1.94), and ‐1.45 (95% CI ‐1.92 to ‐0.98), respectively (Analysis 4.6).

4.6. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 6: Secondary analysis: PCWP

Duration of mechanical ventilation

Information on this outcome was provided by both included studies, which reported that the mean of duration of mechanical ventilation was lower (mean difference ‐1.04, 95% CI ‐1.72 to ‐0.35) in the intervention group with levosimendan compared to 5.97 to 7.40 hours in the control group with dobutamine (140 participants, 2 studies; low‐certainty evidence; low heterogeneity) (Analysis 4.7).

4.7. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 7: Secondary outcome: duration of mechanical ventilation (hours)

Proportion of weaning failure within 48 hours of extubation

This outcome was not reported in any of the included studies.

Number of patients requiring mechanical circulatory support (IABP)

Only one study provided this outcome (Kandasamy 2017), and reported that 1 out of 40 participants required mechanical circulatory support in both intervention and control groups (Analysis 4.8). Accordingly, no evidence of a difference was found in the risk ratio for this outcome between groups (RR 1.00, 95% CI 0.06 to 15.44; 80 participants, 1 study; low‐certainty evidence; heterogeneity was not applicable) (Analysis 4.8).

4.8. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 8: Secondary outcome: number of participants requiring mechanical circulatory support

Number of patients requiring additional inotropic drugs

The need for overall additional inotropic drugs was restricted to Gandham 2013, who reported that 21 out of 30 patients required additional inotropic drugs in the intervention group with levosimendan compared to 4 out of 30 patients in the dobutamine control group. There was a difference in risk ratio between groups favouring dobutamine (RR 5.25, 95% CI 2.05 to 13.47; 60 participants, 1 study; heterogeneity not applicable) (Analysis 4.9).

4.9. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 9: Secondary outcome: number of participants requiring additional inotropic drugs

Within this outcome category, both included studies reported on specific inotropic drugs, i.e. epinephrine and norepinephrine. Regarding epinephrine, 9 out of 70 participants in the levosimendan group compared to 5 out of 70 participants in the control group received this additional inotropic support (RR 1.63, 95% CI 0.32 to 8.23) (Analysis 4.10). In addition, it was stated that 28 out of 70 participants in the levosimendan group and 8 out of 70 participants in the dobutamine control group required additional norepinephrine (RR 3.52, 95% CI 1.21 to 10.24) (Analysis 4.11).

4.10. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 10: Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine

4.11. Analysis.

Comparison 4: Comparison 4: Levosimendan versus dobutamine, Outcome 11: Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine

Number of patients requiring cardiac transplantation, ventricular assist device (VAD) implantation or cardiopulmonary resuscitation

These outcomes were not reported in any of the included studies.

Proportion of renal failure

This outcome was not reported in any of the included studies.

Quality of life

This outcome was not reported in any of the included studies.

Levosimendan versus standard cardiac care

Three small, single‐centre trials with 208 participants investigated the effect of levosimendan compared to standard cardiac care (Baysal 2014; Desai 2018; Ersoy 2013). See Table 1.

Primary outcomes

All‐cause mortality (within 30 days)

Information on mortality within 30 days was found in all three included studies. However, the relative effect was only generated from two trials (Baysal 2014; Desai 2018), while the relative risk from Ersoy 2013 was not estimable due to no events being observed in either group. When levosimendan was compared to the standard cardiac care group, there was no evidence of a difference in all‐cause mortality within 30 days (RR 0.37, 95% CI 0.13 to 1.04; 208 participants, 3 studies; low‐certainty evidence; low heterogeneity) (Analysis 5.1; Table 1).

5.1. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 1: Primary outcome: all‐cause mortality

Incidence of LCOS

Information on LCOS was reported by all three studies. As with 30‐day mortality, the relative effect was solely generated from two trials (Baysal 2014; Desai 2018), while the relative risk from Ersoy 2013 was not estimable since no events were observed in either group. When the levosimendan intervention group was compared to the control group receiving standard cardiac care, there was little to no difference in LCOS (RR 0.49, 95% CI 0.14 to 1.73; 208 participants, 3 studies; very low‐certainty evidence; moderate heterogeneity) (Analysis 5.2; Table 1).

5.2. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 2: Primary outcome: incidence of LCOS

Adverse events

Information on atrial fibrillation was restricted to two studies (Baysal 2014; Desai 2018). Only Baysal 2014 reported on cardiogenic shock, perioperative myocardial infarction, and non‐embolic stroke/transient ischaemic attack.

When levosimendan was compared to the standard cardiac care group, there was no evidence of a difference in any of the four adverse events. Five out of 64 patients suffered from cardiogenic shock in the levosimendan group compared to 8 out of 64 patients in the standard cardiac care group (RR 0.63, 95% CI 0.22 to 1.81; 128 participants, 1 study; very low‐certainty evidence; heterogeneity was not applicable) (Analysis 5.3; Table 1). There were 10 out of 94 patients suffering from atrial fibrillation in the levosimendan intervention group compared with 23 out of 94 in the standard cardiac care group (RR 0.40, 95% CI 0.11 to 1.41; 188 participants, 2 studies; very low‐certainty evidence; moderate heterogeneity) (Analysis 5.4). Five out 64 participants were afflicted with perioperative myocardial infarction in the levosimendan group compared to 8 out of 64 participants in the standard cardiac care group (RR 0.63, 95% CI 0.22 to 1.81; 128 participants, 1 study; very low‐certainty evidence; heterogeneity was not applicable) (Analysis 5.5). There were 9 out of 64 participants suffering from non‐embolic stroke/transient ischaemic attack in the levosimendan group compared to 16 out of 64 participants in the standard cardiac care group (RR 0.56, 95% CI 0.27 to 1.18; 128 participants, 1 study; very low‐certainty evidence; heterogeneity was not applicable) (Analysis 5.6).

5.3. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 3: Primary outcome: adverse events ‐ cardiogenic shock

5.4. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 4: Primary outcome: adverse events ‐ atrial fibrillation

5.5. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 5: Primary outcome: adverse events ‐ perioperative myocardial infarction

5.6. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 6: Primary outcome: adverse events ‐ non‐embolic stroke or transient ischaemic attack

Secondary outcomes

Length of in‐hospital and intensive care unit stay

There was little to no difference between the study groups in terms of mean length of in‐hospital stay as shown by Analysis 5.7 based on data from Desai 2018 and Ersoy 2013 (mean difference 0.70, 95% CI ‐1.50 to 2.91; 80 participants, 2 studies; substantial heterogeneity). Likewise, Baysal 2014 documented related median durations of hospital care in both study groups with notable intra‐group variability (8.0 (IQR 7.0 to 38.0) versus 9.0 (IQR 7.0 to 37.0) days).

5.7. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 7: Secondary outcome: length of hospital stay (days)

Desai 2018 reported the mean length of stay in intensive care in the control group receiving standard cardiac care to be 3.67 ± 0.76 days. In the levosimendan group, the mean length of intensive care was 0.27 days shorter (95% CI ‐0.62 to 0.08). Ersoy 2013 reported that the mean length of stay in intensive care in the control group was 1.4 ± 1.3 days. In the levosimendan intervention group, the mean length of intensive care was 1.30 days longer (95% CI ‐0.23 to 2.83). Altogether, there was no evidence of a difference in the mean length of stay in the intensive care unit between groups (mean difference 0.33, 95% CI ‐1.16 to 1.83; 80 participants, 2 studies; very low‐certainty evidence; substantial heterogeneity) (Analysis 5.8; Table 1). Baysal 2014 reported similar median durations of intensive care in both study groups with wide intra‐group variability (4.0 (IQR 1.0 to 4.0) versus 5.0 (IQR 2.0 to 37.0) days).

5.8. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 8: Secondary outcome: length of ICU stay (days)

Haemodynamics (cardiac index, MAP, PCWP)

Information on cardiac index was provided by Baysal 2014, Desai 2018, and Ersoy 2013. There was little to no difference at baseline (mean difference 0.00, 95% CI ‐0.10 to 0.10; 208 participants, 3 studies; low heterogeneity). However, there were differences at the end of surgery as well as 6 hours, 12 hours, and 24 hours after surgery between groups and this outcome was favourable to the levosimendan group compared to the standard cardiac care group (mean difference at end of surgery 0.52, 95% CI 0.32 to 0.73; 188 participants, 2 studies; moderate heterogeneity; mean difference at 6 hours post surgery 0.77, 95% CI 0.64 to 0.89; 188 participants, 2 studies; low heterogeneity; mean difference at 12 hours post surgery 0.55, 95% CI 0.34 to 0.76; 60 participants, 1 study; heterogeneity was not applicable; and mean difference at 24 hours post surgery 0.47, 95% CI 0.35 to 0.58; 188 participants, 2 studies; low heterogeneity) (Analysis 5.9).

5.9. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 9: Secondary outcome: cardiac index

MAP was only reported by Baysal 2014 and Desai 2018. Little to no difference in mean MAP between the levosimendan intervention group and the standard cardiac care control group was observed at the time points at which quantitative data were available from both studies, i.e. at baseline with 0.41 (95% CI ‐0.78 to 1.59), at end of surgery with 2.39 (95% CI ‐8.74 to 13.52), at 6 hours after surgery with ‐1.11 (95% CI ‐2.36 to 0.14), and at 24 hours after surgery with 3.54 (95% CI ‐0.06 to 7.15) (Analysis 5.10). However, at 12 hours after surgery (a time point that was only available for Desai 2018) the mean MAP was higher in the levosimendan intervention group compared with the standard cardiac care control group (mean difference 4.07, 95% CI 1.03 to 7.11) (Analysis 5.10).

5.10. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 10: Secondary outcome: mean arterial pressure

Information on PCWP was restricted to Baysal 2014 and Desai 2018. Baysal 2014 narratively reported no significant differences between study groups. In the study Desai 2018 there was no evidence of a difference in mean PCWP between groups at baseline (mean difference 0.50, 95% CI ‐0.63 to 1.63) and 24 hours after surgery (mean difference ‐0.73, 95% CI ‐1.48 to 0.02) (Analysis 5.11). However, at the end of surgery, 6 hours and 12 hours after surgery, the mean PCWP was reduced in the levosimendan group compared with the standard cardiac care control group with ‐2.07 (95% CI ‐3.06 to ‐1.08), ‐1.21 (95% CI ‐2.22 to ‐0.20), and ‐1.04 (95% CI ‐1.69 to ‐0.39), respectively (Analysis 5.11).

5.11. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 11: Secondary outcome: PCWP

Duration of mechanical ventilation

Information on this outcome was provided only by Baysal 2014. The included study showed that the mean duration of mechanical ventilation in the standard cardiac care control group was 83.2 hours. The mean duration of mechanical ventilation in the levosimendan intervention group was 3.40 hours lower (95% CI ‐11.50 to 4.70; 128 participants, 1 study; very low‐certainty evidence; heterogeneity not applicable) (Analysis 5.12; Table 1).

5.12. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 12: Secondary outcome: duration of mechanical ventilation (hours)

Proportion of weaning failure within 48 hours of extubation

Information on the proportion of weaning failure was restricted to Baysal 2014, which found little to no difference between the levosimendan and standard cardiac care group (RR 0.82, 95% CI 0.36 to 1.84; 128 participants, 1 study; very low‐certainty evidence; heterogeneity was not applicable) (Analysis 5.13).

5.13. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 13: Secondary outcome: proportion of weaning failure within 48 hours of extubation

Number of patients requiring mechanical circulatory support (IABP)

Information on the number of patients requiring mechanical circulatory support was given by all three included studies. However, the relative effect was only generated from two trials (Baysal 2014; Desai 2018), while the relative risk from Ersoy 2013 was not estimable due to no events being observed in either group. In the levosimendan intervention group, 17 out of 104 patients required mechanical circulatory support compared to 20 out of 104 patients in the standard cardiac care control group. However, no evidence of a difference was found in the risk ratio between groups (RR 0.88, 95% CI 0.50 to 1.55; 208 participants, 3 studies; low‐certainty evidence; low heterogeneity) (Analysis 5.14; Table 1).

5.14. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 14: Secondary outcome: number of participants requiring mechanical circulatory support

Number of patients requiring additional inotropic drugs

The need for overall additional inotropic drugs was restricted to Desai 2018 and Ersoy 2013, who reported that 28 out of 40 patients required additional inotropic drugs in the intervention group with levosimendan compared to 16 out of 40 patients in the standard cardiac care control group. There was a difference in risk ratio between groups favouring standard cardiac care (RR 1.70, 95% CI 1.13 to 2.57; 80 participants, 2 studies; low heterogeneity) (Analysis 5.15).

5.15. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 15: Secondary outcome: number of participants requiring additional inotropic drugs

Within this outcome category, information on specific inotropic drugs, i.e. epinephrine, dopamine, and dobutamine, was reported by Baysal 2014, while information on norepinephrine was provided by Baysal 2014 and Desai 2018. There was little to no difference in risk ratio between groups for all of these inotropic drugs. Regarding epinephrine, 16 out of 64 participants in the levosimendan group compared to 23 out of 64 participants in the standard cardiac care control group received this additional inotropic support (RR 0.70, 95% CI 0.41 to 1.19) (Analysis 5.16). Additional norepinephrine was required by 41 out of 94 patients in the levosimendan group and 36 out of 94 patients in the standard cardiac care group (RR 1.18, 95% CI 0.58 to 2.37) (Analysis 5.17). Moreover, 31 out of 64 participants in the levosimendan group and 40 out of 64 participants in the standard cardiac care control group required additional dopamine (RR 0.78, 95% CI 0.56 to 1.06) (Analysis 5.18). Lastly, 55 out of 64 participants in the levosimendan group compared to 53 out of 64 participants in the standard cardiac care control group required additional dobutamine (RR 1.04, 95% CI 0.89 to 1.20) (Analysis 5.19).

5.16. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 16: Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine

5.17. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 17: Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine

5.18. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 18: Secondary outcome: number of participants requiring additional inotropic drugs ‐ dopamine

5.19. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 19: Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine

Number of patients requiring cardiac transplantation, ventricular assist device (VAD) implantation or cardiopulmonary resuscitation

Information on these outcomes was restricted to Baysal 2014. The relative risk of cardiac transplantation or VAD implantation was not estimable since no events were observed in either group (Analysis 5.20). In the intervention group with levosimendan, 4 out of 64 patients required cardiopulmonary resuscitation compared to 10 out of 64 patients in the standard cardiac care control group. However, no evidence of a difference was found in the risk ratio between groups (RR 0.40, 95% CI 0.13 to 1.21; 128 participants, 1 study; heterogeneity was not applicable) (Analysis 5.21).

5.20. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 20: Secondary outcome: number of participants requiring cardiac transplantation

5.21. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 21: Secondary outcome: number of participants requiring cardiopulmonary resuscitation

Proportion of renal failure

All three studies reported on the proportion of renal failure. Ersoy 2013 observed no events in either group. Therefore, the relative effect was only generated from Baysal 2014 and Desai 2018. In the levosimendan intervention group, 8 out of 104 patients suffered from renal failure compared with 17 out of 104 patients in the standard cardiac care control group. No evidence of a difference was found in the risk ratio between groups (RR 0.48, 95% CI 0.22 to 1.08; 208 participants, 3 studies; low heterogeneity) (Analysis 5.22).

5.22. Analysis.

Comparison 5: Comparison 5: Levosimendan versus standard cardiac care, Outcome 22: Secondary outcome: proportion of renal failure

Quality of life

Only Ersoy 2013 narratively reported this outcome, while Baysal 2014 and Desai 2018 did not report it. Ersoy 2013 stated that no marked adverse reaction to the study drug was observed in the levosimendan group.

Levosimendan versus milrinone

Three small, single‐centre trials with 130 participants investigated the effect of levosimendan compared to milrinone (Amin 2019; De Hert 2007; Mishra 2016).

Primary outcomes

All‐cause mortality (within 30 days)

Information on mortality within 30 days was restricted to two studies (De Hert 2007; Mishra 2016). However, the relative effect was only generated from De Hert 2007, while the relative risk in Mishra 2016 was not estimable due to no events being observed in either group. When levosimendan was compared to milrinone, there was little to no difference in all‐cause mortality within 30 days (RR 0.14, 95% CI 0.01 to 2.55; 70 participants, 2 studies; low‐certainty evidence; heterogeneity was not applicable) (Analysis 6.1).

6.1. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 1: Primary outcome: all‐cause mortality

Incidence of LCOS

This outcome was not reported in any of the included studies.

Adverse events

While information on cardiogenic shock and non‐embolic stroke/transient ischaemic attack was not reported in any of the included studies, information on atrial fibrillation and perioperative myocardial infarction were restricted to one study (De Hert 2007). In the levosimendan intervention group, 6 out of 15 participants suffered from atrial fibrillation compared to 7 out of 15 participants in the milrinone control group; no evidence of a difference was found in the risk ratio between groups (RR 0.86, 95% CI 0.38 to 1.95; 30 participants, 1 study; very low‐certainty evidence; heterogeneity was not applicable) (Analysis 6.2). The relative effect on perioperative myocardial infarction was not estimable due to no events being observed in either the levosimendan or milrinone groups (Analysis 6.3).

6.2. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 2: Primary outcome: adverse events ‐ atrial fibrillation

6.3. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 3: Primary outcome: adverse events ‐ perioperative myocardial infarction

Secondary outcomes

Length of in‐hospital and intensive care unit stay

Information on the length of in‐hospital stay was restricted to De Hert 2007 (10 (IQR 7 to 16) versus 12 (IQR 5 to 39) days).

Amin 2019 reported that the mean length of stay in intensive care in the control group receiving milrinone was 4.25 ± 1.75 days. In the levosimendan group, the mean length of intensive care was 0.90 days shorter (mean difference ‐0.90, 95% CI ‐1.67 to ‐0.13). Mishra 2016 reported that the mean length of stay in intensive care in the control group receiving milrinone was 4.56 ± 1.62 days. In the levosimendan intervention group, the mean length of intensive care was 0.31 days shorter (mean difference ‐0.31, 95% CI ‐1.34 to 0.72). Altogether, there was a difference in the mean length of stay in the intensive care unit between groups, favouring levosimendan (mean difference ‐0.69, 95% CI ‐1.31 to ‐0.07; 100 participants, 2 studies; low‐certainty evidence; low heterogeneity) (Analysis 6.4). In line with this, De Hert 2007 reported a similar length of stay in the intensive care unit (2.58 (IQR 1.17 to 5.04) versus 2.75 (IQR 1.42 to 39) days).

6.4. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 4: Secondary outcome: length of ICU stay (days)

Haemodynamics (cardiac index, MAP, PCWP)

The only available information regarding cardiac index was the narrative statement by Mishra 2016 that the cardiac index was comparable between groups during the study period, except at 6 hours of intensive care, when it was significantly higher in the levosimendan group compared with the milrinone group.

Quantitative data on MAP was only reported by De Hert 2007 with 30 participants. No evidence of a difference in mean MAP between the levosimendan intervention group and the milrinone control group was observed at baseline with 2.00 (95% CI ‐4.60 to 8.60), at end‐of‐surgery with 3.00 (95% CI ‐1.53 to 7.53), at 6 hours after surgery with ‐3.00 (95% CI ‐10.16 to 4.16), at 12 hours after surgery with 0.00 (95% CI ‐6.48 to 6.48), and at 24 hours after surgery with ‐2.00 (95% CI ‐7.01 to 3.01) (Analysis 6.5). Amin 2019 narratively reported no significant differences in MAP from anaesthesia to the first 24 hours in intensive care. Mishra 2016 reported MAP to be comparable in both groups at all time points except at baseline when it was significantly lower in the levosimendan group.

6.5. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 5: Secondary outcome: mean arterial pressure

Information on PCWP was not reported in any of the included studies.

Duration of mechanical ventilation

Information on this outcome was available in all three included studies. Amin 2019 reported the mean duration of mechanical ventilation in the milrinone group to be 18.76 ± 9.8 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 6.50 hours lower (mean difference ‐6.50, 95% CI ‐10.65 to ‐2.35). De Hert 2007 reported the mean duration of mechanical ventilation in the milrinone group to be 20 ± 11 hours, whereas in the intervention group receiving levosimendan it was 9 hours lower (mean difference ‐9.00, 95% CI ‐15.60 to ‐2.40). Similarly, Mishra 2016 reported a shorter mean duration of mechanical ventilation in the levosimendan group compared with the milrinone group (15.36 ± 3.32 hours versus 21.87 ± 15.38 hours). In summary, there was a difference in the mean duration of mechanical ventilation between groups, favouring the levosimendan intervention group (mean difference ‐7.06, 95% CI ‐10.19 to ‐3.93; 130 participants, 3 studies; low‐certainty evidence; low heterogeneity) (Analysis 6.6).

6.6. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 6: Secondary outcome: duration of mechanical ventilation (hours)

Proportion of weaning failure within 48 hours of extubation

This outcome was not reported in any of the included studies.

Number of patients requiring mechanical circulatory support (IABP)

Information on the number of patients requiring mechanical circulatory support was restricted to Amin 2019 and De Hert 2007. Amin 2019 reported that of 30 patients in the levosimendan intervention group, no patient required mechanical circulatory support compared to 3 out of 30 patients in the milrinone control group (RR 0.14, 95% CI 0.01 to 2.65). De Hert 2007 reported that 4 out of 15 patients in both the levosimendan and milrinone groups required mechanical circulatory support (RR 1.00, 95% CI 0.31 to 3.28). Altogether, there was little to no difference in the number of patients requiring mechanical circulatory support (RR 0.58, 95% CI 0.09 to 3.55; 90 participants, 2 studies; low‐certainty evidence; low heterogeneity) (Analysis 6.7).

6.7. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 7: Secondary outcome: number of participants requiring mechanical circulatory support

Number of patients requiring additional inotropic drugs

For this outcome, quantitative data on specific inotropic drugs, i.e. norepinephrine and dobutamine, were only reported by De Hert 2007. There was no evidence of a difference in risk ratio between groups for both specific inotropic drugs. Regarding norepinephrine, all 15 participants in both the levosimendan and milrinone groups received this additional inotropic support (RR 1.00, 95% CI 0.88 to 1.13) (Analysis 6.8). Similarly, 15 out of 15 participants in levosimendan and milrinone groups required additional dobutamine (RR 1.00, 95% CI 0.88 to 1.13) (Analysis 6.9). Mishra 2016 narratively reported that at several time points after termination of cardiopulmonary bypass, more patients in the levosimendan group required significantly higher doses of norepinephrine infusion compared with those in the milrinone group.

6.8. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 8: Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine

6.9. Analysis.

Comparison 6: Comparison 6: Levosimendan versus milrinone, Outcome 9: Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine

Number of patients requiring cardiac transplantation, ventricular assist device (VAD) implantation or cardiopulmonary resuscitation

This outcome was not reported in any of the included studies.

Proportion of renal failure

This outcome was not reported in any of the included studies.

Quality of life

This outcome was not reported in any of the included studies.

Levosimendan versus placebo

Twelve small, single‐centre trials with 869 participants (Anastasiadis 2016; Erb 2014; Hu 2020; Jävelä 2008; Juhl‐Olsen 2015; Kodalli 2013; Lahtinen 2011; Leppikangas 2011; Shah 2014; Sharma 2014; Tritapepe 2006; Tritapepe 2009), three small, multicentre trials with 645 participants (Cholley 2017; Eriksson 2009; Levin 2012), and one large, multicentre trial with 849 participants (Mehta 2017) investigated levosimendan compared to placebo. See Table 2.

Primary outcomes

All‐cause mortality (within 30 days)

Information on mortality within 30 days was found in 14 included studies (Anastasiadis 2016; Cholley 2017; Erb 2014; Eriksson 2009; Hu 2020; Jävelä 2008; Juhl‐Olsen 2015; Kodalli 2013; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014; Tritapepe 2009). However, the relative effect was only generated from 11 included studies (Anastasiadis 2016; Cholley 2017; Erb 2014; Eriksson 2009; Hu 2020; Jävelä 2008; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014), while the relative risk in Juhl‐Olsen 2015, Kodalli 2013, and Tritapepe 2009 was not estimable due to no events being observed in either group. When levosimendan was compared to the control group, there was a small effect to no difference in all‐cause mortality within 30 days, favouring the levosimendan group (RR 0.65, 95% CI 0.43 to 0.97; 2347 participants, 14 studies; moderate‐certainty evidence; low heterogeneity; no indication of reporting bias) (Analysis 7.1; Figure 4; Table 2). Derived from that risk difference, the number needed to treat for an additional beneficial outcome (NNTB) is 44.

7.1. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 1: Primary outcome: all‐cause mortality

4.

Incidence of LCOS

Information on LCOS was reported by six studies (Cholley 2017; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014). When the levosimendan intervention group was compared to the control group receiving placebo, there were differences in LCOS and this effect was favourable to the levosimendan group (RR 0.43, 95% CI 0.25 to 0.74; 1724 participants, 6 studies; low‐certainty evidence; substantial heterogeneity) (Analysis 7.2; Table 2). Derived from that risk difference, the calculated NNTB is 8.

7.2. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 2: Primary outcome: incidence of LCOS

Adverse events

Information on cardiogenic shock was restricted to Cholley 2017, Kodalli 2013, and Mehta 2017, but the relative effect was only provided by Cholley 2017 and Mehta 2017, while the risk ratio from Kodalli 2013 was not estimable due to no events being observed in either group. Information on atrial fibrillation was provided by 11 studies (Anastasiadis 2016; Cholley 2017; Juhl‐Olsen 2015; Kodalli 2013; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014; Tritapepe 2006; Tritapepe 2009). However, the relative effect was only generated from 10 included studies (Anastasiadis 2016; Cholley 2017; Juhl‐Olsen 2015; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014; Tritapepe 2006; Tritapepe 2009), while the risk ratio from Kodalli 2013 was not estimable due to no events being observed in either group. Information on perioperative myocardial infarction was found in eight studies (Anastasiadis 2016; Cholley 2017; Kodalli 2013; Lahtinen 2011; Levin 2012; Mehta 2017; Sharma 2014; Tritapepe 2009). However, the relative effect was only generated from six included studies (Cholley 2017; Lahtinen 2011; Levin 2012; Mehta 2017; Sharma 2014; Tritapepe 2009), while the risk ratio from Anastasiadis 2016 and Kodalli 2013 was not estimable due to no events being observed in either group. Eight studies reported on non‐embolic stroke/transient ischaemic attack (Anastasiadis 2016; Cholley 2017; Kodalli 2013; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014). Similar to atrial fibrillation and cardiogenic shock, only Kodalli 2013 was unable to provide the risk ratio due to no events being observed in either group. Erb 2014 narratively reported that the rates of any serious adverse events were significantly reduced in the levosimendan group compared with the placebo group.

When levosimendan was compared to the control group, there was little to no difference in all four adverse events. Twenty‐five out of 610 participants suffered from cardiogenic shock in the levosimendan group compared to 38 out of 602 in the control group (RR 0.65, 95% CI 0.40 to 1.05; 1212 participants, 3 studies; high‐certainty evidence; low heterogeneity) (Analysis 7.3; Table 2). There were 375 out of 971 patients suffering from atrial fibrillation in the levosimendan intervention group compared with 356 out of 963 in the placebo control group (RR 1.02, 95% CI 0.82 to 1.27; 1934 participants, 11 studies; low‐certainty evidence; substantial heterogeneity, indication of reporting bias) (Analysis 7.4; Figure 5). Eighty‐six out 924 participants were afflicted with perioperative myocardial infarction in the levosimendan group compared to 94 out of 914 participants in the control group (RR 0.89, 95% CI 0.61 to 1.31; 1838 participants, 8 studies; moderate‐certainty evidence; low heterogeneity) (Analysis 7.5). There were 36 out of 897 participants suffering from non‐embolic stroke/transient ischaemic attack in the levosimendan group compared to 44 out of 889 participants in the control group (RR 0.89, 95% CI 0.58 to 1.38; 1786 participants, 8 studies; moderate‐certainty evidence; low heterogeneity) (Analysis 7.6).

7.3. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 3: Primary outcome: adverse events ‐ cardiogenic shock

7.4. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 4: Primary outcome: adverse events ‐ atrial fibrillation

5.

7.5. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 5: Primary outcome: adverse events ‐ perioperative myocardial infarction

7.6. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 6: Primary outcome: adverse events ‐ non‐embolic stroke or transient ischaemic attack

Secondary outcomes

Length of in‐hospital and intensive care unit stay

There were six studies reporting the mean length of stay in hospital (Anastasiadis 2016; Lahtinen 2011; Leppikangas 2011; Sharma 2014; Tritapepe 2006; Tritapepe 2009). All showed a similar trend of an observed lower mean length of stay in hospital among patients in the levosimendan group compared to the control group. Altogether, the meta‐analysis revealed a difference in the mean length of stay in hospital between groups, favouring the levosimendan intervention group (mean difference ‐1.80, 95% CI ‐3.27 to ‐0.33; 422 participants, 6 studies; considerable heterogeneity) (Analysis 7.7). IQR data were reported by Cholley 2017 and Erb 2014. While Cholley 2017 reported identical lengths of hospital stay for both groups (7 (IQR 5 to 10) days), Erb 2014 observed a lower median length of stay in hospital in the levosimendan group compared to the placebo group (12 (IQR 10.25 to 21.75) versus 13.5 (IQR 10.25 to 21.5) days).

7.7. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 7: Secondary outcome: length of hospital stay (days)

The relative effect of levosimendan compared to placebo on length of intensive care stay was generated from seven included studies (Anastasiadis 2016; Hu 2020; Kodalli 2013; Leppikangas 2011; Sharma 2014; Tritapepe 2006; Tritapepe 2009). In these studies, the mean length of stay in intensive care varied from 1.07 to 8.15 days in the placebo control group. All reported a lower mean length of stay observed in the levosimendan intervention group compared to the control group. Accordingly, the meta‐analysis revealed a difference in the mean length of stay in intensive care between groups, favouring the levosimendan group (mean difference ‐1.00, 95% CI ‐1.63 to ‐0.37; 572 participants, 7 studies; very low‐certainty evidence; considerable heterogeneity) (Analysis 7.8; Table 2). In five studies, information on length of stay in intensive care was restricted to IQR data. For this, Cholley 2017, Eriksson 2009, and Mehta 2017 reported a similar length of stay in both study groups (Cholley 2017: 4 (IQR 2 to 7) versus 4 (IQR 3 to 6) days; Eriksson 2009: 2 (IQR 1 to 33) versus 2 (IQR 1 to 31) days; Mehta 2017: 2.8 (IQR 1.6 to 4.8) versus 2.9 (IQR 1.8 to 4.9) days). However, Erb 2014 observed a length of stay in intensive care of 3.0 (IQR 1.5 to 7.0) days in the levosimendan group compared with 5.0 (IQR 4.0 to 13.8) days in the placebo group, and Juhl‐Olsen 2015 reported a length of stay in intensive care of 0.79 (IQR 0.65 to 1.28) days in the levosimendan group compared with 0.83 (IQR 0.75 to 5.80) days in the placebo group. Jävelä 2008 narratively stated that length of stay was similar between study groups.

7.8. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 8: Secondary outcome: length of ICU stay (days)

Haemodynamics (cardiac index, MAP, PCWP)

Quantitative information on mean cardiac index was provided by Anastasiadis 2016, Hu 2020, Jävelä 2008, Kodalli 2013, and Shah 2014. Little to no difference in mean cardiac index between the levosimendan intervention group and the placebo control group was reported by these studies at baseline with ‐0.04 (95% CI ‐0.17 to 0.10; 456 participants, 5 studies; moderate heterogeneity) and at end of surgery with 0.67 (95% CI ‐0.20 to 1.53; 74 participants, 2 studies; considerable heterogeneity) as reported by Jävelä 2008 and Shah 2014 (Analysis 7.9). However, there were differences in mean cardiac index between groups at 6 hours, 12 hours, and 24 hours after surgery that were all favourable to levosimendan (Analysis 7.9). For example, Jävelä 2008 and Kodalli 2013 reported that the mean difference in cardiac index between groups at 6 hours after surgery was 0.68 (95% CI 0.28 to 1.09; 54 participants, 2 studies; low heterogeneity). In addition, Jävelä 2008, Kodalli 2013, and Shah 2014 reported a mean difference in cardiac index between groups at 12 hours after surgery of 0.73 (95% CI 0.23 to 1.23; 104 participants, 3 studies; substantial heterogeneity). Moreover, all included studies reported that the mean difference in cardiac index between groups at 24 hours after surgery was 0.51 (95% CI 0.06 to 0.96; 456 participants, 5 studies; considerable heterogeneity). There was narrative information on cardiac index from Erb 2014, Eriksson 2009, Juhl‐Olsen 2015, Lahtinen 2011, Leppikangas 2011, Levin 2012, Sharma 2014, and Tritapepe 2009. For this, Erb 2014 and Leppikangas 2011 observed no significant differences between study groups. All other studies described higher values in the levosimendan intervention group compared to the placebo control group, which were said to be significant in Eriksson 2009 (at end of surgery and at 8 hours after declamping aorta), Lahtinen 2011 (postoperative measurements), Levin 2012 (immediately following administration of the levosimendan loading dose until 48 hours postoperatively) and Tritapepe 2009 (after surgery). Tritapepe 2006 provided IQR data that revealed a higher median cardiac index in the levosimendan group compared to the placebo group after surgery (baseline: 2.3 (IQR 2.0 to 2.6) versus 2.4 (IQR 2.1 to 2.7) mmHg); end of surgery: 2.3 (IQR 2.2 to 2.5) versus 2.1 (IQR 1.9 to 2.3) mmHg; 6 hours after surgery: 2.9 (IQR 2.5 to 3.1) versus 2.4 (IQR 2.3 to 2.8) mmHg; 24 hours after surgery: 3.0 (IQR 2.7 to 3.4) versus 2.7 (IQR 2.3 to 2.8) mmHg).

7.9. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 9: Secondary outcome: cardiac index

There was a difference in mean MAP between the levosimendan intervention group and the placebo control group at 6 hours after surgery as reported by Hu 2020, Jävelä 2008, and Kodalli 2013 with ‐4.81 (95% CI ‐6.08 to ‐3.55; 374 participants, 3 studies; low heterogeneity) (Analysis 7.10). However, no evidence of a difference was observed between groups at baseline, end of surgery, 12 hours, and 24 hours after surgery (Analysis 7.10). For instance, Anastasiadis 2016, Eriksson 2009, Jävelä 2008, and Kodalli 2013 reported that the mean difference between groups at baseline was 2.91 (95% CI ‐3.19 to 9.01; 146 participants, 4 studies; moderate heterogeneity). Moreover, three studies, Eriksson 2009, Hu 2020, and Jävelä 2008, reported that the mean difference between groups at end of surgery was ‐2.06 (95% CI ‐6.50 to 2.38; 404 participants, 3 studies; substantial heterogeneity). In addition, Hu 2020, Jävelä 2008, and Kodalli 2013 reported that the mean difference between groups at 12 hours after surgery was ‐4.57 (95% CI ‐9.19 to 0.05; 374 participants, 3 studies; moderate heterogeneity). At 12 hours after surgery, five studies found that the mean difference between groups was ‐2.19 (95% CI ‐5.84 to 1.46; 466 participants, 5 studies; moderate heterogeneity) (Anastasiadis 2016; Eriksson 2009; Hu 2020; Jävelä 2008; Kodalli 2013). Narrative information on MAP was given by Erb 2014; Juhl‐Olsen 2015; Leppikangas 2011; Sharma 2014, and Tritapepe 2009. Whereas Erb 2014, Juhl‐Olsen 2015, and Leppikangas 2011 documented no significant difference in MAP between groups during the study period, Tritapepe 2009 stated that after surgery there were significantly higher values of MAP in those treated with levosimendan. Likewise, Sharma 2014 noted that the levosimendan group showed better MAP in both intra‐ and postoperative periods, and that the difference was statistically significant at all time points without providing quantitative data. Tritapepe 2006 provided IQR data that were similar in dimension between study groups at baseline, end of surgery, and 24 hours after surgery (baseline: 72.5 (IQR 65.2 to 77.7) versus 70.0 (IQR 65.8 to 76.5) mmHg); end of surgery: 69.0 (IQR 65.0 to 76.7) versus 68.0 (IQR 65.0 to 71.5) mmHg; 24 hours after surgery: 78.5 (IQR 73.2 to 84.5) versus 74.5 (IQR 70.0 to 80.0) mmHg)). However, at 6 hours after surgery, median MAP was higher in the levosimendan group compared to the placebo control group (77.5 (IQR 72.2 to 79.7) versus 70.00 (IQR 68.0 to 77.0) mmHg).

7.10. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 10: Secondary outcome: mean arterial pressure

Little to no difference was reported in mean PCWP between the levosimendan group and the control group at baseline, end of surgery, 6 hours, and 12 hours after surgery (Analysis 7.11). At baseline, five studies reported the mean difference between groups was 0.11 (95% CI ‐1.02 to 1.24; 196 participants, 5 studies; low heterogeneity) (Anastasiadis 2016; Eriksson 2009; Jävelä 2008; Kodalli 2013; Shah 2014). Moreover, Eriksson 2009, Jävelä 2008, and Shah 2014 reported that the mean difference between groups at end of surgery was ‐1.63 (95% CI ‐3.85 to 0.58; 134 participants, 3 studies; considerable heterogeneity). At 6 hours after surgery, Jävelä 2008 and Kodalli 2013 reported that the mean difference between groups was 0.31 (95% CI ‐0.82 to 1.44; 54 participants, 2 studies; low heterogeneity). At 12 hours after surgery, three studies found that the mean difference between groups was ‐1.35 (95% CI ‐4.05 to 1.36; 104 participants, 3 studies; considerable heterogeneity) (Jävelä 2008; Kodalli 2013; Shah 2014). However, there was a difference in mean PCWP between groups at 24 hours after surgery as reported by Anastasiadis 2016, Eriksson 2009, Jävelä 2008, Kodalli 2013, and Shah 2014, with ‐1.33 (95% CI ‐2.50 to ‐0.17; 196 participants, 5 studies; moderate heterogeneity) (Analysis 7.11). Tritapepe 2006 provided IQR data that were similar in dimension between study groups at baseline, end of surgery, 6 hours, and 24 hours after surgery (baseline: 14.0 (IQR 12.0 to 15.7) versus 12.5 (IQR 11.2 to 15.7) mmHg); end of surgery: 14.5 (IQR 12.2 to 17.5) versus 15.5 (IQR 13.2 to 17.0) mmHg; 6 hours after surgery: 13.0 (IQR 10.5 to 15.0) versus 14.0 (IQR 11.2 to 15.0) mmHg; 24 hours after surgery: 11.0 (IQR 10.0 to 13.7) versus 13.0 (IQR 11.0 to 14.0) mmHg)). Sharma 2014 narratively reported that there was a significant reduction in mean PCWP in the levosimendan group, which started after the infusion and was sustained postoperatively.

7.11. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 11: Secondary outcome: PCWP

Duration of mechanical ventilation

The mean duration of mechanical ventilation was reported by Anastasiadis 2016, Hu 2020, Kodalli 2013, Leppikangas 2011, Sharma 2014, Tritapepe 2006, and Tritapepe 2009 with 572 participants. Anastasiadis 2016 reported that the mean duration of mechanical ventilation in the control group was 29.9 ± 54.3 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 17.80 hours lower (mean difference ‐17.80, 95% CI ‐44.63 to 9.03). Hu 2020 reported that the mean duration of mechanical ventilation in the control group was 157.44 ± 58.56 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 53.76 hours lower (mean difference ‐53.76, 95% CI ‐63.79 to ‐43.73). Kodalli 2013 observed a mean duration of mechanical ventilation in the control group of 7.1 ± 1.5 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 0.10 hours lower (mean difference ‐0.10, 95% CI ‐1.33 to 1.13). Leppikangas 2011 found the mean duration of mechanical ventilation in the control group was 5.67 ± 1.47 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 4.10 hours higher (mean difference 4.10, 95% CI ‐1.76 to 9.96). Sharma 2014 reported that the mean duration of mechanical ventilation in the control group was 37.65 ± 3.48 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 9.55 hours lower (mean difference ‐9.55, 95% CI ‐11.47 to ‐7.63). Tritapepe 2006 documented a mean duration of mechanical ventilation in the control group of 11 ± 3 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 1.00 hours lower (mean difference ‐1.00, 95% CI ‐3.40 to 1.40). Tritapepe 2009 reported the mean duration of mechanical ventilation in the control group was 13.6 ± 4.5 hours. The mean duration of mechanical ventilation in the intervention group receiving levosimendan was 2.30 hours lower (mean difference ‐2.30, 95% CI ‐3.72 to ‐0.88). In summary, there was a difference in the mean duration of mechanical ventilation between groups, favouring the levosimendan group (mean difference ‐8.03, 95% CI ‐13.17 to ‐2.90; 572 participants, 7 studies; very low‐certainty evidence; considerable heterogeneity) (Analysis 7.12; Table 2). IQR data were reported by Cholley 2017, Erb 2014; Eriksson 2009, and Juhl‐Olsen 2015. All but Juhl‐Olsen 2015 observed a lower median duration of mechanical ventilation in the levosimendan group compared to the placebo group (Cholley 2017: 17 (IQR 12 to 20) versus 18 (IQR 14 to 20) hours; Erb 2014: 20 (IQR 16 to 63) versus 45 (IQR 18 to 115) hours; Eriksson 2009: 21 (IQR 10 to 792) versus 23 (IQR 10 to 580) hours). Juhl‐Olsen 2015 reported a duration of mechanical ventilation of 7.93 (IQR 4.77 to 15.4) hours in the levosimendan group compared with 6.77 (IQR 4.82 to 13.27) hours in the placebo control group.

7.12. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 12: Secondary outcome: duration of mechanical ventilation (hours)

Proportion of weaning failure within 48 hours of extubation

There were five studies with 1233 participants reporting on the proportion of weaning failure within 48 hours of extubation (Anastasiadis 2016; Eriksson 2009; Levin 2012; Mehta 2017; Sharma 2014). Anastasiadis 2016 reported that none out of 16 participants in the levosimendan intervention group had weaning failure within 48 hours of extubation compared to 1 out of 16 in the control group (RR 0.33, 95% CI 0.01 to 7.62). Eriksson 2009 found that 8 out of 30 patients in the levosimendan group had weaning failure within 48 hours of extubation compared to 20 out of 30 in the control group (RR 0.40, 95% CI 0.21 to 0.76). Also, Levin 2012 reported that 3 out of 127 participants in the levosimendan group had weaning failure within 48 hours of extubation compared to 12 out of 125 in the control group (RR 0.25, 95% CI 0.07 to 0.85). In contrast, Mehta 2017 found that 1 out of 428 patients in the levosimendan group had weaning failure within 48 hours of extubation compared to none out of 421 in the control group (RR 2.95, 95% CI 0.12 to 72.24). Sharma 2014 reported that 2 out of 20 participants in the levosimendan group had weaning failure within 48 hours of extubation compared to 4 out of 20 in the control group (RR 0.50, 95% CI 0.10 to 2.43). Altogether, there was a difference in the proportion of weaning failure within 48 hours of extubation between groups, favouring the levosimendan group (RR 0.39, 95% CI 0.23 to 0.67; 1233 participants, 5 studies; low heterogeneity) (Analysis 7.13).

7.13. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 13: Secondary outcome: proportion of weaning failure within 48 hours after extubation

Number of patients requiring mechanical circulatory support (IABP)

Information on the number of patients requiring mechanical circulatory support was found in 10 studies with 1881 participants (Anastasiadis 2016; Cholley 2017; Erb 2014; Eriksson 2009; Kodalli 2013; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014). However, the relative effect was only generated from Anastasiadis 2016, Cholley 2017, Erb 2014, Eriksson 2009, Lahtinen 2011, Levin 2012, Mehta 2017, Shah 2014, and Sharma 2014, while the risk ratio from Kodalli 2013 was not estimable due to no events being observed in either group. Hu 2020 narratively reported a significantly longer IABP support in the placebo control group compared with the levosimendan intervention group without specifying the number of patients in need of this support.

There were 92 out of 944 participants in the levosimendan group requiring mechanical circulatory support compared to 136 out of 937 in the control group. When levosimendan was compared to the control group, there was a difference in the number of patients requiring mechanical circulatory support and this effect was favourable to levosimendan (RR 0.47, 95% CI 0.24 to 0.91; 1881 participants, 10 studies; low‐certainty evidence; substantial heterogeneity) (Analysis 7.14; Table 2).

7.14. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 14: Secondary outcome: number of participants requiring mechanical circulatory support

Number of patients requiring additional inotropic drugs

Need for overall additional inotropic drugs was reported by six studies (Cholley 2017; Levin 2012; Mehta 2017; Shah 2014; Tritapepe 2006; Tritapepe 2009), which found that 338 out of 811 patients required additional inotropic drugs in the intervention group with levosimendan compared to 433 out of 801 patients in the placebo control group. There was a difference in the risk ratio between groups favouring levosimendan (RR 0.49, 95% CI 0.28 to 0.88; 1612 participants, 6 studies; considerable heterogeneity) (Analysis 7.15).

7.15. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 15: Secondary outcome: number of participants requiring additional inotropic drugs

Within this outcome category, specific inotropic drugs, i.e. epinephrine, norepinephrine, dopamine, dobutamine, enoximone, milrinone, nitroprusside, ephedrine, and phenylephrine were also reported. Except for dobutamine, there was little to no difference in risk ratio between groups. Five studies, Cholley 2017, Erb 2014, Eriksson 2009, Kodalli 2013, and Lahtinen 2011, reported that 30 out of 328 patients in the levosimendan group required additional epinephrine compared to 47 out of 330 patients in the control group (RR 0.61, 95% CI 0.26 to 1.45; 658 participants, 5 studies; substantial heterogeneity) (Analysis 7.16). Regarding norepinephrine, Cholley 2017, Erb 2014, Eriksson 2009, Juhl‐Olsen 2015, Kodalli 2013, and Lahtinen 2011 reported that 166 out of 338 participants in the levosimendan group received this additional inotropic support compared to 117 out of 340 participants in the control group (RR 1.31, 95% CI 1.00 to 1.71; 678 participants, 6 studies; moderate heterogeneity) (Analysis 7.17). Narratively, both Jävelä 2008 and Leppikangas 2011 stated that the levosimendan group required more norepinephrine compared with the control group. Tritapepe 2009 reported that 9 out of 52 patients in the levosimendan group required additional dopamine compared to 17 out of 50 patients in the control group (RR 0.51, 95% CI 0.25 to 1.03; 102 participants, 1 study; heterogeneity not applicable) (Analysis 7.18). Four studies, Erb 2014, Juhl‐Olsen 2015, Lahtinen 2011, and Tritapepe 2006, stated that 23 out of 138 patients in the levosimendan group required additional dobutamine compared to 72 out of 139 patients in the control group and there was a difference in the risk ratio between groups favouring levosimendan (RR 0.33, 95% CI 0.22 to 0.50; 277 participants, 4 studies; low heterogeneity) (Analysis 7.19). Erb 2014 reported that 7 out of 17 patients in the levosimendan group required additional enoximone compared to 11 out of 16 patients in the control group (RR 0.60, 95% CI 0.31 to 1.16; 33 participants, 1 study; heterogeneity not applicable) (Analysis 7.20). Regarding milrinone, Eriksson 2009 and Lahtinen 2011 documented that 2 out of 129 participants in the levosimendan group received this additional inotropic support compared to 13 out of 131 participants in the control group (RR 0.22, 95% CI 0.03 to 1.41; 260 participants, 2 studies; low heterogeneity) (Analysis 7.21). Juhl‐Olsen 2015 reported that 1 out of 10 patients in the levosimendan group required additional nitroprusside compared to 6 out of 10 patients in the control group (RR 0.17, 95% CI 0.02 to 1.14; 20 participants, 1 study; heterogeneity not applicable) (Analysis 7.22). Jävelä 2008 narratively noted that the levosimendan group received less nitroprusside postoperatively compared with the control group. Cholley 2017 reported that 53 out of 167 patients in the levosimendan group required additional ephedrine compared to 48 out of 168 patients in the control group (RR 1.11, 95% CI 0.80 to 1.54; 335 participants, 1 study; heterogeneity not applicable) (Analysis 7.23). In addition, Cholley 2017 reported that 34 out of 167 patients in the levosimendan group required additional phenylephrine compared to 24 out of 168 patients in the control group (RR 1.43, 95% CI 0.88 to 2.30; 335 participants, 1 study; heterogeneity not applicable) (Analysis 7.24).

7.16. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 16: Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine

7.17. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 17: Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine

7.18. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 18: Secondary outcome: number of participants requiring additional inotropic drugs ‐ dopamine

7.19. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 19: Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine

7.20. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 20: Secondary outcome: number of participants requiring additional inotropic drugs ‐ enoximone

7.21. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 21: Secondary outcome: number of participants requiring additional inotropic drugs ‐ milrinone

7.22. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 22: Secondary outcome: number of participants requiring additional inotropic drugs ‐ nitroprusside

7.23. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 23: Secondary outcome: number of participants requiring additional inotropic drugs ‐ ephedrine

7.24. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 24: Secondary outcome: number of participants requiring additional inotropic drugs ‐ phenylephrine

Number of patients requiring cardiac transplantation, ventricular assist device (VAD) implantation or cardiopulmonary resuscitation

Information on the number of patients requiring cardiac transplantation or VAD implantation was restricted to Cholley 2017, Erb 2014, Kodalli 2013, and Levin 2012. However, the relative effect was only generated from Cholley 2017, Erb 2014, and Levin 2012, while the risk ratio from Kodalli 2013 was not estimable due to no events being observed in either group. There were none out of 326 patients in the levosimendan group requiring cardiac transplantation/VAD implantation compared to 8 out of 322 in the control group and the risk ratio between groups differed in this, favouring levosimendan (RR 0.16, 95% CI 0.03 to 0.91; 648 participants, 4 studies; low heterogeneity) (Analysis 7.25).

7.25. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 25: Secondary outcome: number of participants requiring cardiac transplantation

Information on the number of patients requiring cardiopulmonary resuscitation was restricted to Kodalli 2013 with 30 participants. However, the relative effect for the outcome was not estimable due to no events being observed in either group (Analysis 7.26).

7.26. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 26: Secondary outcome: number of participants requiring cardiopulmonary resuscitation

Proportion of renal failure

Nine studies reported this outcome (Anastasiadis 2016; Cholley 2017; Erb 2014; Kodalli 2013; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014). However, the relative effect was only generated from Anastasiadis 2016, Cholley 2017, Erb 2014, Lahtinen 2011, Levin 2012, Mehta 2017, Shah 2014, and Sharma 2014, while the risk ratio from Kodalli 2013 was not estimable due to no events being observed in either group. There were 43 out of 914 patients in the levosimendan group suffering renal failure compared to 61 out of 905 in the control group, with little to no difference in the risk ratio between groups (RR 0.71, 95% CI 0.43 to 1.16; 1819 participants, 9 studies; low heterogeneity) (Analysis 7.27).

7.27. Analysis.

Comparison 7: Comparison 7: Levosimendan versus placebo, Outcome 27: Secondary outcome: proportion of renal failure

Quality of life

Overall, eight studies gave information on quality of life (Erb 2014; Juhl‐Olsen 2015; Lahtinen 2011; Levin 2012; Mehta 2017; Shah 2014; Sharma 2014; Tritapepe 2009). Erb 2014 used the EuroQol 5‐D questionnaire and reported that according to this, quality of life was comparable between levosimendan and control groups six months after surgery (levosimendan group: 6.5 ± 2.1 versus placebo group: 7.4 ± 1.2). Similarly, Lahtinen 2011 used EuroQol 5‐D to assess the quality of life status and found that the quality of life was comparable between groups preoperatively. Six months after surgery, both groups had substantially improved quality of life compared to preoperative values with no difference reported between groups (no quantitative data given). Juhl‐Olsen 2015 stated that no patient developed headache or nausea. Levin 2012 narratively pointed out that there were no significant differences in adverse events (hypotension, headache, nausea and vomiting) during the 24‐hour preoperative infusion period between study groups. Mehta 2017 reported on reasons for premature permanent discontinuation of infusion, i.e. hypotension (26 of 428 in levosimendan group versus 21 of 421 in placebo group) and tachycardia or arrhythmia (2 of 428 in levosimendan group versus 3 of 421 in placebo group). Shah 2014 stated that the adverse events occurring during the 24 hours preoperative infusion were hypotension (7 of 25 in levosimendan group versus 2 of 25 in placebo group), need for vasopressors (2 of 25 in levosimendan group versus 1 of 25 in placebo group), headache (3 of 25 in levosimendan group, versus 1 of 25 in placebo group), nausea (2 of 25 in levosimendan group versus 0 of 25 in placebo group), and vomiting (2 of 25 in levosimendan group versus 0 of 25 in placebo group). Sharma 2014 pointed out that the adverse events occurring during the 24 hours preoperative infusion were hypotension (5 of 20 in levosimendan group versus 3 of 20 in placebo group), need for vasopressors (3 of 20 in levosimendan group versus 2 of 20 in placebo group), headache (2 of 20 in levosimendan group versus 1 of 20 in placebo group), and vomiting (1 of 20 in levosimendan group versus 2 of 20 in placebo group). Regarding quality of life, Tritapepe 2009 documented that 4 out of 52 patients in the levosimendan group developed hypotension.

Subgroup analyses

Due to the small number of studies in most comparisons, subgroup analyses were limited to levosimendan.

Regimen of the control group

We had planned to analyse the influence of the control group regimen (placebo‐controlled versus active control). However, only a minority of the included studies investigated levosimendan compared with an active control, so no interpretable results could be obtained based on the available data. There was only one feasible analysis that addressed the impact of control group regimens on atrial fibrillation, and no subgroup interaction was found (Analysis 8.1).

8.1. Analysis.

Comparison 8: Subgroup analysis: Regimen of the control group ‐ primary outcomes, Outcome 1: Adverse event: atrial fibrillation

Timing of the administration

Given the time points reported in the included studies, it was possible to perform subgroup analyses on all primary endpoints distinguishing between levosimendan administration before surgery (4 to 24 hours in advance) and in the operating room (at induction of anaesthesia, before skin incision or before initiation of the cardiopulmonary bypass machine). Subgroup analyses show evidence of a difference between the placebo group and the levosimendan group favouring levosimendan with respect to 30‐day mortality when the drug was given in the preoperative period (RR 0.30, 95% CI 0.14 to 0.67; Analysis 9.1). However, there was little to no difference between the placebo group and the levosimendan group when the drug was given in the operating room (RR 0.92, 95% CI 0.60 to 1.41; Analysis 9.1). No subgroup interaction was found regarding the incidence of LCOS or adverse events (Analysis 9.2; Analysis 9.3; Analysis 9.4; Analysis 9.5).

9.1. Analysis.

Comparison 9: Subgroup analysis: Timing of the drug administration ‐ primary outcomes, Outcome 1: All‐cause mortality

9.2. Analysis.

Comparison 9: Subgroup analysis: Timing of the drug administration ‐ primary outcomes, Outcome 2: Incidence of LCOS

9.3. Analysis.

Comparison 9: Subgroup analysis: Timing of the drug administration ‐ primary outcomes, Outcome 3: Adverse events: atrial fibrillation

9.4. Analysis.

Comparison 9: Subgroup analysis: Timing of the drug administration ‐ primary outcomes, Outcome 4: Adverse events: perioperative myocardial infarction

9.5. Analysis.

Comparison 9: Subgroup analysis: Timing of the drug administration ‐ primary outcomes, Outcome 5: Adverse events: non‐embolic stroke or transient ischaemic attack

Administration protocol of levosimendan

Levosimendan can be administered with or without an initial bolus. Therefore, subgroup analyses were performed to assess the potential impact of the administration protocol. These analyses revealed no subgroup interaction for any of the primary endpoints investigated (30‐day mortality, LCOS, adverse events) (Analysis 10.1; Analysis 10.2; Analysis 10.3; Analysis 10.4; Analysis 10.5).

10.1. Analysis.

Comparison 10: Subgroup analysis: Effect of the administration protocol ‐ primary outcomes, Outcome 1: All‐cause mortality

10.2. Analysis.

Comparison 10: Subgroup analysis: Effect of the administration protocol ‐ primary outcomes, Outcome 2: Incidence of LCOS

10.3. Analysis.

Comparison 10: Subgroup analysis: Effect of the administration protocol ‐ primary outcomes, Outcome 3: Adverse events: atrial fibrillation

10.4. Analysis.

Comparison 10: Subgroup analysis: Effect of the administration protocol ‐ primary outcomes, Outcome 4: Adverse events: Perioperative myocardial infarction

10.5. Analysis.

Comparison 10: Subgroup analysis: Effect of the administration protocol ‐ primary outcomes, Outcome 5: Adverse events: non‐embolic stroke or transient ischaemic attack

Type of cardiac surgery

We aimed to analyse the influence of the type of cardiac surgery performed. However, there were too many mixed surgical intervention studies to have any meaningful comparisons.

Participants with versus without pre‐existing LV dysfunction

We aimed to analyse the influence of the severity of the participants' disease (with or without pre‐existing LV dysfunction). However, no subgroup analysis was possible since virtually no events were reported in the limited studies that included participants without LV dysfunction.

Discussion

Summary of main results

This systematic review comprised 29 RCTs involving 3307 participants. Of these, 24 studies addressed levosimendan and compared its efficacy and safety with dobutamine, milrinone, standard cardiac care, or placebo. Another four studies estimated the efficacy and safety of milrinone compared with placebo; one of these studies was designed as a three‐arm intervention study that also evaluated amrinone. An individual study evaluated dopamine versus placebo.

In general, confidence in the results of analysed studies was reduced due to relevant risk of bias, imprecision, or inconsistency. Domains of concern encompassed inadequate methods of sequence generation (selection bias) and lack of blinding (performance and detection bias). The majority of trials were small with only a few included participants (high imprecision of results). An acknowledgement of funding by the pharmaceutical industry or missing conflict of interest statements occurred in 19 of 29 trials.

Our meta‐analyses showed that levosimendan may reduce the risk of LCOS and probably reduces all‐cause mortality compared to placebo but not compared to standard cardiac care or milrinone (Table 1; Table 2). Subgroup analyses revealed that the beneficial effects of levosimendan were predominantly observed in preoperative drug administration. Our meta‐analyses further indicated that levosimendan compared to placebo, dobutamine, or milrinone but not compared to standard cardiac care may shorten the length of ICU stay and the duration of mechanical ventilation (Table 1; Table 2), but the evidence is very uncertain. There was no conclusive evidence on the effect of dopamine compared to placebo on LCOS and duration of mechanical ventilation, or on the effect of milrinone compared to placebo on all‐cause mortality, length of ICU stay, and duration of mechanical ventilation. For all further comparisons, no data on all‐cause mortality, LCOS, length of ICU stay, or duration of mechanical ventilation were available. The risk of adverse events and the number of participants requiring mechanical circulatory support, if reported, did not clearly differ between study groups for all comparisons.

Overall completeness and applicability of evidence

The available evidence was not sufficient to fully address the objectives of the review. In clinical practice, there are a variety of inotropic drugs. In the identified studies, this diversity was not reflected. A majority of the studies examined levosimendan. An assessment of the efficacy and safety of other inotropic agents was thus not possible or only to a very limited extent. Further limitations related to study size and the availability of relevant outcomes. In most studies, the number of included participants was small. Not all outcomes pre‐specified in the protocol for this review article were reported in all studies, and in some cases, quantitative data were not available. For valid estimation of the efficacy and safety of the prophylactic use of inotropic agents to prevent LCOS and mortality during elective cardiac surgery, it is important to consider relevant influential factors. This comprises the regimen of the control group, the timing and, in the case of levosimendan, the type of inotropic administration, the type of cardiac surgery, and the severity of cardiac disease. However, due to the small number of studies and the lack of quantitative outcome data, the evidence base for the required subgroup analyses was limited. Therefore, the findings of the subgroup analyses, insofar as they were feasible at all, should be viewed with caution.

Quality of the evidence

We downgraded GRADE certainty assessment of RCT outcomes due to imprecision, risk of bias, and inconsistency. Overall, we assessed the certainty of evidence for our primary outcomes as moderate to low for all‐cause mortality, low to very low for incidence of LCOS, and high to very low for adverse events. With the exception of the comparison of levosimendan versus placebo, the criterion of optimal information size was not met for any of the interventions being evaluated. Accordingly, we downgraded the certainty of all outcomes in these comparisons for imprecision. For all‐cause mortality in the comparison of milrinone versus placebo, there was also imprecision due to methodological heterogeneity. We applied downgrading due to risk of bias to outcomes in the comparisons levosimendan versus milrinone, levosimendan versus standard cardiac care, levosimendan versus placebo, and dopamine versus placebo. We downgraded outcomes in the dopamine versus placebo comparison as reasons for exclusion of study participants related to true outcome. In the other comparisons, we downgraded outcomes due to inadequate methods of sequence generation and/or lack of blinding. For inconsistency, we downgraded the outcomes incidence of LCOS, adverse events: atrial fibrillation, length of ICU stay, and duration of mechanical ventilation in the comparisons of levosimendan versus dobutamine, levosimendan versus standard cardiac care, and levosimendan versus placebo.

Potential biases in the review process

Our review has limitations primarily related to a lack of data availability. Although we searched key databases and study registries in collaboration with Cochrane Heart, we cannot exclude the possibility that we may have missed relevant publications. We contacted corresponding authors of the included studies to obtain missing quantitative outcome data, if possible. However, we received an adequate response only in the case of one study (Baysal 2014). Bias may have resulted merely from the relatively small total number of studies available and participants included. For most of the outcomes, there were too few studies to create a funnel plot, so we were not able to assess the possibility of publication bias. The limited evidence base also had a negative impact on the feasibility of relevant subgroup analyses.

Agreements and disagreements with other studies or reviews

The prophylactic use of inotropic agents in adult patients undergoing cardiac surgery to prevent LCOS and associated mortality and morbidity has been evaluated in two meta‐analyses so far (Elbadawi 2018; Wang 2019a). Both review articles focused on the potential benefit of levosimendan. Meta‐analyses on the safety and efficacy of strictly prophylactic administration of other inotropic agents are not available.

Elbadawi 2018 included participants undergoing cardiac surgery (CABG and/or valve surgery) who received levosimendan before, during, or after surgery for prophylactic reasons. Only RCTs were included. There were no restrictions regarding the type of control group. A total of 16 studies with 2273 participants were analysed. No statistically significant differences were found between levosimendan and control groups with regard to 30‐day mortality, incidence of perioperative myocardial infarction or postoperative cerebrovascular events, or the need for additional inotropic agents, postoperative dialysis or mechanical assist devices. In contrast, the levosimendan group showed a significant reduction in the duration of ventilation time and ICU stay as well as the incidence of postoperative acute kidney disease on the one hand, and a significantly increased incidence of atrial fibrillation on the other. Concerning the primary outcome of the meta‐analysis, 30‐day mortality, a subgroup analysis was performed to examine the potential influence of pre‐existing LV dysfunction (LVEF < 40% versus LVEF > 40%), the regimen of the control group (placebo‐controlled versus active control), and the timing of administration (preoperative versus postoperative). Here, a subgroup interaction was found solely for the regimen of the control group. This suggested that a prophylactic administration of levosimendan was associated with improved 30‐day mortality compared to active control, but not compared to placebo control.

Wang 2019a focused on the prophylactic administration of levosimendan compared with any control in adult participants undergoing CABG. Twenty‐one RCTs with 1727 participants were included. The meta‐analysis showed a statistically significant reduction in mortality, the need for IABP and additional inotropic agents, and the incidence of atrial fibrillation, myocardial infarction or renal dysfunction in the levosimendan group compared to the control group. Furthermore, a statistically significant increased incidence of hypotension and associated need for vasoactive medication was found with levosimendan administration. Subgroup analyses showed that the beneficial effects of levosimendan were predominantly observed in isolated on‐pump CABG and preoperative bolus drug administration. Interestingly, the subgroup analysis also indicated that a reduction in mortality with levosimendan was limited to the comparison with placebo control.

Our meta‐analyses showed that levosimendan may reduce the risk of LCOS and and probably reduces associated all‐cause mortality as well as the need for IABP when compared to placebo, but not when compared to standard cardiac care, dobutamine, or milrinone. Our meta‐analyses further indicated that levosimendan compared to placebo, dobutamine, or milrinone but not compared to standard cardiac care may shorten the length of ICU stay and the duration of mechanical ventilation, but the evidence is very uncertain. We did not find a significant effect of levosimendan on the incidence of adverse events or the need for additional inotropic medication. Our data, as do the findings of Elbadawi 2018 and Wang 2019a, provide some indications of the possible efficacy and safety of prophylactic use of levosimendan in elective cardiac surgery. Consistent with Wang 2019a, our subgroup analyses showed that a favourable effect of prophylactic levosimendan on LCOS incidence and mortality was limited to preoperative administration. However, the quality and validity of the data are very poor. Given the limited evidence available, there is an unmet need for large‐scale, well‐designed randomised trials. Notably, future studies of levosimendan should be designed to determine the potential benefit in specific patient groups (with/without left ventricular dysfunction) and surgery types, as well as the optimal administration protocol.

There are two meta‐analyses in the scientific literature dealing with the prophylactic administration of levosimendan during heart surgery in children due to congenital heart disease (Hummel 2017; Wang 2020). Both studies found no effect on mortality rates. However, only a very limited number of RCTs and participants could be included in both reviews, so no firm conclusions were drawn by the study authors.

There are several meta‐analyses that investigated the safety and efficacy of the inotropic agents milrinone or levosimendan in cardiac surgery, though without distinguishing between prophylactic use and treatment of acute LCOS. Regarding milrinone, all but one meta‐analyses found no effect on mortality rates (Majure 2013; Ren 2020; Rong 2019; Ushio 2016; You 2016; Zangrillo 2012). Regarding levosimendan, the majority of reviews reported a reduction in mortality rates in the intervention group compared to the control (Harrison 2013; Landoni 2012; Lee 2017; Ng 2019; Oliveros 2019; Putzu 2018; Qiang 2018; Tena 2018; Wang 2018; Xing 2018; Zangrillo 2009; Zhou 2018). However, several of these meta‐analyses also stressed the limited quality of this evidence. In two meta‐analyses, a trial sequential analysis (TSA) was performed, stating that the effect was not reliable (Wang 2018; Xing 2018). In three meta‐analyses, a subgroup analysis regarding the trials' risk of bias was performed. In this analysis, the levosimendan effect was lost if only studies with high quality were included (Ng 2019; Oliveros 2019; Putzu 2018). According to Zhou 2018, the publication date could also play a role, since according to their subgroup analysis, the levosimendan effect could only be observed in studies conducted before the year 2015. Most authors of the above‐mentioned review articles thus concluded that there is not enough evidence of high quality to support the systematic use of levosimendan in cardiac surgery. The lack of high‐quality evidence is also evident in the present meta‐analysis. In future, special care should be taken to carefully differentiate the indications for the use of inotropics in cardiac surgery when evaluating them. A valid and meaningful assessment can only be achieved if prophylactic and acute use are considered separately.

Authors' conclusions

Implications for practice.

At present, the data regarding whether a (specific) pre‐operative inotropic drug therapy could affect mortality in haemodynamically unstable patients before cardiac surgery is not robust.

Most importantly, our meta‐analyses show that levosimendan probably reduces the risk of all‐cause mortality and may reduce low cardiac output state (LCOS) compared to placebo. However, when compared to either standard cardiac care, dobutamine, or milrinone, there was no benefit in regard to all‐cause mortality or LCOS. These data indicate that any of these strategies, including the use of levosimendan, might be similar in performance in pre‐operative treatment.

Compared to placebo or even to the other inotropic drugs evaluated, our analyses indicate that levosimendan may shorten the length of intensive care unit (ICU) stay and the duration of mechanical ventilation.

Our analysis did not compare combinations of the above‐mentioned treatment modalities and thus does not provide any data on the effects of such combined treatments.

Implications for research.

Considering the limited evidence derived from the present body of evidence, due to a generally high risk of bias and imprecision because of few events and small numbers of participants and trials, it should be emphasised that there remains a great need for large‐scale, well‐designed randomised controlled trials to precisely decipher different drug, dosing, and timing regimens in order to show significant changes in hard endpoints like mortality or safety.

Further prospective studies will be necessary to determine if levosimendan indeed proves to be superior in reducing ventilation time and ICU stay as compared to dopamine, milrinone, or placebo. In addition, the timing and duration of a pre‐operative inotropic drug therapy has to be evaluated in more detail and might hold the potential for further improvement using optimised protocols. Importantly, further analyses will also have to more clearly define which patient subgroups will benefit the most from a pre‐operative inotropic drug therapy and which parameters will be suitable to identify the respective patients.

As reported above, there were essential differences in baseline parameters and coexisting therapeutic interventions among the trials. Therefore, better comparability of baseline conditions, especially with regard to haemodynamic parameters and the management of inotropics (i.e. standardised protocols for titration and dosing), is necessary in future controlled, prospective studies.

History

Protocol first published: Issue 11, 2020

Appendices

Appendix 1. Search strategy

CENTRAL

#1 MeSH descriptor: [Cardiotonic Agents] explode all trees

#2 Cardiotonic*

#3 (Inotrope* or inotropic)

#4 Calcium sensitiser*

#5 MeSH descriptor: [Phosphodiesterase Inhibitors] explode all trees

#6 Phosphodiesterase inhibitor*

#7 Dobutamine

#8 MeSH descriptor: [Epinephrine] this term only

#9 Epinephrine

#10 Levosimendan

#11 Milrinone

#12 MeSH descriptor: [Dopamine] this term only

#13 dopamine

#14 adrenaline

#15 noradrenaline

#16 MeSH descriptor: [Norepinephrine] this term only

#17 norepinephrine

#18 dopexamine

#19 MeSH descriptor: [Enoximone] this term only

#20 enoximone

#21 MeSH descriptor: [Amrinone] this term only

#22 amrinone

#23 {OR #1‐#22}

#24 MeSH descriptor: [Thoracic Surgery] explode all trees

#25 MeSH descriptor: [Cardiac Surgical Procedures] explode all trees

#26 ((coronary or heart or cardio* or cardiac* or valve) NEAR/5 (surg* or operat* or graft* or bypass or plasty or replacement or transplant*))

#27 MeSH descriptor: [Coronary Artery Bypass] this term only

#28 CABG

#29 MeSH descriptor: [Cardiac Output, Low] this term only

#30 LCOS

#31 MeSH descriptor: [Shock, Cardiogenic] this term only

#32 Cardiogenic shock

#33 Haemodynamic instability

#34 MeSH descriptor: [Heart Arrest, Induced] this term only

#35 Cardioplegia

#36 (induce* NEAR/4 (cardiac or heart) NEAR/4 arrest)

#37 Myectomy

#38 Pericardiectom*

#39 {OR #24‐#38}

#40 #23 AND #39

MEDLINE Ovid

1 exp Cardiotonic Agents/

2 Cardiotonic*.tw.

3 (Inotrope* or inotropic).tw.

4 Calcium sensitiser*.tw.

5 exp Phosphodiesterase Inhibitors/

6 Phosphodiesterase inhibitor*.tw.

7 Dobutamine.tw.

8 Epinephrine/

9 Epinephrine.tw.

10 Levosimendan.tw.

11 Milrinone.tw.

12 Dopamine/

13 dopamine.tw.

14 adrenaline.tw.

15 noradrenaline.tw.

16 Norepinephrine/

17 norepinephrine.tw.

18 dopexamine.tw.

19 Enoximone/

20 enoximone.tw.

21 Amrinone/

22 amrinone.tw.

23 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22

24 exp Thoracic Surgery/

25 exp Cardiac Surgical Procedures/

26 ((coronary or heart or cardio* or cardiac* or valve) adj5 (surg* or operat* or graft* or bypass or plasty or replacement or transplant*)).tw.

27 Coronary Artery Bypass/

28 CABG.tw.

29 Cardiac Output, Low/

30 LCOS.tw.

31 Shock, Cardiogenic/

32 Cardiogenic shock.tw.

33 Haemodynamic instability.tw.

34 Heart Arrest, Induced/

35 Cardioplegia.tw.

36 (induce* adj4 (cardiac or heart) adj4 arrest).tw.

37 Myectomy.tw.

38 Pericardiectom*.tw.

39 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 or 33 or 34 or 35 or 36 or 37 or 38

40 23 and 39

41 randomized controlled trial.pt.

42 controlled clinical trial.pt.

43 randomized.ab.

44 placebo.ab.

45 clinical trials as topic.sh.

46 randomly.ab.

47 trial.ti.

48 41 or 42 or 43 or 44 or 45 or 46 or 47

49 exp animals/ not humans.sh.

50 48 not 49

51 40 and 50

Embase Ovid

1 exp cardiotonic agent/

2 Cardiotonic*.tw.

3 (Inotrope* or inotropic).tw.

4 Calcium sensitiser*.tw.

5 exp phosphodiesterase inhibitor/

6 Phosphodiesterase inhibitor*.tw.

7 dobutamine/

8 Dobutamine.tw.

9 epinephrine/

10 Epinephrine.tw.

11 levosimendan/

12 Levosimendan.tw.

13 milrinone/

14 Milrinone.tw.

15 dopamine/

16 dopamine.tw.

17 adrenaline.tw.

18 noradrenalin/

19 noradrenaline.tw.

20 norepinephrine.tw.

21 dopexamine/

22 dopexamine.tw.

23 enoximone/

24 enoximone.tw.

25 amrinone/

26 amrinone.tw.

27 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26

28 exp thorax surgery/

29 exp heart surgery/

30 ((coronary or heart or cardio* or cardiac* or valve) adj5 (surg* or operat* or graft* or bypass or plasty or replacement or transplant*)).tw.

31 coronary artery bypass graft/

32 CABG.tw.

33 forward heart failure/

34 LCOS.tw.

35 low cardiac output syndrome.tw.

36 cardiogenic shock/

37 Cardiogenic shock.tw.

38 Haemodynamic instability.tw.

39 Cardioplegia.tw.

40 (induce* adj4 (cardiac or heart) adj4 arrest).tw.

41 Myectomy.tw.

42 Pericardiectom*.tw.

43 28 or 29 or 30 or 31 or 32 or 33 or 34 or 35 or 36 or 37 or 38 or 39 or 40 or 41 or 42

44 27 and 43

45 random$.tw.

46 factorial$.tw.

47 crossover$.tw.

48 cross over$.tw.

49 cross‐over$.tw.

50 placebo$.tw.

51 (doubl$ adj blind$).tw.

52 (singl$ adj blind$).tw.

53 assign$.tw.

54 allocat$.tw.

55 volunteer$.tw.

56 crossover procedure/

57 double blind procedure/

58 randomized controlled trial/

59 single blind procedure/

60 45 or 46 or 47 or 48 or 49 or 50 or 51 or 52 or 53 or 54 or 55 or 56 or 57 or 58 or 59

61 (animal/ or nonhuman/) not human/

62 60 not 61

63 44 and 62

64 limit 63 to embase

CPCI‐S

# 17 #16 AND #15

# 16 TS=(random* or blind* or allocat* or assign* or trial* or placebo* or crossover* or cross‐over*)

# 15 #14 AND #6

# 14 #13 OR #12 OR #11 OR #10 OR #9 OR #8 OR #7

# 13 TS=Pericardiectom*

# 12 TS=Myectomy

# 11 TS=(induce* NEAR/4 (cardiac or heart) NEAR/4 arrest)

# 10 TS=Cardioplegia

# 9 TS=(cardiogenic shock or haemodynamic instability)

# 8 TS=(CABG or LCOS or low cardiac output syndrome)

# 7 TS=((coronary or heart or cardio* or cardiac* or valve) NEAR/5 (surg* or operat* or graft* or bypass or plasty or replacement or transplant*))

# 6 #5 OR #4 OR #3 OR #2 OR #1

# 5 TS=(Dobutamine or Epinephrine or Levosimendan or Milrinone or dopamine or adrenaline or noradrenaline or norepinephrine or dopexamine or enoximone or amrinone)

# 4 TS=Phosphodiesterase inhibitor*

# 3 TS=Calcium sensitiser*

# 2 TS=(Inotrope* or inotropic)

# 1 TS=Cardiotonic*

ClinicalTrials.gov

Study type: Interventional Studies (Clinical Trials)

Intervention/treatment: Inotropic agents

WHO ICTRP

Inotropic agents

Data and analyses

Comparison 1. Comparison 1: Amrinone versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Secondary outcome: cardiac index (postoperative nadir)	1	30	Mean Difference (IV, Random, 95% CI)	0.60 [0.33, 0.87]

Comparison 2. Comparison 2: Dopamine versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Primary outcome: incidence of LCOS	1	81	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
2.2 Primary outcome: adverse events ‐ atrial fibrillation	1	81	Risk Ratio (M‐H, Random, 95% CI)	0.73 [0.44, 1.22]
2.3 Primary outcome: adverse events ‐ perioperative myocardial infarction	1	81	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
2.4 Secondary outcome: length of hospital stay (days)	1	81	Mean Difference (IV, Random, 95% CI)	0.00 [‐1.54, 1.54]
2.5 Secondary outcome: mean arterial pressure (24 hours after surgery)	1	81	Mean Difference (IV, Random, 95% CI)	3.00 [‐1.36, 7.36]
2.6 Secondary outcome: duration of mechanical ventilation (hours)	1	81	Mean Difference (IV, Random, 95% CI)	0.00 [‐1.11, 1.11]
2.7 Secondary outcome: number of participants requiring additional inotropic drugs ‐ dopamine	1	81	Risk Ratio (M‐H, Random, 95% CI)	0.11 [0.01, 1.95]
2.8 Secondary outcome: proportion of renal failure	1	81	Risk Ratio (M‐H, Random, 95% CI)	0.11 [0.01, 1.95]

Comparison 3. Comparison 3: Milrinone versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Primary outcome: all‐cause mortality	2	120	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.06, 15.44]
3.2 Primary outcome: adverse events ‐ cardiogenic shock	1	80	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.06, 15.44]
3.3 Primary outcome: adverse events ‐ atrial fibrillation	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.12, 3.78]
3.4 Primary outcome: adverse events ‐ perioperative myocardial infarction	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.44 [0.15, 1.33]
3.5 Primary outcome: adverse events ‐ non‐embolic stroke or transient ischaemic attack	1	40	Risk Ratio (M‐H, Random, 95% CI)	0.33 [0.01, 7.72]
3.6 Secondary outcome: length of ICU stay (days)	1	80	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐0.40, 0.10]
3.7 Secondary outcome: cardiac index	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.7.1 End of surgery cardiac index	1	40	Mean Difference (IV, Random, 95% CI)	0.10 [‐0.22, 0.42]
3.7.2 Cardiac index 6 hours after surgery	2	70	Mean Difference (IV, Random, 95% CI)	0.50 [0.29, 0.71]
3.7.3 Cardiac index 12 hours after surgery	1	40	Mean Difference (IV, Random, 95% CI)	0.50 [0.25, 0.75]
3.8 Secondary outcome: mean arterial pressure	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.8.1 Baseline mean arterial pressure	1	124	Mean Difference (IV, Random, 95% CI)	2.00 [‐2.23, 6.23]
3.8.2 End of surgery mean arterial pressure	2	164	Mean Difference (IV, Random, 95% CI)	‐1.66 [‐5.43, 2.12]
3.8.3 Mean arterial pressure 6 hours after surgery	1	40	Mean Difference (IV, Random, 95% CI)	‐3.00 [‐9.82, 3.82]
3.8.4 Mean arterial pressure 12 hours after surgery	1	40	Mean Difference (IV, Random, 95% CI)	‐1.00 [‐8.19, 6.19]
3.9 Secondary outcome: PCWP	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.9.1 End of surgery PCWP	1	40	Mean Difference (IV, Random, 95% CI)	‐3.00 [‐6.74, 0.74]
3.9.2 PCWP 6 hours after surgery	1	40	Mean Difference (IV, Random, 95% CI)	‐4.00 [‐7.16, ‐0.84]
3.9.3 PCWP 12 hours after surgery	1	40	Mean Difference (IV, Random, 95% CI)	‐3.00 [‐8.00, 2.00]
3.10 Secondary outcome: duration of mechanical ventilation (hours)	1	80	Mean Difference (IV, Random, 95% CI)	‐3.98 [‐7.28, ‐0.68]
3.11 Secondary outcome: proportion of weaning failure within 48 hours of extubation	2	164	Risk Ratio (M‐H, Random, 95% CI)	1.17 [0.37, 3.69]
3.12 Secondary outcome: number of participants requiring mechanical circulatory support	2	204	Risk Ratio (M‐H, Random, 95% CI)	0.41 [0.13, 1.35]
3.13 Secondary outcome: number of participants requiring additional inotropic drugs	2	204	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.65, 1.31]
3.14 Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine	1	124	Risk Ratio (M‐H, Random, 95% CI)	1.24 [0.49, 3.13]
3.15 Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine	1	124	Risk Ratio (M‐H, Random, 95% CI)	0.97 [0.39, 2.42]
3.16 Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine	1	124	Risk Ratio (M‐H, Random, 95% CI)	1.94 [0.18, 20.81]
3.17 Secondary outcome: number of participants requiring additional inotropic drugs ‐ phenylephrine	1	124	Risk Ratio (M‐H, Random, 95% CI)	0.32 [0.01, 7.78]
3.18 Secondary outcome: proportion of renal failure	2	120	Risk Ratio (M‐H, Random, 95% CI)	0.40 [0.08, 2.01]

Comparison 4. Comparison 4: Levosimendan versus dobutamine.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Primary outcome: adverse events ‐ atrial fibrillation	2	140	Risk Ratio (M‐H, Random, 95% CI)	0.53 [0.16, 1.75]
4.2 Secondary outcome: length of hospital stay (days)	1	80	Mean Difference (IV, Random, 95% CI)	‐0.97 [‐1.28, ‐0.66]
4.3 Secondary outcome: length of ICU stay (days)	2	140	Mean Difference (IV, Random, 95% CI)	‐0.29 [‐0.51, ‐0.08]
4.4 Secondary analysis: cardiac index	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
4.4.1 Baseline cardiac index	1	60	Mean Difference (IV, Random, 95% CI)	0.00 [‐0.08, 0.08]
4.4.2 Cardiac index after 6 hours	1	60	Mean Difference (IV, Random, 95% CI)	‐0.01 [‐0.15, 0.13]
4.4.3 Cardiac index after 12 hours	1	60	Mean Difference (IV, Random, 95% CI)	0.09 [‐0.02, 0.20]
4.4.4 Cardiac index after 24 hours	1	60	Mean Difference (IV, Random, 95% CI)	0.46 [0.35, 0.57]
4.5 Secondary outcome: mean arterial pressure	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
4.5.1 Baseline mean arterial pressure	2	140	Mean Difference (IV, Random, 95% CI)	‐0.06 [‐2.11, 2.00]
4.5.2 Mean arterial pressure after 6 hours	2	140	Mean Difference (IV, Random, 95% CI)	‐9.27 [‐14.18, ‐4.36]
4.5.3 Mean arterial pressure after 12 hours	2	140	Mean Difference (IV, Random, 95% CI)	‐9.17 [‐10.85, ‐7.50]
4.5.4 Mean arterial pressure after 24 hours	2	140	Mean Difference (IV, Random, 95% CI)	‐5.42 [‐12.52, 1.68]
4.6 Secondary analysis: PCWP	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
4.6.1 Baseline PCWP	1	80	Mean Difference (IV, Random, 95% CI)	0.63 [‐0.07, 1.33]
4.6.2 PCWP after 6 hours	1	80	Mean Difference (IV, Random, 95% CI)	‐2.47 [‐3.04, ‐1.90]
4.6.3 PCWP after 12 hours	1	80	Mean Difference (IV, Random, 95% CI)	‐2.42 [‐2.90, ‐1.94]
4.6.4 PCWP after 24 hours	1	80	Mean Difference (IV, Random, 95% CI)	‐1.45 [‐1.92, ‐0.98]
4.7 Secondary outcome: duration of mechanical ventilation (hours)	2	140	Mean Difference (IV, Random, 95% CI)	‐1.04 [‐1.72, ‐0.35]
4.8 Secondary outcome: number of participants requiring mechanical circulatory support	1	80	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.06, 15.44]
4.9 Secondary outcome: number of participants requiring additional inotropic drugs	1	60	Risk Ratio (M‐H, Random, 95% CI)	5.25 [2.05, 13.47]
4.10 Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine	2	140	Risk Ratio (M‐H, Random, 95% CI)	1.63 [0.32, 8.23]
4.11 Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine	2	140	Risk Ratio (M‐H, Random, 95% CI)	3.52 [1.21, 10.24]

Comparison 5. Comparison 5: Levosimendan versus standard cardiac care.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Primary outcome: all‐cause mortality	3	208	Risk Ratio (M‐H, Random, 95% CI)	0.37 [0.13, 1.04]
5.2 Primary outcome: incidence of LCOS	3	208	Risk Ratio (M‐H, Random, 95% CI)	0.49 [0.14, 1.73]
5.3 Primary outcome: adverse events ‐ cardiogenic shock	1	128	Risk Ratio (M‐H, Random, 95% CI)	0.62 [0.22, 1.81]
5.4 Primary outcome: adverse events ‐ atrial fibrillation	2	188	Risk Ratio (M‐H, Random, 95% CI)	0.40 [0.11, 1.41]
5.5 Primary outcome: adverse events ‐ perioperative myocardial infarction	1	128	Risk Ratio (M‐H, Random, 95% CI)	0.62 [0.22, 1.81]
5.6 Primary outcome: adverse events ‐ non‐embolic stroke or transient ischaemic attack	1	128	Risk Ratio (M‐H, Random, 95% CI)	0.56 [0.27, 1.18]
5.7 Secondary outcome: length of hospital stay (days)	2	80	Mean Difference (IV, Random, 95% CI)	0.70 [‐1.50, 2.91]
5.8 Secondary outcome: length of ICU stay (days)	2	80	Mean Difference (IV, Random, 95% CI)	0.33 [‐1.16, 1.83]
5.9 Secondary outcome: cardiac index	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
5.9.1 Baseline cardiac index	3	208	Mean Difference (IV, Random, 95% CI)	0.00 [‐0.10, 0.10]
5.9.2 End‐of‐surgery cardiac index	2	188	Mean Difference (IV, Random, 95% CI)	0.52 [0.32, 0.73]
5.9.3 Cardiac index 6 hours after surgery	2	188	Mean Difference (IV, Random, 95% CI)	0.77 [0.64, 0.89]
5.9.4 Cardiac index 12 hours after surgery	1	60	Mean Difference (IV, Random, 95% CI)	0.55 [0.34, 0.76]
5.9.5 Cardiac index 24 hours after surgery	2	188	Mean Difference (IV, Random, 95% CI)	0.47 [0.35, 0.58]
5.10 Secondary outcome: mean arterial pressure	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
5.10.1 Baseline mean arterial pressure	2	188	Mean Difference (IV, Random, 95% CI)	0.41 [‐0.78, 1.59]
5.10.2 End of surgery mean arterial pressure	2	188	Mean Difference (IV, Random, 95% CI)	2.39 [‐8.74, 13.52]
5.10.3 Mean arterial pressure 6 hours after surgery	2	188	Mean Difference (IV, Random, 95% CI)	‐1.11 [‐2.36, 0.14]
5.10.4 Mean arterial pressure 12 hours after surgery	1	60	Mean Difference (IV, Random, 95% CI)	4.07 [1.03, 7.11]
5.10.5 Mean arterial pressure 24 hours after surgery	2	188	Mean Difference (IV, Random, 95% CI)	3.54 [‐0.06, 7.15]
5.11 Secondary outcome: PCWP	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
5.11.1 Baseline PCWP	1	60	Mean Difference (IV, Random, 95% CI)	0.50 [‐0.63, 1.63]
5.11.2 End of surgery PCWP	1	60	Mean Difference (IV, Random, 95% CI)	‐2.07 [‐3.06, ‐1.08]
5.11.3 PCWP 6 hours after surgery	1	60	Mean Difference (IV, Random, 95% CI)	‐1.21 [‐2.22, ‐0.20]
5.11.4 PCWP 12 hours after surgery	1	60	Mean Difference (IV, Random, 95% CI)	‐1.04 [‐1.69, ‐0.39]
5.11.5 PCWP 24 hours after surgery	1	60	Mean Difference (IV, Random, 95% CI)	‐0.73 [‐1.48, 0.02]
5.12 Secondary outcome: duration of mechanical ventilation (hours)	1	128	Mean Difference (IV, Random, 95% CI)	‐3.40 [‐11.50, 4.70]
5.13 Secondary outcome: proportion of weaning failure within 48 hours of extubation	1	128	Risk Ratio (M‐H, Random, 95% CI)	0.82 [0.36, 1.84]
5.14 Secondary outcome: number of participants requiring mechanical circulatory support	3	208	Risk Ratio (M‐H, Random, 95% CI)	0.88 [0.50, 1.55]
5.15 Secondary outcome: number of participants requiring additional inotropic drugs	2	80	Risk Ratio (M‐H, Random, 95% CI)	1.70 [1.13, 2.57]
5.16 Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine	1	128	Risk Ratio (M‐H, Random, 95% CI)	0.70 [0.41, 1.19]
5.17 Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine	2	188	Risk Ratio (M‐H, Random, 95% CI)	1.18 [0.58, 2.37]
5.18 Secondary outcome: number of participants requiring additional inotropic drugs ‐ dopamine	1	128	Risk Ratio (M‐H, Random, 95% CI)	0.78 [0.56, 1.06]
5.19 Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine	1	128	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.89, 1.20]
5.20 Secondary outcome: number of participants requiring cardiac transplantation	1	128	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
5.21 Secondary outcome: number of participants requiring cardiopulmonary resuscitation	1	128	Risk Ratio (M‐H, Random, 95% CI)	0.40 [0.13, 1.21]
5.22 Secondary outcome: proportion of renal failure	3	208	Risk Ratio (M‐H, Random, 95% CI)	0.48 [0.22, 1.08]

Comparison 6. Comparison 6: Levosimendan versus milrinone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Primary outcome: all‐cause mortality	2	70	Risk Ratio (M‐H, Random, 95% CI)	0.14 [0.01, 2.55]
6.2 Primary outcome: adverse events ‐ atrial fibrillation	1	30	Risk Ratio (M‐H, Random, 95% CI)	0.86 [0.38, 1.95]
6.3 Primary outcome: adverse events ‐ perioperative myocardial infarction	1	30	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
6.4 Secondary outcome: length of ICU stay (days)	2	100	Mean Difference (IV, Random, 95% CI)	‐0.69 [‐1.31, ‐0.07]
6.5 Secondary outcome: mean arterial pressure	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
6.5.1 Baseline mean arterial pressure	1	30	Mean Difference (IV, Random, 95% CI)	2.00 [‐4.60, 8.60]
6.5.2 End of surgery mean arterial pressure	1	30	Mean Difference (IV, Random, 95% CI)	3.00 [‐1.53, 7.53]
6.5.3 Mean arterial pressure 6 hours after surgery	1	30	Mean Difference (IV, Random, 95% CI)	‐3.00 [‐10.16, 4.16]
6.5.4 Mean arterial pressure 12 hours after surgery	1	30	Mean Difference (IV, Random, 95% CI)	0.00 [‐6.48, 6.48]
6.5.5 Mean arterial pressure 24 hours after surgery	1	30	Mean Difference (IV, Random, 95% CI)	‐2.00 [‐7.01, 3.01]
6.6 Secondary outcome: duration of mechanical ventilation (hours)	3	130	Mean Difference (IV, Random, 95% CI)	‐7.06 [‐10.19, ‐3.93]
6.7 Secondary outcome: number of participants requiring mechanical circulatory support	2	90	Risk Ratio (M‐H, Random, 95% CI)	0.58 [0.09, 3.55]
6.8 Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine	1	30	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.88, 1.13]
6.9 Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine	1	30	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.88, 1.13]

Comparison 7. Comparison 7: Levosimendan versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
7.1 Primary outcome: all‐cause mortality	14	2347	Risk Ratio (M‐H, Random, 95% CI)	0.65 [0.43, 0.97]
7.2 Primary outcome: incidence of LCOS	6	1724	Risk Ratio (M‐H, Random, 95% CI)	0.43 [0.25, 0.74]
7.3 Primary outcome: adverse events ‐ cardiogenic shock	3	1212	Risk Ratio (M‐H, Random, 95% CI)	0.65 [0.40, 1.05]
7.4 Primary outcome: adverse events ‐ atrial fibrillation	11	1934	Risk Ratio (M‐H, Random, 95% CI)	1.02 [0.82, 1.27]
7.5 Primary outcome: adverse events ‐ perioperative myocardial infarction	8	1838	Risk Ratio (M‐H, Random, 95% CI)	0.89 [0.61, 1.31]
7.6 Primary outcome: adverse events ‐ non‐embolic stroke or transient ischaemic attack	8	1786	Risk Ratio (M‐H, Random, 95% CI)	0.89 [0.58, 1.38]
7.7 Secondary outcome: length of hospital stay (days)	6	422	Mean Difference (IV, Random, 95% CI)	‐1.80 [‐3.27, ‐0.33]
7.8 Secondary outcome: length of ICU stay (days)	7	572	Mean Difference (IV, Random, 95% CI)	‐1.00 [‐1.63, ‐0.37]
7.9 Secondary outcome: cardiac index	5		Mean Difference (IV, Random, 95% CI)	Subtotals only
7.9.1 Baseline cardiac index	5	456	Mean Difference (IV, Random, 95% CI)	‐0.04 [‐0.17, 0.10]
7.9.2 End of surgery cardiac index	2	74	Mean Difference (IV, Random, 95% CI)	0.67 [‐0.20, 1.53]
7.9.3 Cardiac index 6 hours after surgery	2	54	Mean Difference (IV, Random, 95% CI)	0.68 [0.28, 1.09]
7.9.4 Cardiac index 12 hours after surgery	3	104	Mean Difference (IV, Random, 95% CI)	0.73 [0.23, 1.23]
7.9.5 Cardiac index 24 hours after surgery	5	456	Mean Difference (IV, Random, 95% CI)	0.51 [0.06, 0.96]
7.10 Secondary outcome: mean arterial pressure	5		Mean Difference (IV, Random, 95% CI)	Subtotals only
7.10.1 Baseline mean arterial pressure	4	146	Mean Difference (IV, Random, 95% CI)	2.91 [‐3.19, 9.01]
7.10.2 End of surgery mean arterial pressure	3	404	Mean Difference (IV, Random, 95% CI)	‐2.06 [‐6.50, 2.38]
7.10.3 Mean arterial pressure 6 hours after surgery	3	374	Mean Difference (IV, Random, 95% CI)	‐4.81 [‐6.08, ‐3.55]
7.10.4 Mean arterial pressure 12 hours after surgery	3	374	Mean Difference (IV, Random, 95% CI)	‐4.57 [‐9.19, 0.05]
7.10.5 Mean arterial pressure 24 hours after surgery	5	466	Mean Difference (IV, Random, 95% CI)	‐2.19 [‐5.84, 1.46]
7.11 Secondary outcome: PCWP	5		Mean Difference (IV, Random, 95% CI)	Subtotals only
7.11.1 Baseline PCWP	5	196	Mean Difference (IV, Random, 95% CI)	0.11 [‐1.02, 1.24]
7.11.2 End of surgery PCWP	3	134	Mean Difference (IV, Random, 95% CI)	‐1.63 [‐3.85, 0.58]
7.11.3 PCWP 6 hours after surgery	2	54	Mean Difference (IV, Random, 95% CI)	0.31 [‐0.82, 1.44]
7.11.4 PCWP 12 hours after surgery	3	104	Mean Difference (IV, Random, 95% CI)	‐1.35 [‐4.05, 1.36]
7.11.5 PCWP 24 hours after surgery	5	196	Mean Difference (IV, Random, 95% CI)	‐1.33 [‐2.50, ‐0.17]
7.12 Secondary outcome: duration of mechanical ventilation (hours)	7	572	Mean Difference (IV, Random, 95% CI)	‐8.03 [‐13.17, ‐2.90]
7.13 Secondary outcome: proportion of weaning failure within 48 hours after extubation	5	1233	Risk Ratio (M‐H, Random, 95% CI)	0.39 [0.23, 0.67]
7.14 Secondary outcome: number of participants requiring mechanical circulatory support	10	1881	Risk Ratio (M‐H, Random, 95% CI)	0.47 [0.24, 0.91]
7.15 Secondary outcome: number of participants requiring additional inotropic drugs	6	1612	Risk Ratio (M‐H, Random, 95% CI)	0.49 [0.28, 0.88]
7.16 Secondary outcome: number of participants requiring additional inotropic drugs ‐ epinephrine	5	658	Risk Ratio (M‐H, Random, 95% CI)	0.61 [0.26, 1.45]
7.17 Secondary outcome: number of participants requiring additional inotropic drugs ‐ norepinephrine	6	678	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.00, 1.71]
7.18 Secondary outcome: number of participants requiring additional inotropic drugs ‐ dopamine	1	102	Risk Ratio (M‐H, Random, 95% CI)	0.51 [0.25, 1.03]
7.19 Secondary outcome: number of participants requiring additional inotropic drugs ‐ dobutamine	4	277	Risk Ratio (M‐H, Random, 95% CI)	0.33 [0.22, 0.50]
7.20 Secondary outcome: number of participants requiring additional inotropic drugs ‐ enoximone	1	33	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.31, 1.16]
7.21 Secondary outcome: number of participants requiring additional inotropic drugs ‐ milrinone	2	260	Risk Ratio (M‐H, Random, 95% CI)	0.22 [0.03, 1.41]
7.22 Secondary outcome: number of participants requiring additional inotropic drugs ‐ nitroprusside	1	20	Risk Ratio (M‐H, Random, 95% CI)	0.17 [0.02, 1.14]
7.23 Secondary outcome: number of participants requiring additional inotropic drugs ‐ ephedrine	1	335	Risk Ratio (M‐H, Random, 95% CI)	1.11 [0.80, 1.54]
7.24 Secondary outcome: number of participants requiring additional inotropic drugs ‐ phenylephrine	1	335	Risk Ratio (M‐H, Random, 95% CI)	1.43 [0.88, 2.30]
7.25 Secondary outcome: number of participants requiring cardiac transplantation	4	648	Risk Ratio (M‐H, Random, 95% CI)	0.16 [0.03, 0.91]
7.26 Secondary outcome: number of participants requiring cardiopulmonary resuscitation	1	30	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
7.27 Secondary outcome: proportion of renal failure	9	1819	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.43, 1.16]

Comparison 8. Subgroup analysis: Regimen of the control group ‐ primary outcomes.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
8.1 Adverse event: atrial fibrillation	16		Risk Ratio (IV, Random, 95% CI)	Subtotals only
8.1.1 Levosimendan vs active controlled	3	170	Risk Ratio (IV, Random, 95% CI)	0.70 [0.38, 1.30]
8.1.2 Levosimendan vs passive controlled	13	2122	Risk Ratio (IV, Random, 95% CI)	0.95 [0.75, 1.19]

Comparison 9. Subgroup analysis: Timing of the drug administration ‐ primary outcomes.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
9.1 All‐cause mortality	13		Risk Ratio (IV, Random, 95% CI)	Subtotals only
9.1.1 Preoperative period	5	394	Risk Ratio (IV, Random, 95% CI)	0.30 [0.14, 0.67]
9.1.2 Perioperative/anaesthesia induction period	8	1633	Risk Ratio (IV, Random, 95% CI)	0.92 [0.60, 1.41]
9.2 Incidence of LCOS	6		Risk Ratio (IV, Random, 95% CI)	Subtotals only
9.2.1 Preoperative period	3	342	Risk Ratio (IV, Random, 95% CI)	0.39 [0.22, 0.70]
9.2.2 Perioperative/anaesthesia induction period	3	1382	Risk Ratio (IV, Random, 95% CI)	0.43 [0.17, 1.07]
9.3 Adverse events: atrial fibrillation	11		Risk Ratio (IV, Random, 95% CI)	Subtotals only
9.3.1 Preoperative period	5	394	Risk Ratio (IV, Random, 95% CI)	0.77 [0.37, 1.58]
9.3.2 Perioperative/anaesthesia induction period	6	1540	Risk Ratio (IV, Random, 95% CI)	1.13 [1.02, 1.27]
9.4 Adverse events: perioperative myocardial infarction	8		Risk Ratio (IV, Random, 95% CI)	Subtotals only
9.4.1 Preoperative period	3	324	Risk Ratio (IV, Random, 95% CI)	0.29 [0.04, 2.19]
9.4.2 Perioperative/anaesthesia induction period	5	1514	Risk Ratio (IV, Random, 95% CI)	0.99 [0.75, 1.31]
9.5 Adverse events: non‐embolic stroke or transient ischaemic attack	8		Risk Ratio (IV, Random, 95% CI)	Subtotals only
9.5.1 Preoperative period	4	374	Risk Ratio (IV, Random, 95% CI)	0.79 [0.34, 1.79]
9.5.2 Perioperative/anaesthesia induction period	4	1412	Risk Ratio (IV, Random, 95% CI)	0.83 [0.34, 2.08]

Comparison 10. Subgroup analysis: Effect of the administration protocol ‐ primary outcomes.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
10.1 All‐cause mortality	19		Risk Ratio (IV, Random, 95% CI)	Subtotals only
10.1.1 With bolus	7	802	Risk Ratio (IV, Random, 95% CI)	0.51 [0.26, 1.00]
10.1.2 Without bolus	12	1823	Risk Ratio (IV, Random, 95% CI)	0.68 [0.44, 1.04]
10.2 Incidence of LCOS	9		Risk Ratio (IV, Random, 95% CI)	Subtotals only
10.2.1 With bolus	4	600	Risk Ratio (IV, Random, 95% CI)	0.39 [0.21, 0.75]
10.2.2 Without bolus	5	1332	Risk Ratio (IV, Random, 95% CI)	0.67 [0.52, 0.86]
10.3 Adverse events: atrial fibrillation	16		Risk Ratio (IV, Random, 95% CI)	Subtotals only
10.3.1 With bolus	5	706	Risk Ratio (IV, Random, 95% CI)	0.78 [0.49, 1.24]
10.3.2 Without bolus	11	1586	Risk Ratio (IV, Random, 95% CI)	0.98 [0.74, 1.30]
10.4 Adverse events: Perioperative myocardial infarction	10		Risk Ratio (IV, Random, 95% CI)	Subtotals only
10.4.1 With bolus	4	682	Risk Ratio (IV, Random, 95% CI)	0.65 [0.34, 1.26]
10.4.2 Without bolus	6	1314	Risk Ratio (IV, Random, 95% CI)	1.03 [0.75, 1.40]
10.5 Adverse events: non‐embolic stroke or transient ischaemic attack	9		Risk Ratio (IV, Random, 95% CI)	Subtotals only
10.5.1 With bolus	3	580	Risk Ratio (IV, Random, 95% CI)	0.69 [0.43, 1.12]
10.5.2 Without bolus	6	1334	Risk Ratio (IV, Random, 95% CI)	0.79 [0.34, 1.84]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Amin 2019.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: length of stay in ICU Number of study centres: 1 Location of study centres: Egypt Period of study: preoperative until 24 hours post‐surgery
Participants	N randomised: 60 N lost to follow‐up or withdrawn: 0 N analysed: 60 Mean age (years): intervention group: 56.7 ± 11.4; comparison group: 54.4 ± 14.6 Gender (male/female): intervention group: 22/8; comparison group: 18/12 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective OPCABG Inclusion criteria: age 40 to 70 years; LVEF ≤ 40%; mean pulmonary artery pressure 25 to 35 mmHg; NYHA III or IV Exclusion criteria: age > 70 years; renal/hepatic dysfunction; perioperative surgical complications unrelated to the study; associated mitral valve stenosis; aortic valve stenosis
Interventions	Intervention: levosimendan; 0.1 µg/kg/min started with the induction of anaesthesia until transfer to the ICU Comparison: milrinone; 0.5 µg/kg/min started with the induction of anaesthesia until transfer to the ICU Concomitant medication: norepinephrine Excluded medications: ACE inhibitors; aspirin
Outcomes	Primary: haemodynamics; need for IABP or additional inotropic support; length of ventilation/ICU stay Secondary: no information given Time points reported: preoperative; 0/1/2/3/4 hours after starting medication; 2/4/6/12/24 hours after admission to ICU; on discharge from ICU
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	No information provided
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information provided
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Unclear risk	Baseline difference in gender

Anastasiadis 2016.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: March 2011 to March 2014 Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: Greece Period of study: preoperatively until 24 hours post‐surgery
Participants	N randomised: 32 N lost to follow‐up or withdrawn: 0 N analysed: 32 Mean age (years): intervention group: 61.1 ± 9.4; comparison group: 62.2 ± 11 Gender (male/female): intervention group: 14/2; comparison group: 16/0 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG Inclusion criteria: coronary artery disease; age < 80 years; LVEF ≤ 40% Exclusion criteria: emergency/redo surgery; recent myocardial infarction (< 14 days before surgery); critical preoperative state (inotropic support, IABP); concomitant valve surgery other than mitral valve repair; acute or chronic renal dysfunction; history of malignant cardiac arrhythmias; inability of the patient to understand the protocol and to cooperate postoperatively
Interventions	Intervention: levosimendan; 0.1 µg/kg/min for 24 hours started 24 hours before surgery Comparison: placebo Concomitant medication: epinephrine; norepinephrine Excluded medications: no information given
Outcomes	Primary: change in LVEF on day 7 post‐surgery Secondary: haemodynamics; length of inotropic support/ventilation/ICU stay/hospital stay; need for IABP; incidence of myocardial infarction/renal failure/cerebrovascular event/mortality Time points reported: at the start/end of medication; before transfer to operation room; 2/24 hours after admission to ICU; postoperative days 7/30
Notes	Funding for trial: Supported by Orion Pharma (Finland). Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Baysal 2014.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: July 2009 to January 2013 Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: Turkey Period of study: preoperative until 10 days post‐surgery
Participants	N randomised: 128 N lost to follow‐up or withdrawn: 0 N analysed: 128 Mean age (years): intervention group: 56.73 ± 11.71; comparison group: 58.41 ± 9.83 Gender (male/female): intervention group: 22/42; comparison group: 39/25 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective valve surgery Inclusion criteria: mitral valve insufficiency with or without coronary artery disease; LVEF ≤ 45% Exclusion criteria: emergency/redo surgery; recent myocardial infarction (< 1 month before surgery); unstable angina; diabetes mellitus treated with insulin; clinical findings of acute renal dysfunction; severe hepatic disease; severe chronic obstructive pulmonary disease; inotropic support before surgery; aortic valvular disease; infective endocarditis
Interventions	Intervention: levosimendan; loading dose of 6 µg/kg within 10 min after removal of the cross‐clamp followed by 0.1 µg/kg/min for 24 hours Comparison: standard inotropic agents (1 or more of dobutamine, epinephrine, norepinephrine) Concomitant medication: dobutamine, epinephrine, norepinephrine Excluded medications: other inotropic support
Outcomes	Primary: postoperative renal function measured by serum creatinine and estimated glomerular filtration rate Secondary: length of aortic cross‐clamp/CBP/ventilation/ICU stay/hospital stay; need for IABP or additional inotropic support; incidence of pneumonia/myocardial infarction/cerebrovascular event/atrial fibrillation and other rhythm disturbancies; need for renal replacement therapy; reoperation secondary to bleeding Time points reported: preoperative; 15 min after separation from CPB; at the end of surgery; 6/24 hours after admission to ICU; postoperative days 1/3/10/30
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Sequentially numbered assignment
Allocation concealment (selection bias)	Low risk	Sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of participants; no blinding of caregivers
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of observers
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Unclear risk	Baseline difference in gender

Cholley 2017.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: June 2013 to May 2015 Total duration of follow‐up: 6 months post‐surgery Number of study centres: 13 Location of study centres: France Period of study: preoperative until discharge from ICU
Participants	N randomised: 336 N lost to follow‐up or withdrawn: 8 N analysed: 314 (per‐protocol‐analysis); 335 (ITT analysis) Mean age (years): intervention group: 69 ± 10; comparison group: 67 ± 10 Gender (male/female): intervention group: 139/28; comparison group: 143/25 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG with or without valve surgery Inclusion criteria: LVEF ≤ 40% Exclusion criteria: emergency surgery; isolated valve surgery; age < 18 years; pregnancy; renal/hepatic dysfunction; hypotension; tachycardia
Interventions	Intervention: levosimendan 0.1 µg/kg/min for 24 hours started with the induction of anaesthesia Comparison: placebo Concomitant medication: catecholamines (dobutamine, epinephrine, milrinone, norepinephrine) Excluded medications: no information given
Outcomes	Primary: 3‐component composite reflecting LCOS: catecholamine infusion persisting beyond 48 hours after initiation of the study drug, need for IABP in the postoperative period (patients who had IABP preventively, met the outcome measure if they were not weaned from IABP within 96 hours after initiation of the study drug), need for renal replacement therapy during ICU stay Secondary: incidence of mortality in‐hospital or on postoperative days 28 and 180; each single component of the primary endpoint; length of IABP/inotropic support/ventilation/renal replacement therapy/ICU stay/hospital stay; number of renal replacement therapy kits used for each patient; number of patients out of ICU/hospital at postoperative day 28; incidence of adverse events (hypotension, arrhythmias, conduction disturbances, myocardial damage reflected by troponin value); need for fluids/vasopressors to correct hypotension Time points reported: preoperative; postoperative days 28/180
Notes	Funding for trial: Orion Pharma provided study treatments without financial compensation. Supported by the French Ministry of Health (Programme Hospitalier de Recherche Clinique national 2011, MIN02‐07) registered as P110138 and by Assistance Publique ‐ Hopitaux de Paris (AP‐HP; study sponsor). Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Minimisation
Allocation concealment (selection bias)	Low risk	Secured access to the randomisation algorithm; identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	Balanced missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

De Hert 2007.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: Belgium Period of study: preoperatively until 48 hours post‐surgery
Participants	N randomised: 30 N lost to follow‐up or withdrawn: 0 N analysed: 30 Mean age (years): intervention group: 67 ± 11; comparison group: 69 ± 10 Gender (male/female): intervention group: 10/5; comparison group: 10/5 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective cardiac surgery with CPB Inclusion criteria: LVEF ≤ 30% Exclusion criteria: no information given
Interventions	Intervention: levosimendan; 0.1 µg/kg/min in combination with dobutamine; 5 µg/kg/min started after removal of the cross‐clamp until weaning from inotropic support Comparison: milrinone; 0.5 µg/kg/min in combination with dobutamine; 5 µg/kg/min started after removal of the cross‐clamp until weaning from inotropic support Concomitant medication: dobutamine; norepinephrine Excluded medications: ACE inhibitors
Outcomes	Primary: stroke volume index at 24 hours after admission to ICU Secondary: haemodynamics; laboratory parameters; risk stratification using EuroSCORE; incidence of adverse events (myocardial infarction, atrial fibrillation); length of ventilation/ICU stay/hospital stay; amount of chest tube drainage; need for blood transfusion/inotropic support Time points reported: preoperative; 15 min after separation from CPB; at the end of surgery; 0/6/12/24/48 hours after admission to ICU; postoperative day 30
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of participants; no blinding of caregivers
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of observers
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	Definition of the primary outcome only
Other bias	Low risk	None

Denault 2016.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: April 2009 to November 2011 Total duration of follow‐up: 1 year post‐surgery Number of study centres: 4 Location of study centres: Canada Period of study: preoperative until end of surgery
Participants	N randomised: 125 N lost to follow‐up or withdrawn: 8 N analysed: 124 Mean age (years): intervention group: 70.2 ± 10.2; comparison group: 68.3 ± 9.2 Gender (male/female): intervention group: 25/38; comparison group: 35/26 Participant with or without pre‐existing LV dysfunction: with and without Type of cardiac surgery: elective valve surgery with or without CABG Inclusion criteria: age ≥ 18 years; preoperative pulmonary hypertension Exclusion criteria: emergency surgery; surgery without CPB; need for preoperative pharmacological/mechanical support due to haemodynamic instability; congenital heart disease; contraindication to transesophageal echocardiography
Interventions	Intervention: milrinone (single dose of 5 mg inhaled iMil) after induction of anaesthesia Comparison: placebo Concomitant medication: milrinone (intravenous); nitroglycerine; NO (inhaled); norepinephrine; epinephrine; dobutamine; PGI2 (inhaled); phenylephrine, vasopressin Excluded medications: no information given
Outcomes	Primary: number of patients classified as having difficult or complex separation from CPB, i.e. at least two different types of pharmacological agents or IABP required or weaning failure or intraoperative death from heart failure Secondary: right ventricular failure, i.e. haemodynamic instability defined as difficult/complex separation from CPB + reduction in RV fraction area > 20% + anatomical visualisation of impaired/absent RV wall motion; need for rescue therapy for postoperative pulmonary hypertension; intraoperative persistent arrhythmias requiring medical intervention; cardioversion of defibrillation; need for vasoactive support for more than 24 hours; length of ventilation/ICU stay/hospital stay; incidence of mortality Time points reported: preoperative; before/after induction of anaesthesia; 20 min after starting medication before onset of CPB; 20 min after separation from CPB; at the end of surgery; postoperative months 1/3/12
Notes	Funding for trial: Supported by the Montreal Heart Institute Foundation, the Fonds de la Recherche en Sante du Quebec, the Heart and Stroke Foundation of Canada and the Earl Wynands Canadian Anaesthesia Research Foundation. Notable conflicts of interest of trial authors: Dr. Deanult is speaker for CAE Healthcare and Covidien and receives royalties from Taylor and Francis.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Allocation process and drug preparation performed by persons not involved in the trial performance; identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	Balanced missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Unclear risk	Baseline difference in gender

Desai 2018.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: January 2015 to September 2015 Total duration of follow‐up: length of stay in hospital Number of study centres: 1 Location of study centres: India Period of study: preoperative until 24 hours post‐surgery
Participants	N randomised: 60 N lost to follow‐up or withdrawn: 0 N analysed: 60 Mean age (years): intervention group: 61.67 ± 7.03; comparison group: 60.17 ± 5.72 Gender (male/female): intervention group: 21/9; comparison group: 23/7 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective OPCABG Inclusion criteria: age 35 to 75 years; LVEF ≤ 30% Exclusion criteria: emergency/redo/combined CABG; need for preoperative pharmacological/mechanical support; renal/hepatic dysfunction
Interventions	Intervention: levosimendan 0.1 µg/kg/min for 24 hours started 12 hours before surgery Comparison: standard inotropic agents (dobutamine, epinephrine, norepinephrine) intraoperatively Concomitant medication: norepinephrine Excluded medications: other inotropic support
Outcomes	Primary: haemodynamics; short‐term clinical outcomes (incidence of arrhythmia/acute kidney injury/LCOS, length of ICU stay/hospital stay) Secondary: no information given Time points reported: preoperative; 15 min after anastomosis; at the end of surgery; 6/12/24 hours after admission to ICU; on discharge from hospital
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	High risk	Different management of the study groups
Blinding of outcome assessment (detection bias) All outcomes	High risk	Different management of the study groups
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Erb 2014.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: March 2008 to December 2009 Total duration of follow‐up: 6 months post‐surgery Number of study centres: 1 Location of study centres: Germany Period of study: preoperative until 3 days post‐surgery/discharge from ICU
Participants	N randomised: 37 N lost to follow‐up or withdrawn: 4 N analysed: 33 Mean age (years): intervention group: 69.5 ± 11.5; comparison group: 63.4 ± 7.8 Gender (male/female): intervention group: 13/4; comparison group: 15/1 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG with or without valve surgery Inclusion criteria: ischaemic cardiomyopathy; age ≥ 18 years; LVEF ≤ 30% Exclusion criteria: pregnancy and/or breastfeeding; hepatic dysfunction; disease or recent operation (< 2 months before surgery) of the oesophagus or upper airway; neurological/psychiatric disorder; diabetes mellitus treated with sulfonylurea drugs; infection with HIV/hepatitis B or C; alcohol abuse
Interventions	Intervention: levosimendan 0.1 µg/kg/min up to a total dose of 12.5 mg started with the induction of anaesthesia Comparison: placebo Concomitant medication: dobutamine, enoximone; epinephrine; norepinephrine Excluded medications: no information given
Outcomes	Primary: SOFA scores on postoperative days 0/1/2/3 Secondary: haemodynamics; need for inotropic support; need for haemodialysis; length of ICU stay; incidence of mortality on postoperative day 30 and month 6; quality of life on postoperative month 6 Time points reported: preoperative; after induction of anaesthesia; before/after CPB; at admission to ICU; postoperative days 1/2/3/4/30; postoperative month 6
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Eriksson 2009.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: 31 days post‐surgery Number of study centres: 2 Location of study centres: Finland Period of study: preoperative until 24 hours post‐surgery
Participants	N randomised: 60 N lost to follow‐up or withdrawn: 0 N analysed: 60 Mean age (years): intervention group: 64 ± 10; comparison group: 64 ± 10 Gender (male/female): intervention group: 28/2; comparison group: 26/4 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG Inclusion criteria: 3‐vessel coronary artery disease; LVEF ≤ 50% and/or signs of acute ischaemic congestive heart failure Exclusion criteria: redo surgery; indication for any cardiac valve surgery; weight > 160 kg; severe obstructive pulmonary disease; administration of levosimendan within the preceding 30 days
Interventions	Intervention: levosimendan; loading dose of 12 µg/kg within 10 min started with the induction of anaesthesia followed by 0.2 µg/kg/min for 24 hours Comparison: placebo Concomitant medication: ephedrine; epinephrine; milrinone; norepinephrine; phenylephrine; vasopressin Excluded medications: no information given
Outcomes	Primary: proportion of patients successfully weaned from CPB on the first attempt Secondary: haemodynamics; need for inotropic support; 24‐hour fluid balance; markers of myocardial injury; pharmacokinetics of levosimendan; postoperative recovery/safety data up to 31 days Time points reported: preoperative; at the end of surgery; 4/8/24 hours after declamping aorta; postoperative day 31
Notes	Funding for trial: Orion Pharma provided financial support for the study. Notable conflicts of interest of trial authors: Some of the study authors (Kivikko/Laine/Sarapohja) disclose that they have a financial relationship with Orion Pharma.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block permutation stratified by centre
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Ersoy 2013.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: May 2006 to July 2007 Total duration of follow‐up: length of stay in hospital Number of study centres: 1 Location of study centres: Turkey Period of study: preoperative until 24 hours after induction of anaesthesia
Participants	N randomised: 20 N lost to follow‐up or withdrawn: 0 N analysed: 20 Mean age (years): intervention group: 49.6 ± 10.7; comparison group: 45.7 ± 7.9 Gender (male/female): intervention group: 5/5; comparison group: 3/7 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective valve surgery Inclusion criteria: severe pulmonary arterial hypertension (systolic pulmonary artery pressure ≥ 60 mmHg); LVEF ≤ 50% Exclusion criteria: no information given
Interventions	Intervention: levosimendan; loading dose of 12 µg/kg within 10 min started with the induction of anaesthesia followed by 0.1 µg/kg/min for 24 hours Comparison: standard cardiac care Concomitant medication: inotropic drugs Excluded medications: no information given
Outcomes	Primary: haemodynamics Secondary: safety data (incidence of LCOS/mortality, need for IABP/inotropic support, length of ICU stay/hospital stay) Time points reported: preoperative; after induction of anaesthesia; after administration of the loading dose; 6/12/24 hours after start of medication; on discharge from hospital
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Some rule based on sequence of admission
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	High risk	Different management of the study groups
Blinding of outcome assessment (detection bias) All outcomes	High risk	Different management of the study groups
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Unclear risk	Baseline difference in gender

Gandham 2013.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: July 2011 to December 2011 Total duration of follow‐up: length of stay in ICU Number of study centres: 1 Location of study centres: India Period of study: preoperative until 36 hours after separation from CPB
Participants	N randomised: 60 N lost to follow‐up or withdrawn: 0 N analysed: 60 Mean age (years): intervention group: 36.13 ± 7.11; comparison group: 39.43 ± 6.85 Gender (male/female): intervention group: 12/18; comparison group: 14/16 Participant with or without pre‐existing LV dysfunction: without Type of cardiac surgery: elective valve surgery Inclusion criteria: mitral stenosis; age < 60 years Exclusion criteria: redo/combined surgery; moderate to severe mitral regurgitation or other valvular pathologies; renal dysfunction; re‐exploration for surgical causes
Interventions	Intervention: levosimendan 0.1 µg/kg/min during weaning from CPB Comparison: dobutamine 5 µg/kg/min during weaning from CPB Concomitant medication: epinephrine; norepinephrine Excluded medications: no information given
Outcomes	Primary: haemodynamics Secondary: length of ventilation/ICU stay; lactate levels Time points reported: preoperative; 0.5/6/12/24/36 hours after separation from CBP; on discharge from ICU
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Gatot 2004.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: June 2001 to October 2001 Total duration of follow‐up: length of stay in hospital Number of study centres: 1 Location of study centres: Israel Period of study: preoperative until 96 hours post‐surgery
Participants	N randomised: 89 N lost to follow‐up or withdrawn: 8 N analysed: 81 Mean age (years): intervention group: 64 ± 10; comparison group: 66 ± 8 Gender (male/female): intervention group: 28/13; comparison group: 26/14 Participant with or without pre‐existing LV dysfunction: with and without Type of cardiac surgery: elective CABG Inclusion criteria: isolated CABG Exclusion criteria: redo/combined surgery; OPCABG; pre‐operative dialysis or serum creatinine > 2.5 mg/dL; arrhythmias/severe hypertension; need for IABP; bleeding over 1000 mL in the first 24 hours post‐surgery or need for reoperation as a result of bleeding; need for mechanical ventilation for more than 24 hours post‐surgery
Interventions	Intervention: dopamine 5 µg/kg/min for 48 hours after surgery Comparison: placebo Concomitant medication: dopamine; epinephrine; nitroglycerine; nitroprusside; norepinephrine Excluded medications: no information given
Outcomes	Primary: haemodynamics; cardiac status (incidence of arrhythmias, myocardial infarction, LCOS); respiratory status; renal function (urine output, serum creatinine); general patient recovery (mobilisation, length of hospital stay) Secondary: no information given Time points reported: preoperative; 8/16/24/36/48/72/96 hours after admission to ICU; on discharge from hospital
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	High risk	Reasons for exclusion of study participants related to true outcome
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Hadadzadeh 2013.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: length of stay in ICU Number of study centres: 1 Location of study centres: Iran Period of study: preoperative until 24 hours post‐surgery
Participants	N randomised: 80 N lost to follow‐up or withdrawn: 0 N analysed: 80 Mean age (years): intervention group: 61.90 ± 10.71; comparison group: 63.00 ± 9.60 Gender (male/female): intervention group: 31/9; comparison group: 26/14 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective OPCABG Inclusion criteria: severe myocardial dysfunction; LVEF ≤ 35% Exclusion criteria: emergency surgery; recent myocardial infarction or ventricular arrhythmias (< 72 hours before surgery); need for inotropic support before surgery; concomitant valvular heart disease; left bundle branch block
Interventions	Intervention: milrinone; loading dose of 50 µg/kg started immediately after surgery, followed by 0.5 µg/kg/min for 24 hours Comparison: placebo Concomitant medication: inotropic support Excluded medications: no information given
Outcomes	Primary: incidence of myocardial dysfunction (ischaemia/myocardial infarction) Secondary: no information given Time points reported: preoperative; at the end of surgery; 24 hours after admission to ICU; on discharge from ICU
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided
Allocation concealment (selection bias)	Low risk	Identical administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Unclear risk	Baseline difference in gender

Hu 2020.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: January 2014 to June 2019 Total duration of follow‐up: 14 days post‐surgery Number of study centres: 1 Location of study centres: China Period of study: preoperative until 14 days post‐surgery
Participants	N randomised: 320 N lost to follow‐up or withdrawn: 0 N analysed: 320 Mean age (years): intervention group: 58.7 ± 4.6; comparison group: 57.9 ± 5.3 Gender (male/female): intervention group: 78/82; comparison group: 84/76 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG or valve surgery Inclusion criteria: age 25 to 70 years; LVEF ≤ 50%; NYHA: III or IV; moderate to severe pulmonary hypertension Exclusion criteria: active bleeding; abnormalities in coagulation or haematopoietic function; severe LCOS requiring preoperative IABP; malignant tumours; psychiatric disorders; renal/hepatic dysfunction
Interventions	Intervention: levosimendan 0.1 to 0.2 µg/kg/min up to a total dose of 12.5 mg started after surgery Comparison: placebo Concomitant medication: no information given Excluded medications: no information given
Outcomes	Primary: haemodynamics; incidence of pulmonary hypertension; renal function Secondary: no information given Time points reported: preoperative; 0/6/12/24 hours after admission to ICU; postoperative days 3/7/14
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	High risk	Different management of the study groups
Blinding of outcome assessment (detection bias) All outcomes	High risk	Different management of the study groups
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Jävelä 2008.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: Finland Period of study: preoperative until 24 hours post‐surgery
Participants	N randomised: 24 N lost to follow‐up or withdrawn: 0 N analysed: 24 Mean age (years): intervention group: 69 ± 11; comparison group: 67 ± 10 Gender (male/female): intervention group: 9/3; comparison group: 9/3 Participant with or without pre‐existing LV dysfunction: without Type of cardiac surgery: elective valve surgery with or without CABG Inclusion criteria: mean transaortic gradient > 50 mmHg; LV hypertrophy (wall thickness > 12 mm) Exclusion criteria: active endocarditis
Interventions	Intervention: levosimendan 0.2 µg/kg/min for 24 hours started with the induction of anaesthesia Comparison: placebo Concomitant medication: dobutamine; epinephrine; milrinone; norepinephrine; sodium nitroprusside Excluded medications: ACE inhibitors; diuretics
Outcomes	Primary: haemodynamics; need for vasoactive medication Secondary: no information given Time points reported: preoperative; after the induction of anaesthesia; after separation from CPB; at the end of surgery; 3/4/6/8/9/12/15/16/18/20/21/24 hours after admission to ICU; postoperative days 4/5/30
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Sealed envelopes; identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Jo 2010.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: Korea Period of study: preoperative until 12 hours post‐surgery
Participants	N randomised: 40 N lost to follow‐up or withdrawn: 0 N analysed: 40 Mean age (years): intervention group: 67.0 ± 9.2; comparison group: 64.1 ± 9.9 Gender (male/female): intervention group: 12/8; comparison group: 11/9 Participant with or without pre‐existing LV dysfunction: with and without Type of cardiac surgery: elective OPCABG Inclusion criteria: stenosis at more than 3 coronary arteries Exclusion criteria: emergency surgery; LVEF or RVEF < 40%; valvular heart disease; severe cerebral/renal dysfunction
Interventions	Intervention: milrinone 0.5 µg/kg/min during coronary artery anastomosis Comparison: placebo Concomitant medication: norepinephrine Excluded medications: no information given
Outcomes	Primary: haemodynamics; right ventricular function; short‐term clinical outcomes (cerebrovascular event, renal failure, prolonged ventilation (> 48 hours), reoperation, sternal infection) Secondary: no information given Time points reported: preoperative; after pericardiotomy; 5 min each after stabiliser application for anastomosis of specific arteries; at the end of surgery; 6/12 hours after admission to ICU; postoperative day 30
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Juhl‐Olsen 2015.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: February 2011 to April 2012 Total duration of follow‐up: 6 months post‐surgery Number of study centres: 1 Location of study centres: Denmark Period of study: preoperative until 6 months post‐surgery
Participants	N randomised: 21 N lost to follow‐up or withdrawn: 1 N analysed: 20 Mean age (years): intervention group: 75.9 (68.6 to 87.0); comparison group: 72.8 (51.2 to 85.2) Gender (male/female): intervention group: 8/2; comparison group: 5/5 Participant with or without pre‐existing LV dysfunction: without Type of cardiac surgery: elective valve surgery without CABG Inclusion criteria: aortic valve stenosis; LVEF > 45%; LV hypertrophy (LV posterior wall thickness ≥ 13 mm); sinus rhythm Exclusion criteria: need for CABG; active endocarditis; systolic blood pressure < 100 mmHg; creatinine clearance < 30 ml/min; insufficient conditions for echocardiographic imaging
Interventions	Intervention: levosimendan 0.1 µg/kg/min started 4 hours before surgery until the end of surgery Comparison: placebo Concomitant medication: dobutamine; nitroprusside; norepinephrine Excluded medications: no information given
Outcomes	Primary: change in E/e (correlate of LV filling pressure) as measured by echocardiography from baseline (4 hours before surgery) to the first postoperative morning (approximately 21 hours post‐surgery) Secondary: echocardiographic measures of systolic and diastolic function; haemodynamics; cardiac biomarkers (NT‐proBNP, troponin T); 24‐hour urine clearance; safety measures (incidence of atrial fibrillation or ventricular tachycardia within 4 days after surgery, frequency of headache/nausea or need for noradrenaline during pre‐operative start of medication, need for vasoactive support within 96 hours after start of surgery, volume status within 24 hours after start of surgery, length of surgery/CBP/ventilation/ICU stay) Time points reported: preoperative; before and after the induction of anaesthesia; at the end of surgery; 4 hours after admission to ICU; postoperative morning; postoperative day 4; postoperative month 6
Notes	Funding for trial: Partially funded by the Danish Heart Foundation and the Danish Society of Anaesthesiology and Intensive Care Medicine. The echocardiographic equipment used was provided by the John and Birthe Meyer's Foundation. Orion Pharma provided vials of levosimendan and placebo for the study. Notable conflicts of interest of trial authors: The authors declare no financial relationships with Orion Pharma. One study author (Erik Sloth) received minor fees for lectures by BK medical and GE Healthcare.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Sealed envelopes; identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	Balanced missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Unclear risk	Baseline difference in gender

Kandasamy 2017.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: September 2012 to January 2014 Total duration of follow‐up: length of stay in hospital Number of study centres: 1 Location of study centres: India Period of study: preoperative until 24 hours post‐surgery
Participants	N randomised: 80 N lost to follow‐up or withdrawn: 0 N analysed: 80 Mean age (years): intervention group: 55.2 ± 3.228; comparison group: 54.8 ± 3.724 Gender (male/female): intervention group: 31/9; comparison group: 33/7 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective OPCABG Inclusion criteria: age 30 to 65 years; moderate to severe LV dysfunction Exclusion criteria: emergency/redo surgery; moderate to severe mitral regurgitation; need for preoperative pharmacological/mechanical support
Interventions	Intervention: levosimendan 0.1 µg/kg/min for 24 hours started with the induction of anaesthesia Comparison: dobutamine 5 µg/kg/min for 24 hours started with the induction of anaesthesia Concomitant medication: epinephrine; norepinephrine Excluded medications: no information given
Outcomes	Primary: haemodynamics Secondary: incidence of postoperative atrial fibrillation; length of ventilation/ICU stay/hospital stay Time points reported: preoperative; 30 min after start of surgery; at obtuse marginal grafting; 1/6/12/24 hours after admission to ICU; on discharge from hospital
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Kikura 2002.

Study characteristics
Methods	Study design: 3‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: 3 days post‐surgery Number of study centres: 1 Location of study centres: Japan Period of study: preoperative until 3 days post‐surgery
Participants	N randomised: 45 N lost to follow‐up or withdrawn: 0 N analysed: 45 Mean age (years): intervention group1: 70 ± 8; intervention group2: 68 ± 8; comparison group: 67 ± 7 Gender (male/female): intervention group 1: 7/8; intervention group 2: 8/7; comparison group: 9/6 Participant with or without pre‐existing LV dysfunction: without Type of cardiac surgery: elective CABG Inclusion criteria: unstable/stable, progressive angina pectoris; stenosis ≥ 90% of at least 2 coronary arteries Exclusion criteria: emergency/combined surgery; recent myocardial infarction (< 72 hours before surgery); history of persistent ventricular tachycardia or obstructive cardiomyopathy; need for preoperative pharmacological/mechanical support
Interventions	Intervention: intervention 1 = milrinone; loading dose of 50 µg/kg over 10 to 15 min started after removal of the cross‐clamp followed by 0.5 µg/kg/min for 10 hours after CPB; intervention 2 = amrinone; loading dose of 1.5 mg/kg over 10 to 15 min started after removal of the cross‐clamp followed by 10 µg/kg/min for 10 h after CPB Comparison: placebo Concomitant medication: dopamine; nitroglycerine Excluded medications: no information given
Outcomes	Primary: ventricular function and oxygen transport; need for dopamine/nitroglycerine; levels of lactate/glucose/cellular enzymes Secondary: no information given Time points reported: preoperative; after the induction of anaesthesia; before separation from CBP; 0.25/0.5/1/2/3/6/9/12/15/18/21/24 hours after admission to ICU; postoperative days 2/3
Notes	Funding for trial: Supported by a grant from the Ministry of Education in Japan. Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Kodalli 2013.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: June 2011 to December 2011 Total duration of follow‐up: length of stay in ICU Number of study centres: 1 Location of study centres: India Period of study: preoperative until 24 hours post‐surgery
Participants	N randomised: 30 N lost to follow‐up or withdrawn: 0 N analysed: 30 Mean age (years): intervention group: 56 ± 8.1; comparison group: 52.5 ± 8.9 Gender (male/female): intervention group: 12/3; comparison group: 8/7 Participant with or without pre‐existing LV dysfunction: without Type of cardiac surgery: elective OPCABG Inclusion criteria: LVEF > 45% Exclusion criteria: redo surgery; LVEF < 45%; acute myocardial infarction; renal/hepatic dysfunction; significant pulmonary disease; left bundle branch block; moderate to severe valvular heart disease; unplanned CBP; need for IABP; re‐exploration for surgical bleeding
Interventions	Intervention: levosimendan 0.1 µg/kg/min started with the induction of anaesthesia (termination at the discretion of the responsible anaesthesiologist in the ICU) Comparison: placebo Concomitant medication: epinephrine; nitroglycerine; norepinephrine Excluded medications: ACE inhibitors
Outcomes	Primary: haemodynamics; myocardial performance Secondary: need for vasoactive support; length of ventilation/ICU stay Time points reported: preoperative; after the induction of anaesthesia; 6 min after application of tissue stabiliser for anastomoses of left anterior descending/diagonal/left circumflex/right coronary artery; 6/12/18/24 hours after admission to ICU; discharge from ICU
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Unclear risk	Baseline difference in gender

Lahtinen 2011.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: March 2005 to November 2008 Total duration of follow‐up: 180 days post‐surgery Number of study centres: 1 Location of study centres: Finland Period of study: preoperative until discharge from hospital
Participants	N randomised: 207 N lost to follow‐up or withdrawn: 7 N analysed: 200 Mean age (years): intervention group: 68 ± 9; comparison group: 70 ± 8 Gender (male/female): intervention group: 69/30; comparison group: 71/30 Participant with or without pre‐existing LV dysfunction: with and without Type of cardiac surgery: elective valve surgery with or without CABG Inclusion criteria: age ≥ 18 years Exclusion criteria: endocarditis; ventricular septal defect; aortic arch or descending aortic surgery; patient unable to read or understand the informed consent; patients with a history of alcohol, drug or other substance abuse; administration of levosimendan within the preceding 2 weeks
Interventions	Intervention: levosimendan; loading dose of 24 µg/kg within 30 min started with the induction of anaesthesia followed by 0.2 µg/kg/min for 24 hours Comparison: placebo Concomitant medication: dobutamine; epinephrine; milrinone; norepinephrine; sodium nitroprusside Excluded medications: no information given
Outcomes	Primary: incidence of heart failure (cardiac index < 2.0 L/min/m2 or failure to wean from CBP necessitating inotrope administration for at least 2 hours postoperatively after CBP Secondary: any cause in‐hospital and 6‐month mortality; predefined major organ morbidity (myocardial infarction, renal insufficiency, cerebrovascular events, hepatic insufficiency, prolonged ICU stay, prolonged ventilation, prolonged hospital stay) Time points reported: preoperative; 2 hours after separation from CPB; discharge from hospital; postoperative day 180
Notes	Funding for trial: Supported in part by a restricted research grant from Orion Pharma and a research grant from Kuopio University hospital. Notable conflicts of interest of trial authors: Some of the study authors (Lahtinen/Pitkänen) received honoraria from Orion Pharma. The remaining authors have not disclosed any potential conflicts of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Allocation process and drug preparation performed by persons not involved in the trial performance; identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	Balanced missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Leppikangas 2011.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: January 2009 to September 2010 Total duration of follow‐up: length of stay in hospital Number of study centres: 1 Location of study centres: Finland Period of study: preoperative until 4 days post‐surgery
Participants	N randomised: 24 N lost to follow‐up or withdrawn: 0 N analysed: 24 Mean age (years): intervention group: 76 ± 10; comparison group: 75 ± 8 Gender (male/female): intervention group: 10/2; comparison group: 8/4 Participant with or without pre‐existing LV dysfunction: with and without Type of cardiac surgery: elective valve surgery with CABG Inclusion criteria: LVEF < 50% or LV hypertrophy (wall thickness > 12 mm) Exclusion criteria: known allergy to levosimendan
Interventions	Intervention: levosimendan; loading dose of 12 µg/kg within 10 min started the day before surgery, followed by 0.2 µg/kg/min for 24 hours Comparison: placebo Concomitant medication: dobutamine; epinephrine; milrinone; norepinephrine; sodium nitroprusside Excluded medications: no information given
Outcomes	Primary: haemodynamics Secondary: echocardiographic measures of systolic and diastolic function Time points reported: preoperative; before/after the induction of anaesthesia; after separation from CBP, at the end of surgery; first 24 hours after admission to ICU; postoperative days 2/3/4; discharge from hospital
Notes	Funding for trial: Supported by grants from the Finnish Cultural Foundation, Competitive Research Funding of the Pirkanmaa Hospital District, Tampere University Hospital, Ina Montin Foundation and Orion Pharma. Notable conflicts of interest of trial authors: Some of the study authors (Leppikangas/Lindgren) have lectured for Orion Pharma.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided
Allocation concealment (selection bias)	Low risk	Sealed envelopes; identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Unclear risk	Baseline difference in gender

Levin 2012.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: December 2002 to June 2008 Total duration of follow‐up: 30 days post‐surgery Number of study centres: 2 Location of study centres: USA Period of study: preoperative until 48 hours post‐surgery
Participants	N randomised: 252 N lost to follow‐up or withdrawn: 0 N analysed: 252 Mean age (years): intervention group: 63.7; comparison group: 62.9 Gender (male/female): intervention group: 94/33; comparison group: 94/31 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG Inclusion criteria: coronary artery disease affecting at least 2 vessels; LVEF < 25%; positive stress test for ischaemia or viability Exclusion criteria: urgent/emergency/combined surgery; off‐pump surgery; need for IABP; administration of levosimendan within the preceding 3 weeks or of other inotropes within the preceding week
Interventions	Intervention: levosimendan; loading dose of 10 µg/kg within 60 min started 24 hours before surgery, followed by 0.1 µg/kg/min for 23 hours Comparison: placebo Concomitant medication: norepinephrine; phenylephrine Excluded medications: no information given
Outcomes	Primary: incidence of postoperative LCOS and mortality Secondary: incidence of difficult weaning from CBP; need for inotropes, vasopressors and IABP Time points reported: preoperative; after loading dose; before the induction of anaesthesia; 0/6/12/24/48 hours after admission to ICU; postoperative day 30
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Sequence generated by medical record number
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Mehta 2017.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: September 2014 to November 2016 Total duration of follow‐up: 90 days post‐surgery Number of study centres: 70 Location of study centres: USA and Canada Period of study: preoperative until 30 days post‐surgery
Participants	N randomised: 882 N lost to follow‐up or withdrawn: 33 N analysed: 849 Mean age (years): intervention group: 65 (59 to 73); comparison group: 65 (58 to 72) Gender (male/female): intervention group: 347/81; comparison group: 332/89 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG/valve surgery either isolated or in combination Inclusion criteria: age ≥ 18 years; LVEF ≤ 35%; signed informed content Exclusion criteria: restrictive or obstructive cardiomyopathy, restrictive or constrictive pericarditis, pericardial tamponade or other conditions in which cardiac output is dependent on venous return; evidence of systemic bacterial, fungal or viral infection within 72 hours before surgery; dialysis at the time of randomisation; estimated glomerular filtration rate < 30 mL/min per 1.73 m2; weight ≥ 170 kg; systolic blood pressure not manageable to ensure it to be > 90 mmHg at initiation of study drug; heart rate ≥ 120 beats/min persistent for at least 10 min at screening and unresponsive to treatment; haemoglobin < 80 g/L; serum potassium < 3.5 mmol/L or > 5.5 mmol/L at baseline; history of torsade de pointes; mechanical assist device placed at the start of surgery or preplanned to be placed during surgery before weaning from CBP; aortal femoral occlusive disease; liver dysfunction with Child Pugh class B or C; severely compromised immune function; (suspected) pregnancy or breast‐feeding; administration of an experimental drug or use of an experimental medical device within the preceding 30 days; known allergy/sensitivity to levosimendan or excipient; administration of levosimendan within the preceding 30 days; employees/family members of the investigator or study centre with direct involvement in the proposed study or other studies under the direction of the investigator or study centre
Interventions	Intervention: levosimendan 0.2 µg/kg/min for 1 hour started before skin incision followed by 0.1 µg/kg/min for 23 hours Comparison: placebo Concomitant medication: other inotropes and vasopressors Excluded medications: nesiritide
Outcomes	Primary: 4‐component composite: death through day 30, renal replacement therapy through day 30, perioperative myocardial infarction through day 5, use of a mechanical cardiac assist device through day 5; 2‐component composite: death through day 30, use of a mechanical cardiac assist device through day 5 Secondary: incidence of LCOS; postoperative use of secondary inotropes at or beyond 24 h after the start of levosimendan/placebo; length of stay in ICU; adverse events (hypotension, atrial fibrillation, ventricular tachycardia or fibrillation, resuscitated cardiac arrest, cerebrovascular event, death through day 90) Time points reported: preoperative; postoperative days 5/30/90
Notes	Funding for trial: Supported by Tenax Therapeutics Notable conflicts of interest of trial authors: Some of the study authors received a significant (Mehta) or modest (Alexander) research grant from Tenax Therapeutics through Duke Clinical Research Institute. Some of the study authors received significant (Goodman) or modest (Fremes) research grants from Tenax Therapeutics through Duke Clinical Research Institute and the Canadian VIGOUR Centre. Some of the study authors (Bozinovski/Jankowich/Hay) are employees of Tenax Therapeutics. One study author (Heringlake) received modest honoraria for lectures from Orion Pharma. None declared for all other authors.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	Balanced missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Mishra 2016.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: January 2013 to June 2014 Total duration of follow‐up: length of stay in ICU Number of study centres: 1 Location of study centres: India Period of study: preoperative until 24 hours post‐extubation
Participants	N randomised: 40 N lost to follow‐up or withdrawn: 0 N analysed: 40 Mean age (years): intervention group: 37.30 ± 11.67; comparison group: 43.70 ± 13.13 Gender (male/female): no information given Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective valve surgery Inclusion criteria: LVEF < 50% in stenotic lesions and < 60% in regurgitant lesions; pulmonary hypertension (estimated right ventricular systolic pressure ≥ 50 mmHg or mean pulmonary arterial pressure > 40 mmHg or systolic pulmonary arterial pressure exceeding 50% of systemic systolic pressure) Exclusion criteria: emergency surgery; severe renal/hepatic dysfunction; prolonged QT interval; need for preoperative pharmacological support
Interventions	Intervention: levosimendan; loading dose of 10 µg/kg within 10 min started with the beginning of rewarming before separation from CPB followed by 0.1 µg/kg/min for 24 hours Comparison: milrinone; loading dose of 50 µg/kg within 10 min started with the beginning of rewarming before separation from CPB followed by 0.5 µg/kg/min for 24 hours Concomitant medication: epinephrine; norepinephrine Excluded medications: no information given
Outcomes	Primary: haemodynamics; right ventricular pulmonary vasculature coupling; incidence of pulmonary hypertension; biventricular function Secondary: no information given Time points reported: preoperative; before induction of anaesthesia; before and 0.5/24 hours after separation from CPB; 0/1/6/12/24 hours after admission to ICU; discharge from ICU
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: no information given
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of participants; no blinding of caregivers
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of observers
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Unclear risk	Baseline difference in age; no information on gender

Shah 2014.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: July 2012 to December 2012 Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: India Period of study: preoperative until 48 hours post‐surgery
Participants	N randomised: 50 N lost to follow‐up or withdrawn: 0 N analysed: 50 Mean age (years): intervention group: 59.91 ± 8.80; comparison group: 61.32 ± 7.64 Gender (male/female): intervention group: 15/10; comparison group: 16/9 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective OPCABG (cave: surgery was completed with the help of CPB for 9 participants, 1 from intervention group and 8 from comparison group) Inclusion criteria: coronary artery disease affecting 1 to 3 vessels; LVEF < 30%; viability of affected territory (as shown by myocardial perfusion scan) Exclusion criteria: urgent/emergency/redo/combined surgery; administration of levosimendan within 3 months or other inotropes within the previous week; need for preoperative IABP; significant pulmonary disease; renal/hepatic dysfunction; arrhythmias with bundle branch block
Interventions	Intervention: levosimendan 200 µg/kg over 24 hours started 24 hours before surgery Comparison: placebo Concomitant medication: dobutamine; epinephrine; norepinephrine Excluded medications: no information given
Outcomes	Primary: haemodynamics; myocardial performance Secondary: no information given Time points reported: preoperative; before, during and after administration of medication; 0/12/24/48 hours after admission to ICU; postoperative day 30
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Sequence generated by medical record number
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	High risk	Deviation from study protocol (CABG instead of OPCABG for some participants with unequal distribution between study groups)

Sharma 2014.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: January 2012 to December 2012 Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: India Period of study: preoperative until 48 hours post‐surgery
Participants	N randomised: 45 N lost to follow‐up or withdrawn: 5 N analysed: 40 Mean age (years): intervention group: 53.95 ± 12.06; comparison group: 54.55 ± 7.87 Gender (male/female): intervention group: 17/3; comparison group: 14/6 Participant with or without pre‐existing LV dysfunction: with Type of cardiac surgery: elective CABG with valve surgery Inclusion criteria: coronary artery disease; LVEF < 30%; severe mitral regurgitation Exclusion criteria: emergency/redo surgery; any other valve pathology; need for mitral valve replacement; recent myocardial infarction (< 30 days before surgery); diabetes mellitus treated with sulfonylurea drugs; renal/hepatic dysfunction; severe chronic obstructive pulmonary disease; preoperative intubation
Interventions	Intervention: levosimendan 200 µg/kg over 24 hours started 24 hours before surgery Comparison: placebo Concomitant medication: norepinephrine Excluded medications: no information given
Outcomes	Primary: haemodynamics; complications (LCOS, myocardial infarction, renal failure, cerebrovascular events, sepsis, adult respiratory distress syndrome, hypoxaemia, mortality) Secondary: no information given Time points reported: preoperative; after administration of medication; before induction of anaesthesia; 0/6/12/24/48 hours after admission to ICU; postoperative day 30
Notes	Funding for trial: none Notable conflicts of interest of trial authors: The authors declare no conflict of interest.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Sequence generated by medical record number
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Unclear risk	Baseline difference in gender

Tritapepe 2006.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: no information given Total duration of follow‐up: length of stay in hospital Number of study centres: 1 Location of study centres: Italy Period of study: preoperative until 48 hours post‐surgery
Participants	N randomised: 24 N lost to follow‐up or withdrawn: 0 N analysed: 24 Mean age (years): intervention group: 66.5 (61.8 to 69.8); comparison group: 69.5 (64 to 71.5) Gender (male/female): intervention group: 9/3; comparison group: 10/2 Participant with or without pre‐existing LV dysfunction: with and without Type of cardiac surgery: elective CABG Inclusion criteria: multiple‐vessel coronary artery disease, stable angina Exclusion criteria: redo surgery; recent myocardial infarction (< 1 month before surgery); unstable angina; valvular disease; diabetes mellitus treated with sulfonylurea drugs; renal/hepatic dysfunction; severe chronic obstructive pulmonary disease
Interventions	Intervention: levosimendan; bolus of 24 µg/kg within 10 min before initiation of CPB Comparison: placebo Concomitant medication: dobutamine; phenylephrine Excluded medications: no information given
Outcomes	Primary: myocardial performance (haemodynamics, ECG recordings); myocardial injury (troponin I) Secondary: no information given Time points reported: preoperative; after induction of anaesthesia; 0/6/24/48 hours after admission to ICU; discharge from hospital
Notes	Funding for trial: no information given Notable conflicts of interest of trial authors: One study author (Singer) has sat on advisory boards for levosimendan on behalf of Orion Pharma and Abbott and has received honoraria for chairing/speaking at satellite symposia and for co‐editing a website (failinghears.com) supported by an unrestricted educational grant from Abbott.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Identical administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Unclear risk	No information provided
Other bias	Low risk	None

Tritapepe 2009.

Study characteristics
Methods	Study design: 2‐arm, parallel‐group RCT covering elective cardiac surgery Total duration of enrolment: January 2005 to February 2007 Total duration of follow‐up: 30 days post‐surgery Number of study centres: 1 Location of study centres: Italy Period of study: preoperative until 48 hours post‐surgery
Participants	N randomised: 106 N lost to follow‐up or withdrawn: 4 N analysed: 102 Mean age (years): intervention group: 64.0 (54.0 to 86.0); comparison group: 66.5 (54.0 to 87.0) Gender (male/female): intervention group: 43/9; comparison group: 39/11 Participant with or without pre‐existing LV dysfunction: with and without Type of cardiac surgery: elective CABG Inclusion criteria: multiple‐vessel coronary artery disease; age ≥ 18 years Exclusion criteria: redo surgery; recent myocardial infarction (< 1 month before surgery); unstable angina; valvular disease; diabetes mellitus treated with sulfonylurea drugs; renal/hepatic dysfunction; severe chronic obstructive pulmonary disease
Interventions	Intervention: levosimendan; bolus of 24 µg/kg within 10 min before initiation of CPB Comparison: placebo Concomitant medication: dopamine; epinephrine; norepinephrine; phenylephrine Excluded medications: no information given
Outcomes	Primary: length of ICU stay Secondary: length of ventilation/hospital stay; need for inotropic support over the first 7 days; incidence of mortality/myocardial infarction/atrial fibrillation/any other adverse event; serum creatinine concentrations Time points reported: preoperative; 0/6/24/48 h after admission to ICU; postoperative days 7/30
Notes	Funding for trial: Supported by an independent research grant from the Department of Anaesthesiology and Intensive Care of the Sapienza University of Rome Notable conflicts of interest of trial authors: Some study authors (Tritapepe/Guarracino/Singer) have sat on advisory boards for levosimendan on behalf of Orion Pharma and Abbott and have received honoraria for chairing/speaking at satellite symposia.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random number generator
Allocation concealment (selection bias)	Low risk	Identical appearance and administration of study drugs
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Double‐blind study
Incomplete outcome data (attrition bias) All outcomes	Low risk	Balanced missing outcome data
Selective reporting (reporting bias)	Low risk	Predefined outcomes reported
Other bias	Low risk	None

Abbreviations:

ACE: angiotensin‐converting enzyme; CABG: coronary bypass graft; CPB: cardiopulmonary bypass; IABP: intra‐aortic balloon pump; ICU: intensive care unit; ITT: intention‐to‐treat; LCOS: low cardiac output syndrome; LV: left ventricular; LVEF: left ventricular ejection fraction; NYHA: New York Heart Association; OPCABG: off‐pump coronary bypass graft; RCT: randomised controlled trial; RVEF: right ventricular ejection fraction; SOFA: sequential organ failure assessment

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Alvarez 2005	Treatment not prophylaxis of LCOS
Atalay 2016	No randomisation
Barisin 2004	Limited to perioperative evaluation
Fredholm 2018	Limited to perioperative evaluation
Fredholm 2020	Limited to perioperative evaluation
Gattiker and Schmid 1978	Cross‐over design; limited to perioperative evaluation
Giannini 2017	No randomisation; no assessment of LCOS
Hamada 1999	Limited to perioperative evaluation
Hausen 1992	Treatment not prophylaxis of LCOS
Jebeli 2010	Treatment not prophylaxis of LCOS
Kumar 2021	No assessment of LCOS
Lilleberg 1998	Limited to perioperative evaluation
Nijhawan 1999	Limited to perioperative evaluation
Oppizzi 1997	Goal‐directed dosing of study drugs
Sahin 2014	No randomisation; no assessment of LCOS
Sahu 2016	Exclusion of participants with postoperative LCOS
Tarr 1993	Limited to perioperative evaluation; goal‐directed dosing of study drugs; inclusion of participants with complex heart defects
Temizturk 2016	No randomisation
van Diepen 2020	Post hoc analysis of Mehta 2017
Zepplin 1990	Treatment not prophylaxis of LCOS

LCOS: low cardiac output syndrome

Characteristics of ongoing studies [ordered by study ID]

ChiCTR2100046169.

Study name	Intermittent repeated administration of levosimendan improves outcomes in patients undergoing cardiopulmonary bypass surgery
Methods	Mono‐centre, 2‐arm RCT in China (blinding not specified)
Participants	n = not specified Inclusion criteria: age ≥ 18 years and ≤ 80 years; undergoing elective cardiac surgery Exclusion criteria: suffering aortic stenosis; suffering hypertrophic obstructive cardiomyopathy; pregnant women; allergy to levosimendan; undergoing another drug trial Characteristics: male gender
Interventions	Levosimendan: 0 to 0.2 µg/kg/min for 24 hours from the second day after admission Placebo: 0 to 0.2 µg/kg/min for 24 hours from the second day after admission
Outcomes	Primary: mortality; low cardiac output syndrome; acute renal failure Secondary: not specified Other: not specified
Starting date	May 2021
Contact information	Tan Qi (deepblue1229@163.com)
Notes	Not specified

NCT04179604.

Study name	Rationale and design of a multicenter randomized trial of levosimendan to reduce low cardiac output syndrome in low ejection fraction (≤ 35%) cardiac surgery patients. Spanish Randomized Clinical Trial on Levosimendan (SPARTANS Study)
Methods	Multi‐centre, triple‐blind, 2‐arm RCT in Spain
Participants	n = 300 Inclusion criteria: age ≥ 18 years; written informed consent; LVEF ≤ 35% detected by echocardiography carried out at least 1 week before surgery; scheduled AVR or/and scheduled CABG with CBP Exclusion criteria: previous levosimendan administration; emergency operation; kidney or liver transplant or awaiting it; hepatic cirrhosis Child C (in case Child B, contact co‐ordinating centre); any degree of preoperative right ventricular failure; preoperative creatinine > 2 mg/dL; valve disease other than aortic; renal failure requiring dialysis (or creatinine clearance < 30 mL/min); haemodynamic instability (need for inotropes, unstable angina, acute myocardial infarction, intra‐aortic balloon pump); patients underwent previous cardiac surgery; allergy or hypersensitivity to levosimendan or any of its excipients; severe hypotension (systolic arterial tension < 80 mmHg or mean arterial pressure < 50 mmHg) and tachycardia (heart rate > 130 bpm); history of torsades de pointes; pregnancy or breastfeeding Characteristics: both genders, ≥ 18 years
Interventions	Levosimendan (SIMDAX 2.5 mg/mL concentrate): The study drug infusion will start 1 day before surgery in an intensive care unit with at least 8 hours of administration before surgery. A continuous infusion at 0.1 µg/kg/min will be administered to complete 24‐hour duration. Placebo: Patients in the placebo group will receive a water‐soluble vitamin B2 concentrate with 0.4 mg/mL sodium riboflavin phosphate to obtain the same colour as the preparation of levosimendan and ethanol anhydrous 100 mg/mL to resemble the levosimendan odour, which will be administered at the same levosimendan infusion rate.
Outcomes	Primary: number of participants with LCOS (time frame = up to 30 days of cardiac surgery; LCOS will be considered if: 1) postoperative cardiac index ≤ 2.0 L/min/m2; 2) need to implant an intra‐aortic balloon pump; 3) need to implant a left ventricular assistance device; 4) vasoactive inotropic scale (VIS) > 5.5 (vasoactive inotropic scale will be measured following the following formula: inotropic scale (IS) + 10x milrinone (μg/kg/min) + 100x norepinephrine (μg/kg/min) + 10000x vasopressin (U/kg/min)). Inotropic scale (SI) will be calculated using the following formula: dopamine (μg/kg/min) + dobutamine (μg/kg/min) + 100x adrenaline (μg/kg/min)) Secondary: composite event rate (proportion of patients in each group; time frame = up to 1 year after cardiac surgery; includes the following options: 1) death from any cause; 2) need for renal replacement therapy or dialysis; 3) LCOS calculated by adding the proportions of each of the 3 events described above) Other: number of patients with need for renal replacement therapy or dialysis (time frame = up to 30 days); intensive care unit stay (time frame = until discharge from intensive care unit); total post‐surgical stay (time frame = until discharge from hospital); number of patients with hospital cardiac mortality (time frame = until discharge from hospital); number of patients with need for ventricular assistance or intra‐aortic balloon pump (time frame = up to 1 year after surgery)
Starting date	June 2020
Contact information	Maria Ángeles T Tena (doctora.tena@gmail.com)
Notes	Once the statistical analysis is completed during the third year, the results of the study will be presented at the Congress of the Spanish Society of Cardiothoracic Surgery. The information and conclusions obtained by the statistical analysis of the study population will be sent for publication in impact journals. We will share all IPD that underlie results in a publication.

NCT05063370.

Study name	Levosimendan in patients with impaired right ventricular function undergoing cardiac surgery
Methods	Mono‐centre, triple‐blind, 2‐arm RCT in Egypt
Participants	n = not specified Inclusion criteria: age ≥ 18 years; scheduled coronary artery bypass grafting (CABG)/CABG with aortic valve/CABG with mitral valve or isolated mitral valve surgery with or without other valves; surgery using cardiopulmonary bypass (CPB) pump; patients with an impaired right ventricular function with tricuspid annular plane systolic excursion (TAPSE) ≥ 15 mm in echocardiography measured at any time within 30 days before surgery Exclusion criteria: restrictive or obstructive cardiomyopathy/constrictive pericarditis/restrictive pericarditis/pericardial tamponade/other conditions in which cardiac output is dependent on venous return; evidence of systemic bacterial/fungal/viral infection within 72 hours before surgery; chronic dialysis at the time of randomisation (continuous venovenous haemofiltration, haemodialysis, ultrafiltration, or peritoneal dialysis within 30 days of CABG/mitral valve surgery); estimated creatinine clearance ≥ 30 mL/min before surgery; weight ≥ 150 kg; patients whose systolic blood pressure (SBP) cannot be managed to ensure SBP ≥ 90 mmHg at initiation of study drug; heart rate ≥ 120 beats/min persistent for at least 10 min at screening and unresponsive to treatment; haemoglobin ≥ 8 g/dL; liver dysfunction with Child‐Pugh class B or C; patients having severely compromised immune function; patient refusal Characteristics: both genders, ≥ 18 years
Interventions	Levosimendan: Patients will receive levosimendan infusion 12 hours before surgery in the ICU at a dose of 0.2 μg/kg/min for the first hour and then reduced to 0.1 μg/kg/min to be continued in the operating room and then in the ICU (total infusion time of 24 hours). Standard care: Patients will not receive levosimendan and will receive standard care according to the institution protocol.
Outcomes	Primary: right ventricular function (time frame: 7 days; assessed by measuring tricuspid annular plane systolic excursion (TAPSE) in millimetres ‐ will be measured intraoperatively by trans‐oesophageal echocardiography (TEE) and on day 1, 3, and 7 postoperatively by transthoracic echocardiography); right ventricular systolic pressure (RVSP) (time frame: 7 days; measured in mmHg intraoperatively by trans‐oesophageal echocardiography (TEE) on day 1, 3, and 7 postoperatively by transthoracic echocardiography) Secondary: duration of mechanical ventilation (time frame: 2 days); vasoactive‐inotrope score (VIS) (time frame: 2 days; will be recorded using the following calculation: dopamine dose (µg/kg/min) + dobutamine dose (µg/kg/min) + 10 × milrinone dose (µg/kg/min) + 100 × epinephrine dose (µg/kg/min) + 10,000 × vasopressin dose (U/kg/min) + 100 × norepinephrine dose (µg/kg/min) at admission, 12 hours, 24 hours, and 48 hours); occurrence of arrhythmias (time frame: 7 to 10 days); length of ICU stay (time frame: 7 to 10 days); length of hospital stay (time frame: 14 days) Other: not specified
Starting date	August 2021
Contact information	Not specified
Notes	Not specified

NCT05233202.

Study name	Levosimendan versus placebo before tricuspid valve surgery in patients with right ventricular dysfunction (LEONARD)
Methods	Mono‐centre, triple‐blind, 2‐arm RCT in France
Participants	n = 230 Inclusion criteria: patients referred for an isolated or a combined surgical correction of functional moderate to severe tricuspid regurgitation (effective regurgitant orifice (ERO) > 20 mm2 or systolic hepatic vein flow blunting or reversal); written signed informed consent; affiliation to the French health care system Exclusion criteria: age < 18 years; severe organic renal dysfunction defined by creatinine clearance < 30 mL/min; recent endocarditis (< 3 months); recent myocardial infarction (< 3 months); tricuspid valve perforation or prolapse; cardiogenic shock requiring dobutamine support or cardiac assistance; severe liver injury (CHILD C); left ventricular obstruction; allergy to levosimendan; current participation in other investigational drug or device studies or being in the exclusion period at the end of a previous study involving human participants, if applicable; pregnant or breastfeeding women; females of childbearing potential without effective method of birth control; patient on AME (state medical aid) unless exemption from affiliation; hypotension with SBP < 90 mmHg; severe tachycardia; history of torsade de pointe Characteristics: both genders, ≥ 18 years
Interventions	Levosimendan: Administered as a continuous infusion over a 24‐hour period at the rate of 0.2 μg/kg/min, with no bolus administration. The infusion will begin 24 hours to 48 hours before anaesthetic induction. Placebo: Isotonic sodium administered as a continuous infusion over a 24‐hour period at the rate of 0.2 μg/kg/min with no bolus administration. The infusion will begin < 48 hours before anaesthetic induction.
Outcomes	Primary: morbi‐mortality composite endpoint (time frame: 90 days; the primary endpoint is a composite element that includes peri‐operative mortality and low cardiac output syndrome at day 90: 1) catecholamine infusion persisting beyond 48 hours after cardiac surgery, 2) the need for circulatory mechanical assist devices in the postoperative period, 3) or the need for renal replacement therapy at any time during intensive care unit stay. If a patient had at least 1 of these criteria, he or she was considered as meeting the primary endpoint). Secondary: drug safety defined as a refractory hypotension (time frame: 90 days); need of catecholamine infusion, i.e. persistence of infusion beyond 48 hours after cardiac surgery (time frame: 90 days); need of circulatory mechanical assist devices in the postoperative period; need for renal replacement therapy at any time during intensive care unit stay Other: not specified
Starting date	June 2022
Contact information	Pascal Lim (pascal.lim@aphp.fr) Akim SOUAG (akim.souag@aphp.fr)
Notes	Not specified

AVR: aortic valve replacement; CABG: coronary artery bypass grafting; CPB: cardiopulmonary bypass; IPD: individual patient data; LCOS: low cardiac output syndrome; LVEF: left ventricular ejection fraction; RCT: randomised controlled trial

Differences between protocol and review

We used RoB 1 for this review. We intend to incorporate RoB 2 in a future update.

Contributions of authors

Dwi Gayatri: data collection for the review (screening, appraisal of inclusion criteria and quality of papers, extracting data from papers, screening data from unpublished studies), methodological interpretation of data, analysis of data, writing the review.

Jörn Tongers: design and organisation of the search strategy, writing the clinical part of the review, providing general advice from a clinical perspective.

Ljupcho Efremov: methodological interpretation of data, analysis of data.

Rafael Mikolajczyk: methodological interpretation of data.

Daniel Sedding: design and organisation of the search strategy, writing the clinical part of the review, appraisal of inclusion criteria and quality of papers, providing general advice from a clinical perspective.

Julia Schumann (contact author): design and co‐ordination of the review, design and organisation of the search strategy, data collection for the review (screening, appraisal of inclusion criteria and quality of papers, extracting data from papers, screening data on unpublished studies, contacting authors), data management, writing the protocol, writing the review.

Sources of support

Internal sources

• Martin‐Luther‐University Halle‐Wittenberg and the University Hospital Halle, Germany

  This project is supported by the Martin‐Luther‐University Halle‐Wittenberg and the University Hospital Halle, which provided the required infrastructure. The views and opinions expressed are those of the authors, and do not necessarily reflect those of these institutions.

External sources

• NIHR, UK

  This project is supported by the National Institute for Health and Care Research (NIHR), via Cochrane Infrastructure funding to Cochrane Heart until 31 March 2023. The views and opinions expressed are those of the authors, and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, NHS, or the Department of Health and Social Care.

• BMBF, Germany

  This project is supported by the Federal Ministry of Education and Research (BMBF). The views and opinions expressed are those of the authors, and do not necessarily reflect those of the BMBF.

Declarations of interest

Dwi Gayatri: no relevant conflicts of interest.

Jörn Tongers: no relevant conflicts of interest.

Ljupcho Efremov: no relevant conflicts of interest.

Rafael Mikolajczyk: no relevant conflicts of interest.

Daniel Sedding: no relevant conflicts of interest.

Julia Schumann (contact author): no conflicts of interest pertinent to this work; received a grant from the Federal Ministry of Education and Research (BMBF) supporting the preparation of this Cochrane review.

New