Nineteen RCTs were included (3,650 patients). Only two fulfilled all quality criteria. Five RCTs reported adequate allocation concealment, 12 used intention-to-treat analysis, and seven used blinding. Duration of follow-up ranged from one to 180 days (where reported).
There was no statistically significant difference in mortality between the levosimendan and placebo groups (six RCTs; 1,578 patients), with no substantial statistical heterogeneity.
Levosimendan was associated with significantly lower mortality than dobutamine, with some statistical heterogeneity (OR 0.75, 95% CI 0.61 to 0.92; I2=44.6%, eight RCTs; 1,979 patients). Subgroup analyses showed a stronger effect in lower quality trials.
Dobutamine was associated with significantly higher mortality than placebo (OR 1.82, 95% CI 1.06 to 3.12; I2=0%; three RCTs; 301 patients).
There was no significant difference in mortality between the levosimendan and milrinone groups (two RCTs).
Levosimendan improved haemodynamic parameters compared with controls. Effects were statistically significant for ejection fraction compared with placebo, and for pulmonary artery occlusion pressure and B-type natureitic peptide compared with either placebo or dobutamide. However data were limited, with only two or three RCTs for each comparison.
There was no evidence of significant publication bias. Findings were similar when random-effects models were used.