Impella was not associated with a statistically significant difference in in-hospital mortality compared to VA-ECMO in the primary crude analysis (RR 0.84; 95% CI 0.66-1.07; p=0.15).
Meta-Analysis (n=22,618)
Does Impella reduce in-hospital mortality in adult patients with cardiogenic shock compared to VA-ECMO?
In observational data, Impella showed a neutral effect on in-hospital mortality overall but was associated with lower mortality and fewer vascular/bleeding complications in matched-cohort analyses compared to VA-ECMO, though limited by confounding.
Effect estimate: RR 0.84 (95% CI 0.66-1.07)
p-value: p=0.15
Observational studies comparing Impella and venoarterial extracorporeal membrane oxygenation (VA-ECMO) in cardiogenic shock have reported inconsistent findings. We performed an updated systematic review and meta-analysis, with exploratory matched-cohort analyses where directly reportable propensity score-matched event-level data were available. We searched PubMed, Scopus, Web of Science, and the Cochrane Library from inception through March 2026 for observational studies comparing Impella versus VA-ECMO in adult patients with cardiogenic shock. The primary outcome was in-hospital mortality. Secondary outcomes included all-cause mortality, ICU outcomes, access site bleeding requiring transfusion, peripheral vascular complications, renal outcomes, and stroke. Random-effects models using restricted maximum likelihood were applied. Fifteen observational studies including 22,618 patients were analyzed. In the primary crude analysis, Impella was not associated with a statistically significant difference in in-hospital mortality (RR 0.84, 95% CI 0.66-1.07; p = 0.15) with substantial heterogeneity (I² = 92.9%). However, leave-one-out analysis identified an influential registry study; its exclusion markedly reduced heterogeneity and favored Impella (p < 0.01). In exploratory propensity score-matched analysis, Impella was also associated with lower in-hospital mortality (RR 0.69, 95% CI 0.56-0.85, p < 0.01). Thirty-day and 6-month all-cause mortality were neutral, whereas 12-month mortality modestly favored Impella (p = 0.048). Sensitivity analyses similarly resolved heterogeneity and favored Impella for selected secondary outcomes, including 6-month mortality, acute kidney injury, and ischemic stroke. ICU mortality was neutral, while ICU length of stay was shorter with Impella (p = 0.01). Impella was also associated with lower access site bleeding requiring transfusion (p < 0.01), peripheral vascular complications (p < 0.01), and hemorrhagic stroke (p < 0.01). In this updated meta-analysis of observational studies, the primary crude analysis of in-hospital mortality was neutral overall but highly heterogeneous, whereas sensitivity and exploratory matched-cohort analyses showed a more consistent association favoring Impella. Impella was also associated with lower bleeding and vascular complications and shorter ICU length of stay, although several outcomes remained sensitive to study-level influence. These findings should be interpreted cautiously given the observational design, predominance of serious risk of bias, and strong potential for confounding by indication.
Mohamed et al. (2026) conducted a meta-analysis in cardiogenic shock (n=22,618). Impella vs. VA-ECMO was evaluated on in-hospital mortality (RR 0.84, 95% CI 0.66-1.07, p=0.15). Impella was not associated with a statistically significant difference in in-hospital mortality compared to VA-ECMO in the primary crude analysis (RR 0.84; 95% CI 0.66-1.07; p=0.15).