Interventional procedure overview of surgical insertion of a catheter-based intravascular microaxial flow pump for cardiogenic shock
Closed for comments This consultation ended on at Request commenting lead permission
Evidence summary
Population and studies description
This interventional procedures overview is based on about 6,400 people from 1 systematic review and meta-analysis (Kwon 2024), 1 prospective multicentre observational study (Abraham 2025) and 6 retrospective observational studies (Fried 2024, Ramzy 2023, Iyengar 2023, Gill 2023, Mahesh 2024 and Schumer 2024). The study by Gill (2023) was also included in the systematic review by Kwon (2024). Of the 6,400 people, about 3,000 had the procedure. In addition, there was a retrospective analysis of the US FDA MAUDE database describing 43 events associated with the procedure (Khalil 2025). This interventional procedures overview is a rapid review of the literature, and a flow chart of the complete selection process is shown in figure 1. This overview presents 9 studies as the key evidence in table 2 and table 3, and lists 75 other relevant studies in appendix B, table 5.
Evidence was included if the indication was described as cardiogenic shock, from any aetiology, regardless of whether PCI was offered as part of the treatment pathway. Evidence was excluded from studies that primarily used microaxial flow pumps as support during high-risk PCI, or for left ventricular unloading during VA-ECMO support.
Studies were included if they described surgical insertion of a catheter-based intravascular microaxial flow pump or if they specified use of Impella 5.5, which is inserted surgically. In general, evidence on Impella 5.0 alone was not included in this review unless it specifically stated it was inserted surgically. Most of the studies were based in the US and they all had a higher proportion of males than females. The mean or median age was typically above 55 years old.
The systematic review by Kwon (2024) included 707 people from 15 studies on Impella 5.5, 2 of which were prospective. Some of the studies included a proportion of people who had other Impella devices, so the number who had Impella 5.5 was 440, with a further 156 people having either 5.0 or 5.5. The main indication for the procedure across the studies was heart failure complicated by cardiogenic shock, followed by AMICS and postcardiotomy cardiogenic shock. The pooled mean age was 58.2 years and 86% of the study population was male. In 5 studies that reported it, the proportion of people who needed cardiopulmonary resuscitation before the procedure was 20%. Follow-up was to hospital discharge or 30 days.
The prospective, multicentre study by Abraham (2025) included 444 people with heart-failure related cardiogenic shock. The mean age was 55.1 years and 86% were male. Most people (94%) were in NYHA class 3 or 4 at baseline and the median LVEF was 15%. Before having Impella 5.5, 48% of people were supported by other MCS devices, most commonly an IABP or Impella CP. Follow-up was 12 months.
The study by Fried (2024) reported a retrospective analysis of data from a multicentre US registry. The most common aetiology was heart failure related cardiogenic shock, followed by AMICS. Of the 754 people included, 620 had Impella 5.5. The mean age was 58.6 years and 83% were male. Most people were SCAI stage D or E at baseline and the mean LVEF was 20%. Nearly half (46%) had at least 1 temporary MCS before surgical insertion of a microaxial flow pump. Follow-up was to hospital discharge.
The retrospective study by Ramzy (2023) included multicentre registry data from 1,238 people who had either Impella 5.5 or Impella 5.0. The indications were AMICS, cardiomyopathy or postcardiotomy cardiogenic shock. People who had ECMO before or during Impella use were excluded. Follow-up was to device explant only.
The retrospective study by Iyengar (2023) included 2,839 people who were on a waiting list for heart transplantation, with support from Impella 5.5 or IABP. The median age was younger in the Impella group (56 years) than the IABP group (58 years; p=0.002) and there was a statistically significantly higher proportion of males (87% versus 75%, p<0.001). At baseline, the functional status was described as severe in a higher proportion of those who had Impella compared to IABP (73% versus 64%, p<0.001). Follow-up was 2 years.
Gill (2023) and Mahesh (2024) both described retrospective analyses of prospectively collected registry data. In Gill (2023), 221 people had Impella 5.5 or 5.0 as a bridging strategy to transplant, durable LVAD or recovery. The aetiology of cardiogenic shock was non-ischaemic (51%) or ischaemic (23%) cardiomyopathy or AMICS (26%). The median age was 58 years and 86% were male. At baseline, the median LVEF was 15% and most people were IP-1 or IP-2. Half the study population were supported with another MCS device (Impella CP, IABP or ECMO) before implantation of the microaxial flow pump. A small proportion of people had concomitant ECMO or right VAD. The median follow-up period was 4 months. In Mahesh (2024), 107 people had Impella 5.5 as BTT, BTDD or BTR. A proportion of people had concomitant percutaneous right VAD support and 20 people were transitioned from ECMO support. The mean follow-up was 23 months, with a maximum of 4.5 years.
Schumer (2024) described a single-centre retrospective cohort study of 126 people who had the procedure for cardiogenic shock (including postcardiotomy cardiogenic shock), BTDD, BTT or planned support for high-risk cardiac surgery. The causes of cardiogenic shock were predominantly acute on chronic non-ischaemic cardiomyopathy, AMI and ischaemic cardiomyopathy. The median age was 62 years and 77% were male. At baseline, 32% of people had experienced cardiopulmonary resuscitation.
Khalil (2025) is a review of the FDA MAUDE database, describing 43 adverse events resulting in death, reported in association with use of Impella 5.5. The MAUDE database includes mandatory reports from manufacturers and device importers when a device may have caused injury to a patient, and voluntary reports from other sources including healthcare professionals and patients. Limitations of the database include under-reporting, duplicate reporting, incomplete reports and uncertainty if the device caused the complication being described. The true denominator for these events is not captured and the database is not designed to calculate or compare complication rates.
In addition to the key evidence summarised in table 2 and table 3, individual case reports describing adverse events associated with the procedure have been listed in table 5a in appendix B.
Table 2 presents study details.
Procedure technique
All the studies used the Impella 5.5 device (Abiomed) but some also used Impella 5.0 (n=1,045). In the studies that reported it, vascular access was predominantly through the axillary artery. A direct aortic access was only used in a small proportion of people. In the study by Ramzy (2023), which aimed to compare Impella 5.5 with Impella 5.0, insertion site was similar in people with AMICS and cardiomyopathy. People with postcardiotomy cardiogenic shock who had Impella 5.5 were more likely to have the device placed through the ascending aorta (40% versus 11%; p=0.001), whereas people with Impella 5.0 were more likely to have axillary access (76% versus 55%; p=0.031). In the registry study by Fried (2024), which included Impella 5.5 or 5.0, cannulation sites were available for 69% (524 out of 754) of people, with 94% axillary configuration.
Previous or concomitant support with other temporary MCS devices, including VA-ECMO, IABPs and right VADs, was commonly reported. The mean or median duration of Impella support varied from 7 to 27 days, in the studies that reported it.
Efficacy
Survival to device explant, hospital discharge or 30 days
Survival to hospital discharge or 30 days was reported in 6 studies. One study reported survival to device explant only.
In the systematic review of 15 studies (Kwon 2024), survival to discharge was 68% (95% CI 58 to 78, 11 studies, I2=80%) and 30-day survival was 65% (95% CI 56 to 74, 9 studies, I2=77%). For people who had support with Impella 5.5 only (n=294) rather than any Impella device, survival to discharge was 78% (95% CI 72 to 82, 8 studies, I2=0%).
In the prospective, multicentre observational study of 444 people with cardiogenic shock caused by acute decompensated heart failure treated by Impella 5.5 alone or by multiple MCS devices (Abraham 2025), overall survival to discharge was 75%. For people who had support with Impella 5.5 only (n=207), survival to discharge was 86% (95% CI 82 to 91). For people who had multiple temporary MSC devices (n=237), survival to discharge was 65% (95% CI 59 to 71).
In the retrospective registry study of 754 people with cardiogenic shock that was mostly heart failure-related, including acute on chronic heart failure, or AMICS (Fried 2024), in-hospital mortality was 33% (248 out of 754). Native heart survival was 20% (154 out of 754) and 46% (341 out of 754) of people were bridged to heart transplantation or durable VAD. When Impella 5.0 or 5.5 was used in isolation, in-hospital mortality was lower than when another MCS device was used (20% versus 39%, p<0.001) and heart transplantation or durable VAD implantation was higher (59% versus 43%, p<0.001). The Kaplan-Meier survival curve showed worse survival in people who had an out-of-hospital cardiac arrest (p<0.0001) or whose baseline lactate was 4 mmol/litre or above (p<0.0003). There was higher mortality in the AMICS group compared to those who had cardiogenic shock associated with decompensated heart failure (45% versus 26%, p<0.001). Among the survivors, a higher proportion of people with AMICS had native heart recovery compared to those in the group with heart failure related cardiogenic shock (30% versus 17%; p<0.001). Of the 341 people who had heart transplantation or durable VAD, most were from the decompensated heart failure group rather than the AMICS group (80% versus 14%, p<0.001).
The retrospective registry study by Gill (2023) included 221 people with non-ischaemic cardiomyopathy, ischaemic cardiomyopathy or AMICS, who had Impella 5.5 or 5.0 as a bridging therapy. Overall survival to discharge was 69% (152 out of 221). Survival in the 2 groups were similar when the most recent 75 people who had Impella 5.0 were compared with those who had Impella 5.5 (72% versus 77%, p=0.45). Overall 30-day survival was 73% (95% CI 67 to 79) and there was no statistically significant difference between the 2 groups (p=0.94).
The retrospective registry study by Mahesh (2024) included 107 people who were bridged with Impella 5.5 to transplant, durable device implantation or recovery. In-hospital mortality was 6% (2 out of 34) in the BTT group, 4% (1 out of 25) in the BTDD group and 24% (10 out of 42) in the BTR group (p=0.021). In the logistic regression model, the category of cardiogenic shock was statistically significantly associated with mortality with worse survival in the postcardiotomy cardiogenic shock group compared to the BTT group (OR 4.7, 95% CI 0.9 to 24, p=0.05).
In the single-centre retrospective cohort study by Schumer (2024), 126 people had Impella 5.5 as planned support for high-risk cardiac surgery, cardiogenic shock, BTDD, BTT or postcardiotomy cardiogenic shock. The causes of heart failure leading to cardiogenic shock were predominantly acute on chronic non-ischaemic cardiomyopathy, AMI and ischaemic cardiomyopathy. Overall survival to device explant was 76% (96 out of 126) and survival to discharge was 68% (85 out of 126).
The retrospective multicentre registry study by Ramzy (2023) included 1,238 people with AMICS, cardiomyopathy or postcardiotomy cardiogenic shock who had implantation of Impella 5.5 or 5.0. Survival to device explant was stratified by cardiogenic shock aetiology and device. In all groups, Impella 5.5 was associated with a statistically significantly higher survival than Impella 5.0. In people with AMICS, 30% (46 out of 156) of those who had Impella 5.5 died on support or had care withdrawn compared to 43% (120 out of 278) of those who had Impella 5.0 (p=0.005). In the cardiomyopathy group, 12% (32 out of 270) of those who had Impella 5.5 and 23% (52 out of 225) of those who had Impella 5.0 died on support or had care withdrawn (p=0.001). In the postcardiotomy group, 24% (28 out of 117) of people who had Impella 5.5 and 44% (39 out of 88) of people who had Impella 5.0 died on support or had care withdrawn (p=0.003).
Successful weaning or bridged to heart transplantation or durable LVAD
The rate of successful weaning or bridging to heart transplantation or durable LVAD was reported in 5 studies.
In the systematic review of 15 studies (Kwon 2024), 10% of people were bridged to transplant directly from Impella support.
In the prospective, multicentre observational study of 444 people, 35% (55 out of 444) of people were bridged to transplant (Abraham 2025).
In the retrospective multicentre registry study by Ramzy (2023), 70% (110 out of 156) of people with AMICS who had Impella 5.5 and 57% (158 out of 278) of people who had Impella 5.0 had successful weaning or bridge to heart transplantation or durable LVAD (p=0.005). In the cardiomyopathy group, the rates were 88% (238 out of 270) for people who had Impella 5.5 and 77% (173 out of 225) for those who had Impella 5.0 (p=0.001). In the postcardiotomy group, the rates were 76% (89 out of 117) and 56% (49 out of 88), respectively (p=0.003).
In the retrospective cohort study by Iyengar (2023), 2,839 people on a waiting list for heart transplantation were supported by Impella 5.5 (n=452) or IABP (n=2,484). At baseline people with Impella support had more functional impairment, higher wedge pressures, higher rates of preoperative diabetes and dialysis, and more ventilator support (all p<0.05). Mortality while on the waitlist was 9% (42 out of 452) in the Impella 5.5 group and 2% (60 out of 2,484) in the IABP group (p<0.001). Transplantation was done in 85% (383 out of 452) of those who had Impella and 92% (2,278 out of 2,484) of those who had IABP (p<0.001).
In the retrospective registry study by Gill (2023), survival to bridging destination was 76% (n=167).
Longer term survival
Survival beyond 30 days was reported in 5 studies.
In the prospective, multicentre observational study of 444 people with cardiogenic shock caused by acute decompensated heart failure (Abraham 2025), native heart survival was 71% at 6 months and 64% at 12 months. For people who had heart transplantation or durable LVAD, survival was 95% at 6 months and 94% at 12 months. For people who had support with Impella 5.5 only, survival at 6 months was 78% (95% CI 73 to 84%) and at 12 months it was 77% (95% CI 71 to 83%). For people who had multiple temporary MSC devices, survival was 56% (95% CI 50 to 62%) at 6 months and 54% (95% CI 47 to 60%) at 12 months.
In the retrospective cohort study by Iyengar (2023) of 2,839 people on a waiting list for heart transplantation supported by Impella 5.5 (n=452) or IABP (n=2,484), mortality within 1 year of transplant was 5% and 6%, respectively (p=0.877). In a propensity-matched cohort of 628 people, rates were 6% in both groups. In the matched cohort, 2-year survival was 83% in the Impella group and 88% in the IABP group (p=0.874).
In the retrospective registry study by Gill (2023), survival at 90 days after implantation was 81% (95% CI 73 to 88) in the BTT group, 52% (95% CI 34 to 70) in the BTDD group and 51% (95% CI 39 to 62) in the BTR group. At 1 year, survival was 73% (95% 64 to 82) in the BTT group, 47% (95% CI 29 to 64) in the BTDD group and 39% (95% CI 28 to 51) in the BTR group.
In the retrospective registry study by Mahesh (2024), mortality during follow-up (mean 23 months) was 9% in the BTT group (3 out of 34), 20% (5 out of 25) in the BTDD group and 36% (15 out of 42) in the BTR group (p=0.019). Actuarial survival at 4.5 years was 91% for BTT, 79% for BTDD and 63% for BTR (p=0.01). The category of cardiogenic shock (HR 3.63, 95% 1.03 to 12.9, p=0.04) and long-term postoperative dialysis (HR 3.9, 95% CI 1.6 to 9, p=0.002) were statistically significant predictors of long-term mortality.
In the single-centre retrospective cohort study by Schumer (2024), overall survival at 6 months was 57% (60 out of 106).
Unsuccessful device placement
In the retrospective multicentre registry study by Ramzy (2023), 6% (10 out of 169) of people with AMICS who had Impella 5.5 and 5% (15 out of 305) of people who had Impella 5.0 had aborted device placement (p=0.671). In the cardiomyopathy group, aborted device placement was reported in 4% (11 out of 287) of people who had Impella 5.5 and 3% (7 out of 245) of those who had Impella 5.0 (p=0.634). In the postcardiotomy group, the rates were 5% (6 out of 126) and 13% (14 out of 106), respectively (p=0.033).
Ambulation
In the retrospective registry study by Gill (2023), 24% (42 out of 221) of people were ambulatory within 24 hours of Impella implantation. By device, the rate was 17% (13 out of 75) for Impella 5.5 and 27% (20 out of 75) for Impella 5.0 (p=0.17).
Duration of support
Mean or median duration of support was reported in 4 studies.
In the prospective, multicentre observational study of 444 people (Abraham 2025), the median duration of support for people bridged to transplant was 18 days.
In the retrospective registry study of 754 people (Fried 2024), the median duration of support was 12.9 days in the 337 people where both the device implant and explant date and time were available.
In the retrospective multicentre registry study by Ramzy (2023), the mean duration of support for people with AMICS was 13.2 days in the Impella 5.5 group and 8.7 days in the Impella 5.0 group (p=0.008). In the cardiomyopathy group, the mean duration of support was 15.1 days for people who had Impella 5.5 and 11.4 days for those who had Impella 5.0 (p<0.001). In the postcardiotomy group, the mean durations were 10.2 and 6.6 days, respectively (p=0.127).
In the retrospective cohort study by Iyengar (2023) of 2,839 people on a waiting list for heart transplantation, the median time on support was 15 days for Impella and 9 days for IABP (p<0.001).
Safety
Bleeding
Bleeding was reported as an outcome in all studies.
Bleeding was the most common complication reported across the studies in the systematic review by Kwon (2024). In 11 studies, the rate of bleeding with varying definitions ranged from 9% to 41% of people. Access site bleeding needing re-exploration was reported in 10% of people (6 studies, n=422). Bleeding events of any type were reported in 27% (95% CI 23 to 31) of people in the observational study of 444 people (Abraham 2025). The rate was 24% (95% CI 18 to 30) in the 207 people who were supported with Impella 5.5 alone. Major bleeding was reported in 46% (349 out of 754) of people in the retrospective registry study of 754 people (Fried 2024). When Impella 5.0 or 5.5 was used in isolation, the rate was 34%. Major bleeding was more common in people who had AMICS (58%) compared to heart failure related cardiogenic shock (39%, p<0.001).
In people who had AMICS, the rate of bleeding was 1% (1 out of 156) in the Impella 5.5 group and 2% (5 out of 278) in the Impella 5.0 group (p=0.426) in the retrospective study by Ramzy (2023). In people who had cardiomyopathy related cardiogenic shock, the rate was 1% (3 out of 270) in the Impella 5.5 group and 2% (5 out of 225) in the Impella 5.0 group (p=0.478). In people who had postcardiotomy cardiogenic shock, the rate was 3% (3 out of 117) in the Impella 5.5 group and 7% (6 out of 88) in the Impella 5.0 group (p=0.177).
Bleeding needing return to theatre was reported in 8% (17 out of 221) of people overall, 11% (8 out of 75) in the Impella 5.5 group and 8% (6 out of 75) in the Impella 5.0 group (p=0.57) in the retrospective registry study by Gill (2023).
Axillary haematoma was reported in 15% (5 out of 34) of people in the BTT group, 12% (3 out of 25) of people in the BTDD group and 5% (2 out of 42) of people in the BTR group (p=0.32) in the retrospective registry study by Mahesh (2024). Gastrointestinal bleed was reported in 6% (2 out of 34) of people in the BTT group, 4% (1 out of 25) of people in the BTDD group and 14% (6 out of 42) of people in the BTR group (p=0.3).
Reoperation for bleeding was reported in 14% (17 out of 126) of people in the single-centre retrospective cohort study by Schumer (2024).
Vascular complications
The rate of vascular complications was reported as an outcome in 2 studies.
Major complications, with varying definitions, were reported in 4 studies and ranged from 0% to 15% in the systematic review by Kwon (2024).
Vascular injury was reported in 1 person with AMICS in the Impella 5.5 group and 1 person with cardiomyopathy in the Impella 5.0 group in the retrospective study of 1,238 people by Ramzy (2023).
Stroke
Stroke was reported as an outcome in all studies.
The incidence of stroke ranged from 0% to 14% in 11 out of the 15 studies included in the systematic review by Kwon (2024). The highest incidence was in a study reporting support with Impella combined with ECMO (known as ECPELLA). Stroke was reported in 4% (95% CI 2 to 6) of people in the observational study of 444 people (Abraham 2025). The proportion was 3% (95% CI 1 to 7) in the 207 people who were supported with Impella 5.5 alone.
Stroke was reported in 7% (51 out of 754) of people in the retrospective registry study of 754 people (Fried 2024). When Impella 5.0 or 5.5 was used in isolation, the proportion was 5% (actual numbers not reported).
In people who had AMICS, cerebrovascular accident was reported in 3% (5 out of 156) of people in the Impella 5.5 group and 1% (3 out of 278) of people in the Impella 5.0 group (p=0.143) in the retrospective study by Ramzy (2023). In people who had cardiomyopathy, the proportion was 2% (6 out of 270) in the Impella 5.5 group and 1% (2 out of 225) in the Impella 5.0 group (p=0.301). In people who had postcardiotomy cardiogenic shock, the proportion was 2% (2 out of 117) in the Impella 5.5 group and 1% (1 out of 88) in the Impella 5.0 group (p>0.99).
Ischaemic stroke was reported in 3% (6 out of 221) of people overall, 3% (2 out of 75) in the Impella 5.5 group and 3% (2 out of 75) in the Impella 5.0 group (p=1.00) in the retrospective registry study by Gill (2023).
Ischaemic stroke was reported in 3% (1 out of 34) of people in the BTT group, 8% (2 out of 25) of people in the BTDD group and 19% (8 out of 42) of people in the BTR group (p=0.07) in the retrospective registry study by Mahesh (2024). Of the 11 people, 8 had complete resolution of neurological disabilities and 3 had residual isolated limb weakness or dysphasia.
Stroke was reported in 10% (13 out of 126) of people in the single-centre retrospective cohort study by Schumer (2024).
Haemolysis
Haemolysis was reported as an outcome in 5 studies.
In the systematic review by Kwon (2024), the incidence of haemolysis, with varying definitions, was reported in 9 studies and ranged from 0% to 23%. Clinically significant haemolysis was reported in 14% (95% CI 10 to 17) of people in the observational study of 444 people (Abraham 2025). The proportion was 13% (95% CI 9 to 18) in the 207 people who were supported with Impella 5.5 alone. Haemolysis was reported in 22% (166 out of 754) of people in the retrospective registry study of 754 people (Fried 2024). When Impella 5.0 or 5.5 was used in isolation, the proportion was 17%. Haemolysis was reported in 26% (46 out of 180) of people overall, 23% (14 out of 62) in the Impella 5.5 group and 30% (20 out of 66) in the Impella 5.0 group (p=0.18) in the retrospective registry study by Gill (2023).
In people who had AMICS, haemolysis was reported in 3% (5 out of 156) of people in the Impella 5.5 group and 4% (10 out of 278) in the Impella 5.0 group (p>0.99) in the retrospective study by Ramzy (2023). In people who had cardiomyopathy, the proportion was 3% (8 out of 270) in the Impella 5.5 group and 9% (21 out of 225) in the Impella 5.0 group (p=0.003). In people who had postcardiotomy cardiogenic shock, the proportion was 2% (2 out of 117) in the Impella 5.5 group and 1% (1 out of 88) in the Impella 5.0 group (p>0.99).
Limb ischaemia
Limb ischaemia was reported as an outcome in 3 studies.
Acute limb ischaemia was reported in 1% (95% CI 0.4 to 3) of people in the observational study of 444 people (Abraham 2025) and 0% (95% CI 0 to 2) in the 207 people who were supported with Impella 5.5 alone.
Limb ischaemia was reported in 7% (53 out of 754) of people in the retrospective registry study of 754 people (Fried 2024). When Impella 5.0 or 5.5 was used in isolation, the proportion was 3%. Limb ischaemia was more common in people who had AMICS (12%) compared to heart failure related cardiogenic shock (4%, p<0.001).
Limb ischaemia was reported in 2% (4 out of 221) of people overall, 1% (1 out of 75) in the Impella 5.5 group and 0% (0 out of 75) in the Impella 5.0 group (p=1.00) in the retrospective registry study by Gill (2023).
Acute kidney injury or renal failure needing renal replacement therapy
The need for renal replacement therapy was reported as an outcome in 5 studies.
Acute kidney injury with need for renal replacement therapy was reported in 10% (95% CI 8 to 14) of people in the observational study of 444 people (Abraham 2025). The rate was 5% (95% CI 3 to 9) in the 207 people who were supported with Impella 5.5 alone. New renal replacement therapy for acute renal failure was reported in 35% (265 out of 754) of people in the retrospective registry study of 754 people (Fried 2024). When Impella 5.0 or 5.5 was used in isolation, the proportion was 24%. New renal failure needing dialysis was reported in 22% (49 out of 221) of people overall, 24% (18 out of 75) in the Impella 5.5 group and 19% (14 out of 75) in the Impella 5.0 group (p=0.43) in the retrospective registry study by Gill (2023).
Dialysis was reported in 12% (4 out of 34) of people in the BTT group, 20% (5 out of 25) of people in the BTDD group and 29% (12 out of 42) of people in the BTR group (p=0.195) in the retrospective registry study by Mahesh (2024).
New dialysis was reported in 28% (35 out of 126) of people in the single-centre retrospective cohort study by Schumer (2024).
Infection
Infection was reported as an outcome in 2 studies.
Bacteraemia or sepsis was reported in 5% (95% CI 3 to 7) of people in the observational study of 444 people (Abraham 2025). The proportion was 3% (95% CI 1 to 7) in the 207 people who were supported with Impella 5.5 alone. In the same study, infection was reported in 17% (95% CI 13 to 20) of people in the whole cohort and 14% (95% CI 10 to 20) of those who had Impella 5.5 alone.
Local infection was reported in 3% (4 out of 126) of people in the single-centre retrospective cohort study by Schumer (2024).
Arrhythmia or heart failure
Destabilising arrhythmia was reported in 33% (95% CI 29 to 38) of people in the observational study of 444 people (Abraham 2025). The proportion was 28% (95% CI 22 to 35) in the 207 people who were supported with Impella 5.5 alone. In the same study, right heart failure was reported in 5% (95% CI 4 to 8) of the whole cohort and 3% (95% CI 1 to 7) of those who had Impella 5.5 alone.
Cardiac arrest
In-hospital cardiac arrest was reported in 20% (154 out of 754) of people in the retrospective registry study of 754 people (Fried 2024). When Impella 5.0 or 5.5 was used in isolation, the proportion was 9% (actual numbers not reported).
Thrombocytopenia
Thrombocytopenia needing platelet transfusion was reported in 13% (95% CI 10 to 16) of people in the observational study of 444 people (Abraham 2025). The proportion was 5% (95% CI 2 to 9) in the 207 people who were supported with Impella 5.5 alone.
Device dislodgement
In the systematic review by Kwon (2024), the incidence of device dislodgement was reported in 5 studies and ranged from 0% to 22%.
Pump thrombosis
In the systematic review by Kwon (2024), the incidence of pump thrombosis was reported in 5 studies and ranged from 0% to 15%.
Device exchange
Device exchange (not otherwise described) was needed in 8% (18 out of 221) of people overall, 4% (3 out of 75) in the Impella 5.5 group and 13% (10 out of 75) in the Impella 5.0 group (p=0.04) in the retrospective registry study by Gill (2023).
Adverse events reported on the FDA MAUDE database
The review by Khalil (2025) searched the FDA MAUDE database for reports of events associated with Impella 5.5 that resulted in death. Of the 43 events identified, 26 were described as procedural, 14 were device-related and the cause was uncertain in 3. The most common procedural complication was cardiac perforation, which was often attributed to improper positioning or repositioning of the device within the left ventricle. Other procedural complications were graft or vessel-related issues, such as bleeding, and valvular damage of device mispositioning. Device-related complications included device malfunction with pump stoppage and clot formation and embolic stroke.
Case reports
A number of case reports describing individual adverse events associated with Impella 5.0 or 5.5 are listed at the beginning of table 5. These include infected pseudoaneurysm of an outflow graft, aortic valve leaflet injury, entrapment of the Impella pump in the mitral subvalvular apparatus, coiled Impella drive line in the left ventricle, and severe aortic regurgitation.
MHRA Field Safety Notice
An MHRA field safety notice was issued for all Impella heart pumps in April 2024. In summary, information on safe use of Impella pumps had been issued with 2 technical bulletins, but the instructions for use were not updated to include the same level of detail covered in the bulletins and 1 of the bulletins was not distributed to European customers. These included a technical bulletin for operator mishandling of the Impella left-sided devices resulting in iatrogenic ventricular wall perforation and an Impella Product Update for an issue with fibres entrapped in the impeller. The action to mitigate the risk was for users to take note of amendment and reinforcement of instructions for use.
Anecdotal and theoretical adverse events
Expert advice was sought from consultants who have been nominated or ratified by their professional society or royal college. They were asked if they knew of any other adverse events for this procedure that they had heard about (anecdotal), which were not reported in the literature. They were also asked if they thought there were other adverse events that might possibly occur, even if they had never happened (theoretical).
The anecdotal and theoretical adverse events listed have all been reported in the literature and are described elsewhere in the overview.
Eight professional expert questionnaires were submitted for surgical insertion of a catheter-based intravascular microaxial flow pump for cardiogenic shock. Find full details of what the professional experts said about the procedure in the https://www.nice.org.uk/guidance/awaiting-development/gid-ipg10461/documents.
Validity and generalisability
There were no randomised controlled trials identified.
Most of the evidence is from the US and it may not be generalisable to practice within the UK. Most studies were observational and retrospective, which have more potential for bias than randomised controlled trials.
The evidence is likely to include early experience with Impella 5.5 for most study centres. There is likely to be a learning curve for this procedure and outcomes may improve with experience.
Most of the studies include cardiogenic shock from a variety of causes. Some of the evidence has reported separate outcomes by aetiology, which suggests there are differences in safety and efficacy.
There was high study heterogeneity for outcomes reported in the systematic review (Kwon 2024), likely driven by highly selected patient populations and inclusion of other Impella devices (mostly Impella 5.0) in some studies.
The systematic review included a high proportion of males (86%) and the aetiology was mainly heart failure complicated by cardiogenic shock, primarily due to cardiomyopathy. The authors note that outcomes may be better in this population compared with people who have AMICS or postcardiotomy cardiogenic shock.
Registry studies may use data that has been collected for other purposes and may have missing data, such as baseline characteristics to define the stage of cardiogenic shock. Also, they may not distinguish between different models of Impella pumps. This is particularly important when they are used to compare different MCS devices. Also, there may not be sufficient data to establish a temporal relationship between adverse events and the use of a microaxial flow pump or other temporary MCS devices.
In the retrospective registry by Fried (2024), data from Impella 5.0 and Impella 5.5 were combined in most of the analyses because of their similarity as large-bore transvalvular micro-axial heart pumps. The authors note that there are improvements in the design of the Impella 5.5 and it may be superior to the Impella 5.0.
The studies by Gill (2023) and Schumer (2024) were done at single high-volume centres and they might not be generalisable to all centres. Safety and efficacy outcomes, such as vascular complication rates, are likely to be affected by the expertise of the study centre.
In the study by Schumer (2024), the authors noted that for people with cardiogenic shock, only those who survived long enough to reach the study centre were able to have Impella 5.5 implantation, which introduced selection bias.
Definitions varied for outcomes such as bleeding and haemolysis.
People commonly had other MCS devices either before, after or at the same time as the catheter-based intravascular microaxial flow pump.
There was no quality-of-life evidence identified.
Several studies declared potential conflicts of interest. One of the authors of the systematic review by Kwon (2024) is a speaker and consultant for Abiomed, 3ive, LivaNova, and Abbott, another is employed by Abiomed and a third author is a consultant for Abiomed. The study by Fried (2024) was supported by institutional grants from Abiomed Inc, Boston Scientific Inc, Abbott Laboratories, Getinge Inc, and LivaNova Inc to Tufts Medical Center. Several of the authors have received consulting honoraria, institutional grant support, or have worked on the advisory board or acted as consultants for companies including Abiomed. For Ramzy (2023), several authors report honoraria, consulting fees or funding from companies including Abiomed, and auxiliary medical writing services were provided by the device manufacturer. For Gill (2023), 1 author has received honoraria from Abiomed. For Schumer (2024), 1 author is a speaker for Abbott and Abiomed and receives honoraria from both and 1 author is a speaker for Abiomed and does not receive honoraria.
Impella 5.0 is no longer available for use in the NHS.
Ongoing trials
IMPELLA, Complications and Tolerance (IMPACT; NCT06644963); Observational study; France; n=800; estimated study completion: January 2025
Cardiogenic Shock Working Group Registry (CSWG; NCT04682483); cohort study; US; n=5,000; estimated study completion: June 2026
How are you taking part in this consultation?
You will not be able to change how you comment later.
You must be signed in to answer questions