Impella in MAUDE: 39,275 Adverse Event Reports and a Recall Cluster That Is Still Growing

Medtronic pulled the HVAD in 2021 and the story closed. Abiomed's Impella is on the opposite trajectory. Across all Abiomed catalog entries under FDA product code OZD, MAUDE has logged 39,275 adverse event reports since January 2019, with 4,923 patient deaths and 19,860 injuries. The Claripulse signal pipeline ran on May 1, 2026, and returned more than two dozen active high- and critical-severity signals against Impella, most of them tied to a Class I recall opened in February 2026. February 2026 alone produced 1,979 reports and 390 deaths against the device.

This post walks through the trend since 2023, the failure-mode signature in the reports, and how the signature lines up with seven open Class I recalls Abiomed has accumulated since June 2025. Standard MAUDE caveats apply: counts are reports, not rates; the event_type = Death field records the worst patient outcome reported but does not establish device causation; reports are de-identified and should stay that way.

#What the Impella is

The Impella is a catheter-mounted axial-flow pump that crosses the aortic valve and unloads the left ventricle. The variants in clinical use are the Impella 2.5 (510(k)-cleared in 2008, PMA-supplemented in 2015), the Impella CP (2012), the Impella 5.0 and LP 5.0 (surgical insertion through the axillary artery), the Impella 5.5 with SmartAssist (PMA expanded in 2019), the right-heart Impella RP (PMA in 2017), and the smaller Impella ECP. Every variant in current U.S. use connects to an Automated Impella Controller, paired with a single-use purge cassette that maintains the dextrose-and-heparin purge fluid keeping blood out of the motor housing.

Abiomed Inc. of Danvers, Massachusetts, was acquired by Johnson and Johnson MedTech on December 22, 2022, for roughly $16.6 billion. The deal made Impella the headline cardiovascular device in J and J's portfolio. The clinical case for that valuation strengthened in April 2024 with the DanGer Shock trial in NEJM, which reported a 12.7 percentage-point absolute reduction in 180-day mortality for Impella CP plus standard care versus standard care alone in STEMI-associated cardiogenic shock. The trial was the first positive randomized mortality readout for any percutaneous mechanical circulatory support device.

That commercial and clinical context matters because Impella's MAUDE trajectory after the J and J acquisition does not look like the trajectory of a device on a clean glidepath.

#The reporting trend turned in 2023

Through 2019 and 2020, the Impella variants in our catalog generated 15 to 35 MAUDE reports per month. By the second half of 2023 the line broke.

Period	Avg reports/month	Notable months
2019	22
2020	27
2021	44
2022	45
2023 Q1 to Q2	92
2023 Q3 to Q4	472	Oct: 1,161
2024	1,549	Mar: 1,802; Aug: 3,778 (623 deaths); Oct: 2,742; Nov: 2,693 (476 deaths)
2025	815	Sep: 1,449
2026 Q1	1,774	Feb: 1,979 (390 deaths); Mar: 1,633 (284 deaths)

The August 2024 spike of 3,778 reports including 623 deaths is the single largest month in the seven-year window. November 2024 logged 476 deaths in a single month against the device, the second largest. These are not historical disclosures of old events. The date_received distribution is dominated by recent filings.

Two changes in 2023 partly account for the ramp. The FDA tightened MDR submission expectations for high-risk PMA devices, and J and J inherited Abiomed's complaint-handling backlog after the December 2022 acquisition. Neither explains the magnitude. By the first quarter of 2026, with no plausible backlog left to disclose, monthly Impella reporting still ran near 1,700 per month with deaths at 280 to 390 per month.

#What the reports describe

The Impella product-problem field in MAUDE is dominated by hardware-and-software failure modes, not clinical complications. The top categories across all years:

Product Problem	Reports
Restricted Flow rate	2,796
Device Sensing Problem	2,538
Patient-Device Incompatibility	2,007
Pumping Stopped	1,632
Positioning Failure	1,482
Failure to Advance	1,457
Application Program Problem	1,320
Malposition of Device	1,291
Increase in Pressure	1,246
Patient Device Interaction Problem	1,009
Fluid/Blood Leak	912
Deformation Due to Compressive Stress	827
Battery Problem	639

The pattern is consistent across years. Reports cluster on pump performance (restricted flow, pumping stopped, increase in pressure), positioning (malposition, positioning failure, failure to advance), the purge system (fluid leak, increase in pressure), the controller software (application program problem, device sensing problem), and the catheter shaft itself (deformation due to compressive stress). The catch-all category "Adverse Event Without Identified Device or Use Problem" accounts for 18,376 additional reports, which is a separate problem with MAUDE's coding discipline rather than a clinical statement about the device.

Two failure modes deserve their own line. "Pumping Stopped" appears in 1,632 reports. "Application Program Problem" appears in 1,320 reports, with related software entries such as "Unintended Application Program Shut Down" and "Computer Operating System Problem" surfacing as top items in the disproportionate-reporting analysis. Both categories trace directly to open recalls.

#The recall cluster

Between June 2025 and February 2026, the FDA classified seven Abiomed Class I recalls. Six of them target the Automated Impella Controller or its purge subsystem. The seventh targets the pump purge cassette itself. Class I is the FDA's highest severity tier and means there is a reasonable probability of serious adverse health consequences or death.

Recall	Date	Class	Target	Reason
Z-2164-2025	2025-06-23	I	AIC	Controller may fail to detect connected pump
Z-2568-2025	2025-08-20	I	AIC	Pump Driver Circuit Assembly may cause pump stop
Z-0069-2026	2025-10-01	I	AIC	Cybersecurity vulnerabilities in the controller OS
Z-0136-2026	2025-09-16	I	AIC	Purge retainer failures cause purge pressure issues
Z-1589-2026	2026-02-18	I	Purge cassette	Gen 1 purge cassette leaks
Z-1591-2026	2026-02-16	I	AIC	"Purge System Blocked" alarm delayed on first-gen 5.5 pumps
Z-1030-2026	2025-12-04	II	5.5 SmartAssist S2	Incorrect outer box carton

Twelve Class II recalls from August 2024 describe a previously undocumented device interaction between the Impella pump and the Shockwave coronary intravascular lithotripsy catheter during PCI. Abiomed updated its labeling rather than removing product.

The Claripulse signal pipeline picked up the recall cluster directly. As of the May 1, 2026 run, four active sequential-spike signals on Impella are flagged "recall-associated: Z-1591-2026" by the MaxSPRT detector. The largest of them observed 1,379 adverse events against an expected 561 over 11 weeks, with a log-likelihood ratio of 422.26 against a critical value of 9.87. Seven concurrent "increasing trend" signals at critical severity show consecutive monthly upticks against the variant rollups. The EBGM disproportionality scan returned more than a dozen high-severity peer-outlier signals against other OZD devices, with the top product-problem signatures clustering on controller software ("Application Program Problem" EB05 of 79.9, "Unintended Application Program Shut Down" EB05 of 95.1) and pump-purge mechanics ("Priming Problem" EB05 of 100.0, "Decrease in Pressure" EB05 of 39.0). Those are the same failure modes the seven Class I recalls describe.

A single weld defect drove the HVAD's post-withdrawal recalls. Abiomed's current cluster spans the controller, the purge subsystem, and the pump driver circuit. Each Class I notice identifies a distinct failure mode on a device whose clinical use is expanding.

#How it compares to the HeartMate 3

The HeartMate 3 is the durable LVAD that displaced the HVAD after 2021. Across the same January 2019 to March 2026 window, the HeartMate 3 generated 50,977 MAUDE reports with 11,176 deaths and 20,681 injuries. On absolute numbers, HeartMate 3 outpaces Impella in both report volume and deaths.

That comparison is not a relative safety ratio. The two devices are not in the same use case. HeartMate 3 is an implanted destination-therapy or bridge-to-transplant pump expected to support a patient for years. Impella is a short-duration percutaneous support device, used during high-risk PCI and in cardiogenic shock, that dwells in the body for hours to days. A patient on HeartMate 3 generates years of opportunity for reportable events. A patient on Impella generates a much narrower exposure window. The two MAUDE record sizes reflect that asymmetry as much as anything else.

The signature in the reports is what differs. HeartMate 3 reports cluster on infection, GI bleeding, and right heart failure, the durable-implant complication profile. Impella reports cluster on flow restriction, positioning, and controller software, the percutaneous deployment-and-hardware profile. One number does not capture two devices that fail in two different ways.

#What MAUDE will and will not show

MAUDE is the public surveillance record for medical devices in the United States. It is not a registry. Reports are not rates. The denominator (how many Impella devices were used in a given period) is not in MAUDE. Without it, no per-procedure complication rate can be computed from MAUDE data alone. The August and November 2024 death spikes look extreme in absolute numbers and are extreme even in proportional terms across the broader peer set, which is why the disproportionality detector flagged them. They do not, by themselves, establish that Impella caused the deaths. Cardiogenic shock has a high baseline mortality regardless of device.

MAUDE is also a lagging surface. Recall-associated reporting follows the recall by weeks to months. Manufacturer reports follow patient reports. Date-received is not date-of-event. Any rate analysis using date-received as the time index will overweight backlogs.

Pattern detection and component-level resolution are where MAUDE earns its keep. The seven Class I recalls in our table are not the same recall renamed seven times. They identify distinct subsystem failures in the controller, the purge cassette, the pump driver circuit, and the alarm logic. The reporting pattern in MAUDE lines up with what those recall notices describe.

#For clinicians, patients, and counsel

Cardiologists running PCI labs and cardiogenic shock programs are already managing the recall cluster. Hospitals received Urgent Medical Device Communications for each Class I recall and have been updating standard operating procedures. The FDA's recall pages and Abiomed's own customer-facing communications are the operational sources of truth. The current Claripulse recalls view surfaces the same set with cardiology filters.

For patients and families, the full reporting record is searchable on Claripulse. The main Impella catalog entry, with all linked variants and the underlying report stream, is at the Impella device page. The public lookup tool accepts brand name or product code queries: search Impella, Abiomed, or product code OZD.

For counsel evaluating individual matters, the MAUDE record is the public evidence base, not the case. The cluster pattern visible in surveillance data does not displace the need for individual medical record review, expert review, or causation analysis. Our guide to searching MAUDE walks through the structural gaps that make raw MAUDE searches incomplete on their own.

This system analyzes publicly available FDA MAUDE data. It does not provide medical advice. MAUDE data is de-identified, and no attempt should be made to re-identify any individual patient.