The HVAD in MAUDE: 49,000 Adverse Event Reports on a Withdrawn Heart Pump

Medtronic stopped selling the HeartWare HVAD on June 3, 2021. The reason the company gave was a "growing body of observational clinical comparisons" showing higher stroke and mortality rates against other circulatory support devices. Five years later, the public FDA adverse event record contains 49,391 reports tied to HVAD components, including 5,073 patient deaths and 28,980 injuries. The device is gone from the market. The reports are still arriving.

What follows is a quantitative retrospective using public data from the FDA's Manufacturer and User Facility Device Experience (MAUDE) database, queried through the Claripulse pipeline. It looks at what was reported about the HVAD before the withdrawal, what surfaced after, and what the pattern says about the limits of post-market surveillance for high-risk implants. None of this constitutes a clinical opinion or an attempt to re-identify any patient. MAUDE data is what it is: reports submitted under FDA regulation, with all the gaps and biases that come with that.

#What the HVAD was

The HVAD was a small, fully implantable left ventricular assist device. The pump used a magnetically and hydrodynamically suspended impeller, no contact bearings, and a flexible driveline routed through the abdominal wall to an external controller and batteries. HeartWare International received FDA premarket approval in 2012 for bridge-to-transplant and again in 2017 for destination therapy. Medtronic acquired HeartWare in 2016 for roughly $1.1 billion. By the time of withdrawal, around 19,000 patients had received an HVAD globally, and approximately 4,000 were still living with one implanted.

Two devices competed in the same space. The Thoratec HeartMate II was an older axial-flow pump. The Abbott HeartMate 3 was a newer fully magnetically levitated centrifugal pump that received FDA approval in 2017. The HVAD's clinical trajectory ran into the HeartMate 3's growth. By 2021 the question was no longer whether the HVAD was a useful device. It was whether the additional risk over the HeartMate 3 was justifiable.

#Five years of reports, by component

Across all HVAD-related products in the device catalog, MAUDE has logged 49,391 reports since January 2019. Pumps, drivelines, controllers, and batteries each generate their own filings. The breakdown:

Component	Reports
HVAD Pump Implant Kit	36,994
HVAD Controller	3,882
HVAD Driveline	3,361
HeartWare Battery (primary SKU)	2,576
HeartWare Controller (secondary SKU)	2,293
HeartWare Battery (secondary SKU)	1,397
Other accessories	< 1,000 each

The annual trend is not flat. 2019 logged 13,923 reports — the largest single year — driven by a backlog of historical events Medtronic disclosed during a period of enhanced FDA reporting requirements. Volume settled to roughly 7,000 to 8,000 per year through 2020, 2021, and 2022. The first quarter of 2022 produced a visible spike of 4,016 reports tied to Medtronic's post-withdrawal letter campaign, when the company was contacting clinicians about the population of still-implanted patients. By 2023 the rate had dropped to about 4,500 reports per year and stayed near that level through 2025. The volume is declining. It is not zero. The HVAD continues to generate adverse event reports five years after Medtronic pulled it from the market.

#The failure-mode signature was already in MAUDE

Medtronic's 2021 statement attributed the withdrawal to clinical observational comparisons. That is true. What is also true: MAUDE reflected the same pattern earlier and at larger scale.

Stroke and cerebrovascular events appeared as free-text mentions in 1,168 HVAD narratives filed in 2019 alone. Thrombosis and pump thrombus, recorded in either the structured product-problem fields or patient-problem fields, account for 3,159 entries across the five-year window. Hemolysis appears in 1,973 reports. These three failure modes describe the hemocompatibility profile of a centrifugal LVAD that does not isolate blood from contact-induced clot formation. They are also the specific dimensions on which the HeartMate 3 demonstrated superiority in the MOMENTUM 3 trial: lower stroke, lower pump thrombosis, lower bleeding.

The HVAD's structured product-problem fields reinforce the picture:

Product Problem	Count
Pumping Problem	6,824
Battery Problem	5,500
Power Problem	3,608
Break (component)	1,050
Connection Problem	847
Pumping Stopped	809
Obstruction of Flow	660
Failure to Pump	240

Pumping problems and power problems together account for over 11,000 reports. None of this was hidden. Any clinician, patient, or attorney with the patience to query MAUDE could have surfaced the pattern. The challenge with MAUDE has never been access. It is the absence of structure that makes the trend visible at a glance. We discussed that gap in detail in an earlier post on searching the FDA MAUDE database.

Reporting about the FDA's own awareness of HVAD failures predates the withdrawal by almost a decade. ProPublica's 2024 investigation traced FDA inspection findings going back to 2014, when the agency issued HeartWare a warning letter citing manufacturing and complaint-handling deficiencies. The agency continued to clear new HVAD components and approve label expansions through 2017 while inspection findings accumulated. That pattern matters for any reader of MAUDE data. The reports do not arrive in a regulatory vacuum. The question of when a signal becomes actionable is not only a statistical one. It is a regulatory and commercial one.

#What surfaced after the withdrawal

Pulling a device from the market does not stop reporting. In April 2022, ten months after the sales halt, Medtronic and the FDA issued a Letter to Health Care Providers describing a previously unreported failure mode: a weld defect in the pump's center post that allowed moisture to corrode the impeller magnets. Three confirmed cases. Two deaths after pump exchange. The FDA followed with a Class I recall in June 2022.

The "failure to restart" language Medtronic flagged in its 2021 communications also remained an active surveillance item. Free-text searches of HVAD narratives identify 624 mentions of "failure to restart," "delay to restart," or close variants in 2024 reports alone. That number was near zero before 2021. Whether the spike reflects new failures or back-reporting of previously unreported events is impossible to determine from MAUDE alone, which is one of the reasons rate-based MAUDE analysis is fraught. What is verifiable: three years post-withdrawal, the FDA was still receiving reports against this failure mode at a steady pace.

Litigation followed the recall. Patients and surviving families have filed claims alleging defective design, failure to warn, and inadequate post-market surveillance. Medtronic has settled several cases privately, with terms and amounts undisclosed. Plaintiffs' counsel have organized intake portals and litigation guides since 2022. The legal calendar continues to advance, and the MAUDE record is the public evidence base most filings cite first. ProPublica's reporting has been used in several complaints. The product is gone. The legal exposure is not.

#How the HVAD compares to the HeartMate 3

The temptation in a retrospective like this is to declare the device unsafe by absolute numbers and stop. The honest comparison is harder. Across the same window from 2019 through early 2026, the HeartMate 3 generated approximately 22,500 reports and roughly 4,900 deaths in MAUDE. On absolute numbers, the HVAD has roughly twice the report volume and slightly more deaths.

The denominator is the problem. The HeartMate 3 became the dominant LVAD globally after 2017 and was the only available LVAD in most U.S. centers after the HVAD withdrawal. Its implanted base has grown every year. The HVAD's implanted base has shrunk every year through transplant, explant, and patient death. A ratio of MAUDE reports to deaths between the two devices is not interpretable as a relative safety ratio.

What is interpretable is the failure-mode mix. The HVAD's MAUDE record is dominated by stroke, thrombosis, and pump-stoppage events. The HeartMate 3's record is dominated by infection, GI bleeding, and right heart failure. These are different signatures, and they reflect what the MOMENTUM 3 trial measured directly: HM3 reduces hemocompatibility-related events at the cost of a different complication profile. Five-year survival in the MOMENTUM 3 long-term cohort was approximately 60 percent for the HeartMate 3 versus 40 percent for the HeartMate II. The HVAD never appeared in MOMENTUM 3 — and that is part of the story.

#What the HVAD case says about MAUDE

MAUDE lags reality. The 2019 reporting volume reflected events that occurred years earlier, disclosed in bulk during a regulatory push. Any rate analysis that uses date-received as a proxy for date-of-event will misread the trend. Date-of-event is more reliable where it is populated, which is not always.

MAUDE is structurally biased toward what manufacturers report. The withdrawal cited "observational clinical comparisons" rather than MAUDE because those comparisons used registry data with proper denominators. MAUDE has no denominator. It records reports, not rates. An analyst looking at the HVAD in 2018 with only MAUDE access would have seen the failure-mode pattern, but not the comparative risk against the HeartMate 3, because HM3's implanted base was still ramping.

MAUDE rewards pattern detection more than total counting. The signal that mattered for the HVAD was the consistent over-representation of stroke and thrombosis in narratives, not the headline report total. Surfacing that pattern is the gap that motivates structured post-market surveillance tools, including ours. For background on what MAUDE-based surveillance is and what it cannot do, see what post-market surveillance actually is in the U.S..

#For patients and clinicians

Roughly 4,000 patients globally are still living with an HVAD. The FDA does not recommend elective explant. The risks of pump exchange, when one is medically indicated, are documented in Medtronic's Urgent Medical Device Communications and updated by the FDA's HVAD recall tracker. Clinicians caring for these patients are monitoring for the documented failure modes: pump stoppage with delayed restart, weld-defect-associated thrombosis, controller and battery failures. Plaintiff counsel and patient advocacy groups have continued to file individual claims, with private settlements reported through 2024 and 2025.

The full record of HVAD-related reports is searchable on Claripulse at the device detail page for the HVAD Pump Implant Kit. The same data feeds the public lookup tool, which accepts brand-name and product-code queries: search HVAD or HeartWare.

#Methodology and caveats

The numbers in this post come from a Claripulse query of MAUDE reports linked to device catalog entries matching "HVAD" or "HeartWare" by brand name or canonical name. Counts include all components: pumps, controllers, drivelines, batteries, and accessories. Death counts reflect the FDA-defined event_type field, which records the most serious patient outcome reported but does not establish device causation. MAUDE data is de-identified and is not a complete record of every adverse event. It is a record of events submitted under the FDA's Medical Device Reporting regulation. Cross-device comparisons in this post are not normalized to implanted patient populations and should not be read as relative safety estimates.

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.