MAUDE drug-delivery adverse events: insulin pumps dominate 1.39M reports

FDA's Manufacturer and User Facility Device Experience (MAUDE) database represents the front line of post-market medical device surveillance in the United States. Between 2022 and 2026, the database logged 12,223,321 total report-by-device rows. Of these, drug-delivery devices—including insulin infusion pumps, automated insulin dosing (AID) systems, ambulatory infusion pumps, hospital large-volume infusion pumps, patient-controlled analgesia (PCA) pumps, autoinjectors, and prefilled syringes—accounted for 1,393,568 reports. This represents approximately 11.4% of all medical device adverse event reports filed with the FDA over the last four and a half years.

A deep-dive analysis of this 1.39M report cohort reveals that post-market safety reporting for drug-delivery hardware is highly concentrated. It is not a broad, evenly distributed sample of the drug-delivery industry. Instead, it is overwhelmingly dominated by diabetes care technologies—specifically, insulin infusion pumps and Alternate Controller Enabled (ACE) pump systems. Tandem Diabetes Care alone generated 913,475 reports, representing more than 65% of the entire drug-delivery cohort. The clinical fingerprint of these reports is also highly specific: while 90.4% (1,259,318 reports) are classified as device malfunctions with "no clinical signs or symptoms" in the patient, the remaining clinical signal points directly to insulin over- or under-delivery, with 169,400 reports of hyperglycemia, 43,592 reports of hypoglycemia, and 11,351 reports of elevated ketones or diabetic ketoacidosis (DKA).

For biopharma quality, manufacturing, combination-product engineering, and regulatory affairs teams, these numbers provide critical post-market safety intelligence. Understanding the baseline rate of malfunctions, the distribution of failure modes across different device classes, and the clinical outcomes associated with hardware errors is essential for designing robust drug-delivery devices and navigating FDA's post-market surveillance expectations.

---

How big is the drug-delivery-device adverse-event footprint in MAUDE, and is it rising?

The post-market reporting volume in MAUDE for all medical devices has hovered at historic highs over the last few years, driven by increased adoption of connected health technologies and more rigorous regulatory compliance pathways. The drug-delivery-device cohort has tracked this overall trend closely, maintaining a consistent share of approximately 10% to 12.5% of all MAUDE reports.

Below is the year-by-year trend from January 1, 2022, through mid-July 2026, comparing the total number of report-x-device rows in MAUDE against the drug-delivery-device subset:

Data Source: openFDA Device Event Database (2022–2026).

Several key takeaways emerge from these trends:

1. Steady Reporting Volume: Drug-delivery device adverse events run at an average of 278,000 to 331,000 reports per full calendar year. This steady baseline represents an immense volume of post-market data that manufacturers must monitor, triage, and investigate.
2. 2024–2025 Peak: The peak of both overall reporting and drug-delivery reporting occurred in 2024 and 2025, corresponding with the massive commercial expansion of automated insulin delivery systems (like the Tandem t:slim X2 with Control-IQ and the Insulet Omnipod 5) and the resolution of several high-profile hospital infusion pump recalls.
3. Caveat on "Report-x-Device" Rows: It is crucial to note that MAUDE database rows are structured as report-x-device relationships, not unique patient events. A single adverse event report submitted to the FDA can contain multiple device rows if multiple pieces of hardware (e.g., an insulin pump, an infusion set, and a transmitter) were involved in the incident. Consequently, these figures represent the volume of device-level records analyzed rather than the absolute number of unique clinical occurrences.

---

Which devices and manufacturers generate the most reports, and why is it mostly insulin pumps?

When the drug-delivery cohort is broken down by manufacturer, the high level of concentration becomes immediately apparent. Diabetes management hardware dominates the reporting landscape.

Below are the top 15 manufacturers within the drug-delivery adverse event cohort for the 2022–2026 period:

Note: medtech legal-entity variants (such as Beta Bionics Inc. and Beta Bionics, Inc.) are listed as they appear in the raw FDA files.

The data indicates that Tandem Diabetes Care and Medtronic (combining its Puerto Rico Operations, MiniMed, and St Paul divisions for a total of ~265,267 reports) account for over 84% of all drug-delivery device reports. Insulet Corporation contributes another 6.20% (86,416 reports), and Beta Bionics (combining its two database entity listings for a total of 46,636 reports) represents approximately 3.35%.

Why do diabetes devices—specifically insulin pumps—so heavily dominate a database meant to cover all drug-delivery devices, including oncology infusion pumps, PCA pumps, and clinical syringe drivers?

There are three primary reasons for this asymmetry:

1. The Size of the Active User Base: Unlike a hospital infusion pump, which is operated by a trained clinician and shared among multiple patients in an acute care setting, an insulin pump is a personal medical device. Hundreds of thousands of patients with Type 1 diabetes wear these pumps 24 hours a day, 365 days a year. The cumulative number of operating hours for the global insulin-pump population is orders of magnitude larger than that of any other class of active drug-delivery hardware.
2. Mandatory Reporting of Malfunctions: Under 21 CFR Part 803, medical device manufacturers are legally obligated to report any device malfunction to the FDA if the malfunction is likely to cause or contribute to a death or serious injury if it were to recur. Because insulin pumps deliver a high-potency hormone (insulin) where minor over- or under-delivery can lead to life-threatening hypoglycemia or diabetic ketoacidosis, almost any firmware bug, occlusion alarm, battery depletion, or screen crack is technically reportable as a malfunction. This generates a massive volume of reports for events where no patient was actually harmed.
3. Connectivity and Data Logging: Modern insulin pumps and automated insulin dosing (AID) systems are highly connected. They pair with continuous glucose monitors (CGMs) and smartphones. They log every bolus, basal adjustment, sensor reading, and hardware state. This high degree of logging makes it easier for both users and manufacturers to detect and document minor anomalies, which are then systematically compiled and reported to the FDA. (Note: CGMs themselves are diagnostic devices and are categorized under clinical chemistry monitors; they are excluded from this delivery-device cohort because they monitor glucose levels rather than actively delivering insulin).

To understand which specific devices are driving these numbers, we can analyze the top brands and product codes within the cohort.

Brand-Level Reporting Volume

The brand-level distribution reflects the dominance of Tandem’s t:slim platform and Insulet’s Omnipod system:

Product code QFG, which represents "alternate controller enabled insulin infusion pump" (often referred to as interoperable or ACE insulin pumps), accounts for 1,014,029 reports—approximately 73% of the entire drug-delivery cohort. Other diabetes-related product codes include OYC (interoperable insulin pumps, 95,830 reports), OZO (insulin infusion pumps to be used with invasive glucose sensors, 90,469 reports), and OZP (automated insulin dosing, threshold suspend systems, 50,788 reports).

---

What is actually failing: malfunction, injury, or death?

To understand the real-world safety impact of these devices, we must analyze the severity of the reported events. The FDA categorizes MAUDE reports into five primary event types. Within our drug-delivery cohort, the breakdown is as follows:

Data Source: openFDA Device Event Database (2022–2026).

1. Malfunctions (90.37% of reports)

Malfunctions make up the overwhelming majority of drug-delivery adverse events. This high percentage should reassure clinicians and patients: the vast majority of MAUDE reports represent proactive, compliance-driven reporting of hardware or software anomalies before they cause clinical harm. Common malfunctions include:

• Battery and Power Issues: Premature battery depletion, charging contact corrosion, or unexpected shutdowns.
• Occlusion and Flow Failures: Blockages in the infusion tubing or pump mechanism that prevent the programmed delivery of the drug.
• Software and Connectivity Bugs: Loss of Bluetooth pairing between the pump and the CGM, screen freezes, or sensor-integration lag.
• Physical Damage: Cracked casings, broken belt clips, or water ingress exceeding the device's IPX rating.

Under FDA's post-market surveillance rules, if a manufacturer detects any of these malfunctions, they must file a report if a repeat of the malfunction could result in injury. Since insulin delivery errors carry a theoretical risk of severe hypo- or hyperglycemia, they are reported systematically.

2. Injuries (9.55% of reports)

With 133,049 injury reports, drug-delivery hardware errors do sometimes result in patient harm. In the context of insulin pumps, an "injury" typically refers to an acute metabolic episode (hyperglycemia or hypoglycemia) that requires medical intervention, such as an emergency room visit, administration of IV dextrose, or treatment with intramuscular glucagon. It can also include infusion-site infections, severe skin irritation, lipohypertrophy, or physical trauma from a failing catheter insertion needle.

3. Deaths (0.08% of reports)

Between 2022 and 2026, the MAUDE database logged 1,140 death reports associated with drug-delivery devices.

[!IMPORTANT] A critical epidemiological caveat applies to MAUDE death data: The presence of a death report in the FDA database does not prove that the medical device caused or contributed to the patient's death. Manufacturers are required to submit a report whenever they become aware of a patient death involving a user of their device, even if the primary cause of death was unrelated (e.g., cardiovascular disease, trauma, or natural causes). Adverse event databases are passive surveillance systems that capture reports of temporal association; they are not controlled clinical trials and cannot establish direct causality.

---

What is the real patient-harm signal beneath 1.16M 'no symptom' reports?

The raw event-type classification only tells part of the story. To understand what is actually happening to patients, we must examine the specific clinical outcomes documented in the "patient problems" field of the MAUDE reports.

Because many reports contain multiple problem codes, the total number of problem occurrences exceeds the number of reports. The distribution of the top patient problems within the drug-delivery cohort is detailed below:

Data Source: openFDA Device Event Database (2022–2026). Multiple codes may be reported per row.

The clinical fingerprint of this dataset is clear:

• Asymptomatic Failures (83.6% of problem entries): The dominant code is "No Clinical Signs, Symptoms or Conditions" (1,165,558 entries). This matches the high proportion of malfunctions, confirming that the vast majority of reports represent hardware failures caught by safety alarms before they impacted the patient's physiology.
• The Under-Delivery Fingerprint: Hyperglycemia (169,400) and DKA/Elevated Ketones (11,351) represent the clinical consequence of insulin under-delivery. This occurs when a pump fails to deliver basal insulin or boluses, often due to silent occlusions, infusion set dislodgement, bent cannulas, software delivery pauses, or reservoir leaks.
• The Over-Delivery Fingerprint: Hypoglycemia (43,592) and Loss of Consciousness (3,461) represent the clinical consequence of insulin over-delivery. This is a higher-risk signal, as acute, severe hypoglycemia can lead to rapid neurological decline, seizures, and death. Over-delivery can be caused by programming errors (user inputs), software calculation bugs, sensor-integration failures (where the AID system responds to a falsely elevated glucose reading), or physical pump malfunctions (e.g., siphoning or mechanical drive failures).

For combination-product manufacturers, this data underscores the importance of human-factors validation. Designing clear user interfaces, implementing robust occlusion-detection algorithms, and optimizing sensor-pump communication are critical to minimizing these metabolic risks. For a broader look at pre-market human-factors engineering, refer to our guide on combination product human factors validation.

---

What do the non-diabetes delivery devices — hospital infusion and PCA pumps — look like?

While diabetes devices represent the vast majority of MAUDE drug-delivery reports, hospital-grade large volume infusion pumps (LVPs), syringe drivers, and patient-controlled analgesia (PCA) pumps represent a distinct and critical clinical subset.

Though their total report volume is smaller, hospital infusion pumps are associated with different safety risks because they are used to deliver high-alert medications—such as chemotherapy, intravenous anesthetics, vasopressors, and opioids—to critically ill patients.

The primary manufacturers of these non-diabetes delivery systems in the MAUDE cohort include:

• Baxter Healthcare Corporation: 18,577 reports (driven by the Spectrum and Novum IQ platforms).
• Smiths Medical ASD, Inc.: 12,107 reports (driven by the CADD ambulatory infusion systems).
• Carefusion SD (Becton Dickinson): 11,851 reports (driven by the Alaris infusion platform).
• ICU Medical, Inc.: 7,422 reports (driven by the Plum and LifeCare systems).

Comparing Hospital Infusion Safety vs. Personal Insulin Pumps

The post-market risk profile of hospital infusion hardware differs from personal diabetes tech in several key areas:

In hospital settings, the dominant failure modes are often related to programming errors (e.g., mistyping a infusion rate or selecting the wrong drug concentration) and mechanical interactions (e.g., air-in-line false alarms, sensor failures, or physical damage to the pump door latch).

Because hospital pumps deliver drugs like norepinephrine or fentanyl, an unexpected pump shutdown or flow interruption can cause rapid hemodynamic instability or severe withdrawal. Conversely, a programming error that leads to over-delivery of an opioid can result in fatal respiratory depression. This makes the design of hospital infusion software—specifically the integration of Dose Error Reduction Systems (DERS) or drug libraries—a critical focus for manufacturing and quality teams. For more information on post-market quality agreements, see our briefing on CDMO quality agreement red flags.

---

How does MAUDE reporting work, and what changes with the 2026 AEMS transition?

To interpret MAUDE statistics correctly, it is necessary to understand the regulatory mechanics of medical device reporting (MDR) under 21 CFR Part 803. The database compiles reports from four sources, each with different legal requirements:

1. Manufacturer Reports (99.65% of the cohort): Device manufacturers are legally obligated to submit MDRs to the FDA within 30 calendar days of becoming aware of information that reasonably suggests one of their marketed devices may have caused or contributed to a death or serious injury, or has malfunctioned in a way that would be likely to cause death or serious injury if the malfunction were to recur. Within our drug-delivery cohort, manufacturers filed 1,388,673 of the 1,393,568 reports (99.65%), reflecting their extensive compliance monitoring systems.
2. Voluntary Reports (0.19% of the cohort): Healthcare professionals, patients, and consumers can submit voluntary adverse event reports directly to the FDA via the MedWatch program. Voluntary reports accounted for 2,607 records in the cohort.
3. Distributor / Importer Reports (0.15% of the cohort): Distributors and importers must report device-related deaths, serious injuries, and certain malfunctions. These accounted for 2,045 records in the cohort.
4. User Facility Reports (0.02% of the cohort): Hospitals, nursing homes, and outpatient treatment centers are legally required to report device-related deaths to both the FDA and the manufacturer, and device-related serious injuries to the manufacturer (or the FDA if the manufacturer is unknown). User facilities accounted for just 243 reports in our drug-delivery cohort.

The 2026 transition to the Adverse Event Monitoring System (AEMS)

The legacy MAUDE database has long faced criticism from industry analysts and epidemiologists for its clunky search interface, lack of standardized vocabulary, and lag in processing times. In 2026, the FDA began replacing it with the Adverse Event Monitoring System (AEMS).

AEMS is best understood not as a from-scratch MAUDE rebuild but as a unified platform that consolidates the agency's previously fragmented post-market databases into one system. It evolved out of FAERS (the former drug/biologic adverse-event database, expanded after vaccine reports from VAERS were integrated) and is being extended to absorb medical-device reports. On March 11, 2026, the FDA announced the AEMS launch; the agency has indicated that medical-device (MAUDE) data migration into AEMS is scheduled over the course of 2026, with the legacy MAUDE interface being retired once that migration completes.

What the transition changes—and what it does not:

• Unified, real-time public dashboard: AEMS consolidates adverse-event reports across FDA-regulated product categories into a single, searchable public interface with enhanced analytics and improved application programming interfaces (APIs), replacing the siloed legacy search tools.
• eMDR coding changes: As part of the consolidation, the FDA announced changes to the coding used in electronic Medical Device Reporting submissions (including certain country codes and adverse-event codes no longer supported), effective May 11, 2026.
• Reporting obligations are unchanged: AEMS does not alter what manufacturers, user facilities, importers, and voluntary reporters must submit under 21 CFR Part 803. The legal triggers, 30-day timelines, and MDR content requirements remain the same; only the system that stores, searches, and surfaces those reports is changing.

For the cohort numbers in this article, the migration matters as a freshness caveat: the 2022–2026 figures above were computed from the openFDA device-event export of the legacy MAUDE data. Future annual refreshes should re-baseline against the AEMS device export once device data has fully migrated, and the new public dashboard is likely to make late filings and coding inconsistencies more visible than the legacy system did.

---

Frequently Asked Questions

What is MAUDE and who has to report to it?

MAUDE (Manufacturer and User Facility Device Experience) is FDA's post-market surveillance database for medical devices. Under 21 CFR Part 803, medical device manufacturers must report all device-related deaths, serious injuries, and reportable malfunctions. Hospitals and user facilities must report device-related deaths and serious injuries. Patients, consumers, and clinicians can file reports voluntarily.

Why do insulin pumps dominate drug-delivery-device adverse events?

Insulin pumps dominate due to their large active user base (hundreds of thousands of patients wearing them 24/7), strict mandatory malfunction reporting (because insulin dosing errors carry high risk), and advanced connectivity features that log and capture minor hardware and software anomalies.

Does a high MAUDE report count mean a device is dangerous?

No, a high report count does not necessarily indicate a defective or dangerous device. It often reflects a large market share (installed base), a highly active user population, or rigorous corporate compliance with mandatory malfunction reporting rules. Because 90% of these reports are malfunctions with no patient symptoms, they indicate that device safety mechanisms are working as intended to catch errors before they cause harm.

What share of drug-delivery-device reports involve death?

In our 2022–2026 cohort, death reports accounted for 1,140 out of 1,393,568 total records, which is approximately 0.08% of all reports. It is important to note that MAUDE records temporal associations, meaning a report is filed if a patient dies while using a device, regardless of whether the device caused the death.

What is the FDA AEMS transition and will it change these numbers?

The FDA AEMS (Adverse Event Monitoring System) is a 2026 modernization initiative that consolidates the agency's previously fragmented post-market databases—including the legacy MAUDE device data—into a single, real-time public dashboard with enhanced analytics and APIs. It does not change what manufacturers must report under 21 CFR Part 803, and it will not change the historical MAUDE data behind these 2022–2026 figures. Over time it is expected to make post-market device data more searchable and to surface coding inconsistencies and late filings more visibly than the legacy system did.

---

Sources

1. FDA Manufacturer and User Facility Device Experience (MAUDE) Database. Center for Devices and Radiological Health (CDRH). FDA MAUDE Database
2. openFDA Device Event API. U.S. Food and Drug Administration. openFDA Device Event JSON API
3. Medical Device Reporting (MDR): 21 CFR Part 803. U.S. Code of Federal Regulations. FDA MDR Regulations
4. FDA Manufacturer and User Facility Device Experience (MAUDE) overview. U.S. Food and Drug Administration. FDA MAUDE Overview
5. FDA Adverse Event Monitoring System (AEMS). FDA announced the unified AEMS platform on March 11, 2026; medical-device (MAUDE) data migration into AEMS is scheduled over the course of 2026. FDA AEMS Announcement (Morgan Lewis summary)
6. More Focus is Needed to Reduce Adverse Events for Diabetes Devices. PubMed Central (PMC8861793). PMC Article
7. Late Reporting of Device Adverse Events in the MAUDE Database. The BMJ (2024–2025 Investigation). BMJ Study
8. More Focus is Needed to Reduce Adverse Events for Diabetes Devices. PubMed Central (PMC8861793). PMC Article
9. Late Reporting of Device Adverse Events in the MAUDE Database. The BMJ (2024–2025 Investigation). BMJ Study