The practical way to research FDA medical device adverse event reports is to treat MAUDE as a searchable archive of suspected associations, not a rate calculator. Start by defining the device characteristics you can anchor a query to—brand name, manufacturer, model number, product code, or unique device identifier—then choose simple search for a selected calendar year or advanced search for a date range and specific fields. As you review returned MDRs, read the report source, event type, device problem codes, and manufacturer narrative together. For larger questions, move to the downloadable MDR Data Files. The core discipline is this: every report reflects an unverified submission, and you should document what the record supports before drawing any conclusion about causation, frequency, or comparative safety.
What MAUDE Contains and What It Excludes
The Manufacturer and User Facility Device Experience database contains medical device reports of adverse events submitted by mandatory reporters—manufacturers, importers, and device user facilities—and voluntary reporters such as health care professionals, patients, and consumers [1][3]. The public database covers the last ten years of MDR data and is updated monthly [1]. Mandatory manufacturer and importer reports date from August 1996, user facility reports from 1991, and voluntary reports from after June 1993 [3][7]. The record therefore mixes different reporter classes, reporting obligations, and levels of clinical detail.
Two exclusions matter for research coverage. Reports received under alternative summary reporting exemptions from 1999 through April 2019 are not available in MAUDE and appear instead on the MDR Data Files page [2]. Reports older than ten years are also outside the searchable MAUDE window and are provided as downloadable files [4][5]. If a research question spans a product introduced before the ten-year window or a device subject to a prior ASR exemption, a MAUDE-only search will undercount the available record set [2][4].
Choosing Simple Search Versus Advanced Search
Simple search accepts any terms appearing in an MDR received in a selected year. You can enter a single word, an exact phrase in quotation marks, or multiple words separated by "and," and the device UDI can also be used as a query [5]. This mode is useful when you have a distinctive term—such as a component name or an unusual event description—but no structured identifiers. The return is capped at 500 results for the selected year, so a broad term in a high-volume device category may truncate before you see the complete set [5].
Advanced search lets you query specific fields such as device brand name or manufacturer over a selected timeframe, also capped at 500 results [5]. This is the better starting point when you know the manufacturer or brand but need to bound the period of interest. The same cap applies, so a multi-year query for a widely used device family can exceed the return limit. In that situation, the FDA directs researchers to download MDR files for larger searches and additional options [5].
A Suggested Workflow for a New Device Signal
Suppose a biomedical engineer is asked to research a suspected problem with a branded electromechanical pump. As a suggested sequence, begin with advanced search using the brand name and a narrow date range, then export or copy the returned MDRs and record the report source code and report number for each [5][2]. Next, run a simple search on a distinctive quoted phrase from the suspected failure mode within a selected year [5]. Compare the advanced and simple result sets for duplicates and review the manufacturer narrative, event type, and patient problem codes on each unique record [1][4].
Then, because simple and advanced searches each cap at 500 results, check whether the query likely hit the cap; if so, download the relevant MDR Data Files for the reporting years and filter by manufacturer and brand in the pipe-delimited records [5][2]. Note whether any records fall under the legacy ASR files, which were removed from MAUDE [2]. Throughout the review, do not calculate an event rate: MAUDE records cannot establish incidence, prevalence, causation, or frequency of device use [1][7].
Reading an MDR Without Overreading It
The displayed MDR fields include model number, event date, event type, patient problem codes, implant and explant years, third-party servicing status, report type, and, for manufacturer reports, a manufacturer narrative [1][4]. Patient sex, weight, ethnicity, and race may also appear [4]. If a submitter left a field blank, it does not show [1]. When reading, check the report source code because the same event may generate separate records from a user facility, distributor, manufacturer, and voluntary submitter [2]. A single clinical incident can therefore appear as multiple master event records across different sources, which is important when you are tallying records rather than distinct events.
FOIA redactions further condition the text. Trade secret or confidential business information is replaced by "(b)(4)," and personal or medical file information by "(b)(6)" [1]. A manufacturer narrative may therefore be partially or substantially redacted. Type of reportable event should be read carefully: the appearance of the word "death" does not make a report a death report unless the reporter believed the device was or may have been attributed to, or a factor in, the death [1][3]. Treat the narrative and coded fields as the submitter's account, not a verified clinical finding.
Moving to MDR Data Files for Volume and History
The MDR Data Files page provides downloadable zipped files containing voluntary reports since June 1993, user facility reports since 1991, distributor reports since 1993, and manufacturer reports since August 1996, updated monthly [4]. The files use pipe-delimited records, and field reference documentation is designed to be used with MedWatch forms 3500A and 3500 [2]. The MDRFOI file, for example, contains 82 fields including MDR report key, report number, report source code, and manufacturer link flag [2]. This is the route when the 500-result cap is inadequate or when the question predates the ten-year MAUDE window.
Downloading brings its own handling decisions. The FDA recommends opening and saving files in Microsoft Word before importing into Microsoft Access to add an end-of-record marker [2]. For FOIDEV files, an extra character may be needed at the end of the first record to avoid losing the last column [2]. The legacy ASR files use a separate format with device problem codes documented in a companion CSV, and reports from 1999 through April 2019 under that program are absent from MAUDE [2]. Plan to reconcile identifiers across the MAUDE display and the file schemas rather than assuming the same field appears in both.
Sources & references
- About Manufacturer and User Facility Device Experience (MAUDE) Database | FDA · www.fda.gov
- MDR Data Files | FDA · www.fda.gov
- Medical Device Reporting (MDR): How to Report Medical Device Problems | FDA · www.fda.gov
- About Manufacturer and User Facility Device Experience (MAUDE) Database | FDA · www.fda.gov
- Manufacturer and User Facility Device Experience (MAUDE) Database · www.accessdata.fda.gov
- Mandatory Reporting Requirements: Manufacturers, Importers and Device User Facilities | FDA · www.fda.gov
- Medical Device Reporting (MDR): How to Report Medical Device Problems | FDA · fda.gov
- Medical Device Reporting (MDR): How to Report Medical Device Problems | FDA · www.fda.gov