Sterilization equipment failures are recorded in two very different public databases, and both are routinely misread. Health Canada publishes recall notices for medical devices sold in Canada. The United States Food and Drug Administration publishes the Manufacturer and User Facility Device Experience database, known as MAUDE, which holds medical device reports of adverse events.
The two look interchangeable from the outside — both are searchable, both return records with a device name, a date and a description — and they are regularly quoted side by side as if they measured the same thing on two sides of the border. They do not. They record different events, from different reporters, under different legal duties, and neither one is designed to produce a failure rate. This article explains what each database actually contains, why a count from either one cannot be used as a reliability figure, and how to check whether your own sterilizer batch or serial number appears in the public record at all.
What each database actually contains
The single most useful habit when reading either source is to ask, first, who put the record there. The answer determines what the record can and cannot tell you.
| Field | Health Canada recall notices | FDA MAUDE |
|---|---|---|
| What it records | Recalls and safety alerts for health products, including medical devices, published by Health Canada | Medical device reports (MDRs) of adverse events involving a device |
| Who supplies the record | The manufacturer's recall report, published by Health Canada | Mandatory reporters (manufacturers, importers, device user facilities) and voluntary reporters such as health care professionals, patients and consumers |
| Unit of a record | One recall action against one product or affected product list | One report of one event |
| Time span | Active notices plus archived items; recalls appear from the recall start date onward | The last ten years of MDR data, updated monthly to include reports received through the last day of the previous month |
| Older material | Archived and earlier notices remain searchable on the site | Voluntary reports since June 1993, user facility reports since 1991, distributor reports since 1993 and manufacturer reports since August 1996 are held in separate downloadable MDR data files |
| Does a record imply the device caused harm? | The notice describes a recall action and the issue prompting it | No. FDA states plainly that the submission of a report does not necessarily demonstrate that the device caused or contributed to the adverse outcome or event |
| Severity or class field | Recall class is recorded on the notice | Report type and event description fields; there is no equivalent recall class |
The first structural point is that a Canadian recall notice and an American event report are not the same kind of statement. A recall notice is an administrative action: a device has been recalled, and the notice tells you which units and why. A MAUDE record is a report: somebody alleged an association between a device and an event, and the report is published largely as received.
Who is legally required to report what
The reporting duties differ in a way that shapes the data in each database.
In Canada, the Medical Devices Regulations place the reporting duty on the manufacturer and the importer. Subsection 59(1) of the regulations requires that the manufacturer and the importer of a medical device each make a preliminary and a final report to the Minister concerning any incident that comes to their attention occurring in Canada that involves the device, where the device is sold in Canada and the incident is related to a failure of the device, a deterioration in its effectiveness, or an inadequacy in its labelling or directions for use, and has led to death or a serious deterioration in the state of health of a patient, user or other person, or could do so were the incident to recur.
In the United States, reporting duties extend beyond the manufacturer. Under 21 CFR 803.30, a user facility must submit a report to the FDA as soon as practicable but no more than ten work days after becoming aware of information that reasonably suggests a device has or may have caused or contributed to the death of a patient, and must also submit the report to the device manufacturer if known. For a serious injury, the facility must report to the manufacturer within the same ten work day window, and to the FDA if the manufacturer is not known.
| Question | Canada | United States |
|---|---|---|
| Primary mandatory reporter | Manufacturer and importer | Manufacturer; device user facilities also report |
| Facility reporting duty | Not the reporting party under section 59 | Yes — user facilities report deaths and serious injuries |
| Threshold for a report | Failure, deterioration in effectiveness or labelling inadequacy linked to death or serious deterioration in health, or that could do so if repeated | Reasonably suggests the device caused or contributed to a death or serious injury |
| Where the record lands | Published recall and safety alert notice | MAUDE report entry |
That asymmetry matters when you compare the two. A Canadian figure and an American figure are produced by different populations of reporters under different triggers, so treating them as two measurements of one quantity is a category error. It also explains why the American dataset contains a large volume of facility-originated reports with narrative detail, while the Canadian public-facing record is more concentrated in formal recall actions.
Why a count from either database is not a failure rate
This is the part that is stated officially and still ignored in practice. The FDA's own description of the MAUDE database sets out the limits of the data in explicit terms. It states that "MDR data is not intended to be used either to evaluate rates of adverse events, evaluate a change in event rates over time, or to compare adverse event occurrence rates across devices."
The same page explains why. The system is described as a passive surveillance system with limitations: the incidence, prevalence or cause of an event cannot be determined from the reporting system alone, because of under-reporting of events, inaccuracies in reports, lack of verification that the device caused the reported event, and lack of information about how often the device was used. Establishing causation is described as especially difficult where the circumstances have not been verified or the device has not been directly evaluated.
The FDA also publishes the statutory disclaimer that applies to every report in the file. Section 21 CFR 803.16 states that a report or other information submitted by a reporting entity, and any release by the FDA of that report or information, does not necessarily reflect a conclusion by the party submitting the report or by the FDA that the report or information constitutes an admission that the device, or the reporting entity or its employees, caused or contributed to the reportable event.
Three practical consequences follow, and they apply to the Canadian database as well:
- The numerator is unknown. What is published is the number of reports that were made, not the number of events that happened. A rise in reports can mean a rise in problems, or a rise in reporting after a policy change, or both.
- The denominator is unknown. Neither database tells you how many sterilizers of that model were in service, how many cycles each ran, or how many facilities reported. Without a denominator there is no rate, only a count.
- Attribution is unverified. A record naming a sterilizer is not a finding that the sterilizer caused the outcome. Recall notices describe an issue prompting a recall action; MAUDE records describe what a reporter submitted.
There is a fourth consequence specific to keyword searching. Health Canada's recall search is a text search across recall and safety alert records, and the word you type does not match one concept. On 29 September 2026, a search of the Canadian database for the term autoclave returned six records, while a search for the term sterilizer returned 479 records — and the overwhelming majority of the second set concerned the sterility of a product, such as a sterile applicator or a sterile ophthalmic ointment, not sterilization equipment. If you search the word sterilizer and read the result count as an equipment figure, you will be wrong by two orders of magnitude before you have read a single record.
How to search each database without over-reading it
A repeatable method is more useful than a headline number. The following sequence keeps the output interpretable.
- Search by equipment term, not by concept. Use the equipment noun — autoclave, sterilizer brand name, or the model designation — and then read the titles. Discard every record whose subject is the sterility of a product rather than the equipment itself.
- Record the method alongside the number. Write down the search term, the database, the date of the search and the number of records returned. A count without its method cannot be reproduced and should not be published.
- Open the individual notice rather than the result list. Result lists give a title and a date; the notice gives the affected product list, the lot or serial numbers, the issue and the recall class.
- For American data, quote the FDA's limitations language whenever you quote a count. If the source itself says the data is not intended for rate comparison, a rate comparison is not a defensible use of it.
- Treat the absence of a record as no information, not as a clean bill of health. Neither database is a register of every device in service, and under-reporting is documented.
Worked example: one Canadian sterilizer recall, read line by line
A single notice shows how much practical detail the Canadian record actually carries. The Statclave G4 Chamber Autoclaves notice, published on 10 July 2023 with a recall start date of 16 June 2023, records a performance issue: an error in production resulted in missing gaskets in some units on float sensors in the reservoirs, and while the units were not leaking when they left the facility, a leak would develop over time with a potential to cause an electrical short.
| Record field | Value in this notice |
|---|---|
| Brand | Scican Ltd. |
| Product | Statclave G4 Chamber Autoclaves |
| Issue category | Medical devices — Performance |
| Model or catalog number | G4-622301 |
| Lot or serial numbers | Ten listed serial numbers, from 710223E00007 to 710223E00017 |
| Recall class | Type II |
| Identification number | RA-73915 |
| Recall start date | 16 June 2023 |
Three things are worth noticing. First, the record identifies the affected units precisely, by model and serial number, which makes it actionable for an owner. Second, the recall class and identification number are administrative markers of the recall action, not measures of the hazard. Third, the notice tells you what to do: contact the manufacturer if you require additional information.
That is the realistic ceiling of what a recall notice does. It is a precise instruction about a specific set of units. It is not evidence about how often the model failed.
A batch and serial number self-check you can run this week
The public databases are most useful as a screening tool for equipment you already own. The check below takes an afternoon across a small clinic and does not depend on any subscription.
- Build an asset list. For each sterilizer, record the manufacturer, the model or catalog number, and the serial number. Take the serial from the data plate, not from an invoice or a memory of the purchase.
- Search the Canadian recall database for each manufacturer name and each model designation, in separate searches. Note the search term and date next to the result.
- Search the American database for the same equipment. Because MAUDE is a report database rather than a recall register, use it to find narrative information about the model, not to establish whether a recall exists.
- Compare the serial numbers you hold against the affected product lists in any notice you find. A model-level match is not an affected unit; a serial-level match is.
- If a unit is listed, quarantine it from clinical use and contact the manufacturer. Record the date, the notice reference and the action taken in the same log that holds your sterilization records.
- Regardless of the outcome, record the search. A dated record that you checked your equipment against the public recall record is a documentable infection prevention and control activity, and it is the kind of evidence an inspector can read.
What to do when a device you own appears in a notice
A recall notice is a prompt for a documented action, not a conclusion about your site. The steps that hold up are the ones written down while they happen: confirm whether your unit is affected by serial number, remove it from use if it is, contact the manufacturer, and log the date and the reference number. If the equipment is out of service for any period, the clinic's monitoring and record-keeping obligations for the remaining sterilizers continue unchanged, and loads processed before the notice was received should be reviewed against your own cycle records rather than assumed to be acceptable.
That last point is where record quality decides the cost of the event. A clinic that can produce cycle records, indicator results and maintenance entries can bound the question to a specific date range. A clinic that cannot has to treat the whole record as unknown, which is exactly the situation that produces large exposure notifications in other settings. The practical value of an inspection-ready record set is not the inspection; it is the ability to answer a bounded question quickly and cheaply when something like a recall notice arrives.
Related reading
- Autoclave failure rates and recall data in Canada: what the published numbers show
- Rebuilding a sterilization log after a public health inspection
- What Ontario clinic inspections ask to see: the reprocessing evidence set
- Biological indicator testing frequency by facility type
- Bowie-Dick test failures: which retest and recall rules apply in Canada versus the United States
- Sterilization compliance hub: cadence, records and monitoring
- Wholesale and institutional ordering
- Sterilization monitoring collection
- B2B wholesale collection
- Biological indicator five-pack trial
- 24-hour self-contained biological indicator, 25-pack
- Autoclave indicator tape, 3-roll set
Frequently Asked Questions
Is there a published failure rate for sterilization equipment in Canada?
No. Health Canada publishes recall notices, which record recall actions against specific products and affected unit lists. That is a different quantity from a failure rate, which would require the number of events divided by the number of units in service and cycles run. Neither figure appears in the public recall record, so no rate can be derived from it.
Why can a MAUDE search result not be used as a failure rate?
Because the FDA states that it cannot. The agency's description of the database says MDR data is not intended to be used to evaluate rates of adverse events, evaluate a change in event rates over time, or compare adverse event occurrence rates across devices, and it attributes this to under-reporting, inaccuracies in reports, lack of verification of causation and lack of information about how often the device was used.
Does a recall notice mean the device caused harm to a patient?
Not by itself. A recall notice describes an action taken against a device and the issue that prompted it. The notice is not a finding about causation in any individual case, and the FDA publishes a statutory disclaimer stating that a submitted report does not necessarily reflect a conclusion by the reporter or the agency that the device caused or contributed to the reportable event.
In Ontario, how often must a sterilizer be monitored with a biological indicator?
In Ontario the expectation is a biological indicator test for each sterilizer on each day the sterilizer is used, together with a control test. A weekly cadence is a different baseline that appears in United States guidance and in some general standards, and it should not be described as the Ontario requirement.
How do I find out whether my sterilizer's serial number is affected by a recall?
Search the Canadian recall database by manufacturer name and by model designation, then open any notice that concerns your equipment and compare the affected product list against the serial numbers you hold. The Statclave G4 notice, for example, lists ten specific serial numbers for one model number, so the comparison is mechanical once you have the serial in front of you.
What should a clinic keep as evidence that it checked?
A dated record showing the search term, the database searched, the date, and the result, filed with the equipment's maintenance record. If a unit was affected, add the notice reference, the quarantine date and the manufacturer contact. This is the same log discipline that supports sterilization cycle records and indicator results.
Are the two databases ever useful to compare?
They are useful to read together for structural reasons, not for numerical comparison. The source, the reporter and the legal trigger differ, which explains why one set leans toward formal recall actions and the other contains a high volume of narrative reports from facilities. Reading them side by side shows how differently two systems capture the same category of problem.
Sources
- Health Canada — Recalls and safety alerts
- Health Canada — Statclave G4 Chamber Autoclaves recall notice
- Health Canada — recall search, term "autoclave", searched 29 September 2026
- FDA — MAUDE database search
- FDA — About the Manufacturer and User Facility Device Experience database
- FDA — MDR data files
- 21 CFR 803.30 — user facility reporting requirements
- 21 CFR 803.16 — disclaimer
- Medical Devices Regulations, SOR/98-282, section 59
CliniEco Medical supplies sterilization monitoring consumables for Canadian clinics and care settings under MDEL #35334.
0 commentaire