Research study
Garage door injuries: what the federal data does and does not show
The injury numbers you see online name no source. Here is what federal records actually show, and how to check it yourself.
Written by HyreGarage Research Desk Primary-source research and data analysis
Audited by HyreGarage Research Desk Citation, scope and retrieval-date audit
The finding
No source gives a current national count of garage door injuries. The firm number is historical: CPSC Release 92-024 (December 3, 1991) recorded 48 children aged 2 to 14 killed by entrapment under automatic garage doors since 1982.
Those deaths led to the federal standard, 16 CFR Part 1211. We could not query NEISS, because cpsc.gov blocked our requests.
What did we find?
The real, primary number is almost never quoted. CPSC Release 92-024, dated December 3, 1991, recorded “48 children between the ages of two and 14 who died after becoming entrapped under doors with automatic garage door openers” since 1982. Those 48 deaths are why Public Law 101-608 § 203 and 16 CFR Part 1211 exist.
We cannot give a current national injury estimate for garage doors. cpsc.gov returned HTTP 403 (access blocked) to every request from our environment. So we did not query NEISS, the federal emergency-room injury database.
Instead we publish what makes the query repeatable: the four NEISS product codes, set out below. An estimate that does not name its codes cannot be checked.
One peer-reviewed NEISS study exists, and its scope matters most. Shapovalov and colleagues, in the American Journal of Emergency Medicine (January 2026), studied automatic doors in general, not garage doors alone, over 2015–2024.
They estimated 38,720 weighted emergency department visits. Older adults were admitted 18.4 percent of the time, against 4.2 percent of adults. Whether that age pattern holds for garage doors is an open question, not a finding.
What are the limits of this data?
On an injury page, getting the boundaries wrong is how bad statistics get made.
cpsc.gov returned HTTP 403 to every request from our environment on the retrieval date. So this page has no HyreGarage count of garage door injuries. Every figure here is credited to whoever produced it.
The study covered injuries from automatic doors. Garage doors are one kind; shop entrances and powered sliding doors are others. Do not quote 38,720 as a garage door figure. We repeat this warning wherever the number appears.
They cover 1982 to 1991, before the federal standard applied to a single opener. They explain why the law exists. They say nothing about the risk from a compliant opener today.
A powered garage door (138), an unpowered garage door (1886), the garage building (1830) and a non-garage automatic door (137) are four different things. A statistic that does not name its codes cannot be judged.
The garage door injury numbers that fill search results trace, in our checking, to law firm marketing pages. They carry no product code, no query and no retrieval date. Leaving them out is deliberate.
NEISS samples hospital emergency departments across the country. It is not a full count. Estimates carry sampling error, and small groups carry a lot of it. Read any single-year, single-product NEISS number with that in mind.
Which 48 deaths led to the federal law?
CPSC counted 48 children killed under automatic garage doors between 1982 and 1991. Most accounts of opener safety mention this period. Very few cite the count, even though it is on the record.
CPSC Release 92-024, dated December 3, 1991, announced a joint program between the Commission and the Industry Coalition for Automatic Garage Door Opener Safety “to address the problem of child entrapment by garage doors with automatic openers.”
The release records “48 children between the ages of two and 14 who died after becoming entrapped under doors with automatic garage door openers” since 1982.
HyreGarage analysis: 48 deaths over about a decade is small in raw terms and huge in regulatory terms. It was a clear group (young children), a well-understood cause, one product and a fix that already existed.
That mix is what moves federal safety rules, and it moved this one fast. Congress wrote the requirement straight into law instead of asking the Commission to write a rule.
The order of events matters. Public Law 101-608 was approved on November 16, 1990. The CPSC release came 13 months later, in December 1991. By then the first legal deadline had passed and the second was a year away.
The Commission published its rule at 57 FR 60455 on December 21, 1992. The second requirement, a separate sensor that acts before the door touches anyone, applied to openers made from January 1, 1993. Our UL 325 and federal rule timeline lists each step.
The 48 tells you nothing about risk today. It is a count from before the standard. Its value is as the reason for the rule, and as the baseline any claim that the rule worked must be measured against. Measuring that needs the NEISS access we could not get.
How does federal injury data classify garage doors?
NEISS files each injury under a product code. These are the codes that matter for garage doors. Most figures you see online never say which codes they counted, or whether they added several together.
| Code | NEISS product label | What it actually covers |
|---|---|---|
| Code 138 | Automatic garage doors or door openers | The powered door. This is the code that matches the product Congress legislated about in 1990. |
| Code 1886 | Garage doors (excl. automatic garage doors) | The unpowered door. Separating these two is essential and almost never done in circulating figures. |
| Code 1830 | Separate garages or tool sheds | The structure rather than the door. Included here because it is frequently swept into “garage injury” totals. |
| Code 137 | Automatic doors or door openers | Non-garage automatic doors — shop entrances, sliding doors. A different product, and the code most likely to be confused with 138. |
Product codes from the NEISS format definitions file, retrieved 2026-09-06. Note how close 137 and 138 are: “automatic doors or door openers” and “automatic garage doors or door openers” differ by one word and a whole product category. HyreGarage did not query NEISS; we publish these codes so a reader with access can.
What does the one peer-reviewed study cover?
It covers automatic doors of all kinds, not garage doors alone. The only modern peer-reviewed analysis we found is: Shapovalov V, Lagunzad I, Tran QK, Groussis M, Abdelsayed N, Pourmand A. “Injury patterns and epidemiology of automatic door-related trauma: A retrospective analysis.” Am J Emerg Med. 2026 Jan;99:411–415. DOI 10.1016/j.ajem.2025.10.042, Epub October 24, 2025.
The authors describe a look-back study “with the use of the National Electronic Injury Surveillance System (NEISS) to identify Emergency Department visits related to automated doors between 2015 and 2024”, with weighted national estimates.
Automatic doors, not garage doors. Garage doors are inside that group, and so are supermarket entrances and powered sliding doors. The abstract does not say which product codes were used. So no one reading it can tell what share of the estimate is garage doors.
With that said, the study estimates 38,720 weighted ER visits over the ten years. Of those, “11 % were pediatric, 52.4 % were adult, and 36.5 % were older adult”, with average ages of 6.97, 44.37 and 78.21.
Injuries differ sharply by age. “Upper extremity injuries predominated in pediatric (56.8 %) and adult (44.7 %) groups, while older adults more frequently sustained head/neck (39.1 %) and trunk/internal injuries (23.8 %).”
Outcomes follow the same pattern: “Most patients were discharged, but admission rates were highest in older adults (18.4 %), compared to adults (4.2 %).”
In the statistical model, “pediatric patients had lower odds of severe injury (OR 0.03, 95 % CI 0.003–0.332) while older adults had higher odds (OR 3.18, 95 % CI 1.33–7.56) compared to adults.”
The authors also report a trend over time: “Injury prevalence peaked in 2015, declined in subsequent years, and rebounded in 2023–2024”, with severe injuries most likely in 2020 and 2024.
HyreGarage analysis, offered as a question, not a finding: public safety messaging about powered doors is about children, because children are why the standard exists. This study puts over a third of visits, and most serious outcomes, among older adults.
A head or trunk injury in someone aged 78 on average is a different event from a hand injury in a six-year-old. An 18.4 percent admission rate is serious.
We are not saying this describes garage doors. It may. The cause is plausible, and older adults spend a lot of time in garages. But the study did not separate garage doors, and we could not run the query that would. For now it is the most interesting open question here, not an answer.
Older adults are hurt most seriously by automatic doors
From the same study, with the same limit: automatic doors in general. The column to read is the share of each group admitted to hospital instead of sent home.
| Age group | Share of visits | Mean age | Predominant injury | Admitted | Odds of severe injury |
|---|---|---|---|---|---|
| Pediatric | 11% | 6.97 (SD ±0.57) | Upper extremity injuries predominated (56.8%) | lowest | OR 0.03 (95% CI 0.003–0.332) for severe injury, against adults |
| Adult | 52.4% | 44.37 (SD ±0.75) | Upper extremity injuries predominated (44.7%) | 4.2% admitted | reference group |
| Older adult | 36.5% | 78.21 (SD ±0.61) | Head/neck (39.1%) and trunk/internal (23.8%) | 18.4% admitted | OR 3.18 (95% CI 1.33–7.56) for severe injury, against adults |
Figures as published in the study abstract: an estimated 38,720 weighted ER visits, 2015–2024, for automatic door injuries. NEISS is a national sample, and the abstract does not give error ranges for subgroups. HyreGarage computed none of these figures and did not query NEISS.
What does the federal standard protect against?
The injury picture is incomplete, but the rule itself is fully on the record. It also shows where the remaining risk sits.
Reversing on contact, since 1991
16 CFR 1211.7 requires built-in (“inherent”) entrapment protection: the opener senses resistance and reverses. Where it tracks door position it “shall measure or monitor the position of the door at increments not greater than 1 inch”.
The limit is built in: the door has already touched whatever it reversed off.
Detecting before contact, since 1993
A second, separate safeguard is required: a photo-eye sensor (a light beam across the opening), an edge sensor, an independent door sensor, or an equivalent device.
Each must reverse a closing door to fully open and “prevent an operator from closing an open door”. This layer protects a child before the door strikes.
A way out without power
§ 1211.9 requires a manual release whose handle “shall be colored red and shall be easily distinguishable from the rest of the operator”. It must adjust to six feet above the floor and release under at most 50 pounds of force with the door blocked in the down position.
What it does not cover: the springs
Springs, cables and drums (the counterbalance, which carries the door’s weight) are not entrapment protection and are not in Part 1211. A failing spring is a separate hazard with a separate cause. Spring and cable work is for a trained technician; see garage door spring repair.
What it does not cover: doors already installed
The standard governs what may be made and sold. An opener made before 1993 was legal when made and stays legal in place. Nothing requires an upgrade.
What it does not cover: what happens later
A sensor that is misaligned, blocked or deliberately bypassed leaves a compliant opener with no working backup protection. No part of the federal system detects or records that. Our photo-eye sensor failure study covers it.
How can you get the number we could not?
If you can reach CPSC’s systems, as a researcher, a journalist or anyone outside our network limits, this query answers the question properly. We would rather publish the method than a borrowed figure.
CPSC runs an online NEISS estimates tool. On the retrieval date its front page said “NEISS data for 2025 Treatment Dates are available for query and download as of May 7th, 2026.” It asks for your affiliation before running a query.
“Automatic garage doors or door openers” is the powered product, and the one the federal standard covers. Run it on its own before combining anything.
“Garage doors (excl. automatic garage doors)” is an unpowered door. Merging 138 and 1886 gives a bigger, less useful number. How the two differ over time is far more telling than their sum.
“Separate garages or tool sheds” is the building, not the door. Including or excluding it are both fair choices. Doing either silently is not.
Given what the automatic-door study found, the age split is the interesting output. As far as we can tell, no one has checked whether garage doors show the same older-adult severity pattern.
Leaving this out is what makes almost every shared garage door injury statistic useless. A NEISS estimate without its product codes cannot be repeated, checked or compared.
What do these terms mean?
- NEISS
- The National Electronic Injury Surveillance System: CPSC’s national sample of hospital emergency departments. It produces weighted national estimates of product-related ER injuries. It is not a full count, and it misses injuries treated elsewhere or not treated at all.
- Weighted estimate
- A national figure scaled up from the sampled hospitals. Every such figure carries sampling error, and the error grows as the group gets smaller. That is why single-product, single-year NEISS numbers need caution.
- Product code
- The category NEISS assigns an injury. Codes 138, 1886, 1830 and 137 matter here, and they cover four different products. A statistic that does not name its codes cannot be repeated.
- Entrapment
- The hazard the federal standard addresses: a person or object caught under a closing door. It is named in three of the four garage door recalls on the CPSC record (see our full garage door recall list) and in the 1991 child death count.
- Odds ratio
- A measure of how likely something is in one group compared with another. The study puts older adults at OR 3.18 (95% CI 1.33–7.56) for severe injury against adults. The wide range means the effect is clearly there but not precisely measured.
- Release 92-024
- CPSC’s number for its December 3, 1991 news release announcing the joint program with the Industry Coalition for Automatic Garage Door Opener Safety. It is the source of the 48-death count on this page.
What could we not find?
NEISS itself. This gap shaped the whole page. cpsc.gov returned HTTP 403 to every request from our environment on the retrieval date. That covered the online query tool, the archived yearly data files and the site root.
CPSC’s SaferProducts REST API, a separate endpoint, answered normally. That is why our recall list exists and this injury count does not. This page has no HyreGarage estimate of garage door injuries, and will not until we can run the query.
The study’s product codes. The abstract says “automated doors” without naming the NEISS codes. Whether it used 138, 137 or both changes how much of its estimate is garage doors. We did not get the full text, so we report the study at the scope its abstract states.
Any official death count after 1991. CPSC’s 48-death figure stops in 1991. Whether entrapment deaths fell after the standard took effect is the key question about this rule, and we could not answer it. It needs CPSC’s death certificate and In-Depth Investigation files, which sit behind the same block.
The widely shared figures. We checked the garage door injury statistics that fill search results. We could not trace any to a query, a product code or a retrieval date. Several appeared on law firm marketing pages.
We are not repeating them, not even to debunk them. Repeating a number is how it survives. We would rather leave the space empty and say why.
One thing we chose not to do: take the 38,720 automatic-door estimate and assign some share to garage doors. There is no published basis for that split.
Any factor we picked would be invented, and the result would spread as a garage door statistic with our name on it. The scope note stays, and that math does not get done.
Questions
How many garage door injuries happen each year in the US?
How many children died from garage door entrapment?
Are garage door injuries mostly to children?
What is the NEISS product code for garage doors?
Do older adults get hurt more seriously by powered doors?
Did the 1993 safety standard reduce garage door deaths?
Why don’t you quote the injury numbers other sites use?
Is a modern garage door opener safe?
What does NEISS actually measure?
Where can I read the study cited here?
Written and audited by
HyreGarage Research Desk
Primary-source research, data analysis and fact checking
We are not a garage door company. We read the agency file, the code record, the standards document or the public register ourselves, compute the figure from it, and publish it with the source and the date we retrieved it.
Where a number cannot be traced to a primary source, we publish the shorter page and say what we could not verify. Our own company records cover ten states; nothing national is ever derived from them.
- 10
- states our own company records cover — and the limit of any claim made from them
- 3,901
- garage door companies in the store
- 457
- license records verified against a state board
- 0
- national claims made from a ten-state store
How this desk works
- Primary sources only. Injury counts come from CPSC. Housing counts come from the Census file, not from a summary of it. Code history comes from the building commission that adopted the code. We do not cite an article that cites a source; we retrieve the source and do the arithmetic ourselves.
- Every figure carries its retrieval date. Registers change, datasets are revised and codes are amended. A number without the date it was read cannot be checked, so every study states one.
- Fact, calculation and analysis are labeled apart. A quote is a quote, a HyreGarage computation says so, and an interpretation says “HyreGarage analysis”. Presenting our reading of a dataset as something the agency stated would be the easiest way to lose the only thing this desk is for.
- Limitations go above the fold. If a figure is an upper bound, a bracket, or an association rather than a cause, that is said before the figure is quoted rather than in a footnote underneath it.
- No DIY instructions for spring, cable or track work. Those components hold enough stored energy to cause serious injury, and CPSC records the consequences. We describe what has failed and what a competent repair involves; we do not tell you how to do it.
Data as of CPSC Release 92-024 and Shapovalov et al., Am J Emerg Med 2026, retrieved 2026-09-06. Authorship on this site is organizational: the analysis belongs to the desk rather than to a named individual, and we do not publish credentials we do not hold.
Our editorial policy sets out how we source, date and correct what we publish.
Sources & retrieval dates
Worried about the door your family uses every day?
Test the reverse and the photo-eyes using your opener’s own published steps. If it fails, pull the red release handle to disconnect the opener, and have it checked before anyone uses it again.
HyreGarage is not a garage door company and does not perform, supervise or warrant garage door work. This page reports published injury research. It is not medical or safety advice, and it contains no HyreGarage estimate of how often injuries happen.