Trampoline Head Injuries: NEISS 2024 Odds for Under-Sixes
On 17 November 2026, PLOS One published a cross-sectional analysis of the 2024 National Electronic Injury Surveillance System file. It found 3,548 recorded trampoline injuries in children aged 0 to 18, which the survey weights turn into a national estimate of 101,474 emergency department visits for the year.
Of that total, 21.7% involved the mouth, face, head, eye, ear or neck. Weighted, that is 22,033 injuries. The same paper reports that children under six carried 1.66 times the adjusted odds of a head, face or neck injury compared with adolescents aged 12 to 18.
That ratio is the useful part for anyone building or refurbishing a trampoline park, and it is also the part most likely to be misread. It is an odds ratio drawn from emergency department records, not a rate per visit, and the dataset records neither the equipment nor the supervision in place at the time of injury.
This article does three things with the study. It sets out what was counted and how. It shows where the numbers can and cannot be pushed, including one result that runs against the obvious reading. Then it converts the finding into the equipment and staffing decisions that a park actually signs off on.
One disclosure before the arithmetic: we sell trampoline park equipment. Every figure below comes from the published paper or the public data it draws on, and every derived number is labelled as derived.
1. What the Study Counted
The authors, R. Constance Wiener and Christopher Waters of West Virginia University, used the 2024 NEISS file, which collects product-related emergency department visits from a sample of about 100 hospitals and weights them to represent roughly 5,300 emergency departments nationally. Trampoline cases were extracted with product code 1233.
Children were split into three bands: under 6, 6 to under 12, and 12 to 18. Six body-part codes were combined into a single oral-facial-head group, because the individual codes were too thin to analyse separately. Disposition, race and location were carried as descriptors or covariates.
| Measure | Weighted estimate | Share | Sample cases |
|---|---|---|---|
| All trampoline injuries, ages 0 to 18 | 101,474 | 100% | 3,548 |
| Oral-facial-head injuries | 22,033 | 21.7% | 730 |
| Other body parts | 79,441 | 78.3% | 2,818 |
| Age under 6 | 33,465 | 33.0% | 1,332 |
| Age 6 to under 12 | 52,692 | 51.9% | 1,781 |
| Age 12 to 18 | 15,318 | 15.1% | 435 |
| Male | 54,785 | 54.0% | 1,845 |
| Female | 46,689 | 46.0% | 1,703 |
| Injury at home | 24,287 | 23.9% | 782 |
| Injury at a recreation location | 27,818 | 27.4% | 1,037 |
| Location not recorded | 47,532 | 46.8% | 1,674 |
| Treated and released | 98,431 | 97.0% | 3,342 |
| Treated and admitted | 2,100 | 2.1% | 159 |
Two checks are worth running on a table like this before trusting it. The three age bands sum to 101,474 exactly. The six body-part components sum to 22,032 against a stated 22,033, a rounding difference of one. Those are the marks of a table transcribed from the source rather than reconstructed from memory.
The internal split of the head group matters for equipment, because the parts are not equally exposed. Head injuries were 9,309 (9.2%), face injuries 5,537 (5.5%), neck injuries 4,021 (4.0%), mouth injuries 2,226 (2.2%), eye injuries about 545 (0.5%) and ear injuries about 394 (0.4%). Head and face together account for roughly two thirds of the group.
Ninety-seven in a hundred of these children were treated and released. Just over two in a hundred were admitted. That is the severity profile of a consumer injury dataset, and it is the reason the numbers cannot be dropped straight into a risk model for a commercial venue.
2. The Odds Ratio, Read Correctly
The paper reports both unadjusted and adjusted odds ratios, with adolescents aged 12 to 18 as the reference group. Adjustment controls for sex and for the location where the injury happened. It does not control for hours of use, equipment type or supervision, because NEISS carries none of those fields.
| Age band | Head, face or neck share within band | Per 1,000 injuries | Unadjusted OR (95% CI) | Adjusted OR (95% CI) |
|---|---|---|---|---|
| Under 6 | 25.0% | 250 | 1.77 (1.17 to 2.67), p = 0.0071 | 1.66 (1.11 to 2.50), p = 0.0155 |
| 6 to under 12 | 21.3% | 213 | 1.44 (1.03 to 2.00), p = 0.0312 | 1.36 (0.99 to 1.90), p = 0.0759 |
| 12 to 18 | 15.9% | 159 | reference | 1.00 by definition |
The conversion to a per-1,000 basis is simple division. Head-group injuries inside the under-six band, 8,371, divided by the 33,465 total injuries in that band, give 25.0%, or 250 per 1,000. The same operation on the middle band gives 11,232 divided by 52,692, or 213 per 1,000. The oldest band gives 2,430 divided by 15,318, or 159 per 1,000.
Read that as a gradient rather than a cliff. The gap between the youngest and oldest band is 9.1 percentage points, and the middle band sits between them. What the regression adds is a confidence interval around the youngest band that stays clear of 1 after adjustment, and an interval around the middle band that does not.
That second detail is the one most summaries of this paper will skip. The adjusted ratio for ages 6 to 12 is 1.36, with a confidence interval running from 0.99 to 1.90 and a p value of 0.076. The paper does not claim a statistically significant difference for that band. Anyone who reports that all children under 12 face higher odds is reporting more than the study found.
The adjustment is instructive in itself. Moving from 1.77 to 1.66 cut the ratio by 0.11, about six percent of its value. Sex and location explain very little of why the youngest children arrive with head injuries. Whatever drives the pattern sits mostly inside the child, which is the argument the authors make in their discussion.
Sex is the stronger covariate in the same model: males had 26.4% head-group injuries against 16.2% for females, an adjusted ratio of 1.86 with an interval from 1.49 to 2.33. For a venue operator that is a staffing signal as much as a design one, and it lines up with the injury and supervision material in our piece on injury rates by zone.
3. The Result That Runs Against the Obvious Reading
If commercial trampoline parks were the main source of head injuries, the location breakdown should show it. It does not. Recreation locations produced 27.4% of all trampoline injuries, but only 14.4% of the injuries recorded at those locations were head-group. Home produced 23.9% of all injuries, and 28.5% of the injuries recorded there were head-group.
| Location code | Share of all injuries | Head-group share within that location | Adjusted OR for head-group injury |
|---|---|---|---|
| Home (reference) | 23.9% | 28.5% | 1.00 |
| Recreation | 27.4% | 14.4% | 0.45 (0.31 to 0.67) |
| Public property | 1.2% | not stable in the source table | 0.78 (0.33 to 1.86) |
| Not recorded | 46.8% | 22.4% | 0.73 (0.49 to 1.09) |
An adjusted ratio of 0.45 means that, in this dataset, an injury sustained at a recreation location had about 55% lower odds of involving the head group than one sustained at home. The interval is tight enough to take seriously, running from 0.31 to 0.67.
Two readings are plausible and the data cannot separate them. The first is compositional: a trampoline park produces a different injury mix, with more ankles, knees and wrists from landings on a shared floor, while a home trampoline with a single jumper produces more falls onto grass, frames and surrounding objects. The second is behavioural: home use involves less supervision and more mixed-age jumping, which the authors name directly when they discuss the hazards of mixed-age jumping.
Neither reading lets a park off. A lower share of head injuries inside a much larger volume of injuries can still be a larger absolute number of head injuries, and this dataset cannot test that, because 46.8% of the location field is missing.
Where this stops: recreation is a NEISS location code. It covers commercial venues, clubs, camps and organised settings together, and the paper does not disaggregate them. There is no park-specific rate in this study, and we are not going to manufacture one.
4. Why the Youngest Band Behaves Differently
The authors explain the finding in biomechanical terms: a higher head-to-body mass ratio, and therefore a higher centre of gravity, thinner cranial bones, and less developed cervical musculoskeletal support. A three-year-old falling from the same height as a ten-year-old is not exposed to the same event.
That sentence reorganises how a build should think about height. Most equipment discussions treat fall height as the variable and the surface as the control. For the under-six band the head arrives first and the neck absorbs more of the deceleration, so the parts that matter are the ones the head meets on the way down and immediately after: the padded edge of the frame, the spring pad, the gap between beds, the exposed steel at the perimeter, and the barrier that stops a child continuing into the next bay.
The paper’s own event-phase recommendations follow that logic. It names mandatory safety enclosures and improved padding at frame edges as controls that reduce severity in a fall or collision. In the pre-event phase it names age-based access restrictions and parent education that addresses mixed-age jumping rather than general awareness.
Note what is absent from that list: nothing about making the middle of the bed softer. The injury pattern concentrates at the boundary of the play surface, not at its centre. That is a purchasing statement rather than a medical one, and it is the part of the paper an equipment brief can use.
5. Three Equipment Decisions This Touches
The study does not prescribe equipment. It does tell you where to spend the next dollar if the under-six band is part of your business. Three decisions follow, in the order they get made on a project.
5.1 Whether under-sixes get their own footprint
What it is A physically separate court or soft-play bay with its own entry, sized for children under six, rather than a corner of the main floor with a sign on it.
What it decides It decides whether the under-six band meets adolescent jumpers at all. The paper recommends age-based access restrictions and singles out mixed-age jumping as a hazard that parent education must address. A separate footprint is the physical version of that restriction, and it holds when the briefing at the door does not.
What to verify Ask for the entry detail: can a four-year-old open the gate alone, and can a parent see the whole bay from the seating area? Then ask what the bay is built to. Soft contained play sits under ASTM F1918, whose stated user envelope runs from the fifth-percentile two-year-old to the ninety-fifth-percentile twelve-year-old, and trampoline courts sit under ASTM F2970. Do not let a supplier quote one standard for the other. Our safety standards page sets out how we document that split per project.
5.2 Whether edge padding is specified as a component or a finish
What it is The pad that covers the frame, the springs and the gap between beds. On many quotes it appears as a line item priced by the metre, alongside the vinyl and the labour.
What it decides It decides what a head meets at the perimeter, which is where this dataset says the youngest children get hurt. It also decides your replacement cost, because padding is a wear item, not a structural one.
What to verify Four things in writing: the attachment method and whether it survives a child pulling on it, plus the foam specification rather than the cover colour, the compression test the factory runs, and the lead time and price for replacement pads for the exact model you are buying. That last clause decides your year-four budget, and our article on the equipment upgrade clause works through how to write it.
5.3 How the under-six court is staffed and priced
What it is A session model: open all-ages jumping, banded sessions by age, or a separately ticketed under-six slot with a supervising adult required.
What it decides It decides supervision cost per hour and ticket yield per square metre at the same time, and those two move in opposite directions. The commercial models are compared in our piece on toddler zone economics, so this article will not reprice them.
What to verify The ratio you can actually staff on a Saturday, not the one in the operations manual. The 1:15 figure that circulates in this industry came out of an Australian sentencing decision as a remedial measure, not out of a code, and we set the case out in the 1:15 ratio article.
6. The Five-Year Clock Nobody Puts in the Safety Section
The paper’s introduction carries a detail that belongs in every maintenance schedule. Citing the American Academy of Pediatrics policy statement, it notes that the typical lifetime of a trampoline is five years, and that the owner is expected to inspect and replace the padding, frame, springs, mat and enclosure during that period.
Set that next to the injury pattern. The components expected to be replaced inside five years are the same components that sit at the boundary where the head-group injuries concentrate. Padding is not a cosmetic finish that outlives the warranty. It is a consumable with a service life, and it degrades in ways a walk-round does not catch: compression at the frame edge, cover abrasion over the spring line, and attachment points that loosen as the mat settles.
| Component | Why it matters for this injury pattern | Replacement trigger | Who signs off |
|---|---|---|---|
| Frame and perimeter padding | First surface a falling head meets at the edge | Compression set, cover split, loose attachment | Operations manager, monthly walk |
| Spring pad | Covers the spring line between beds, the second contact surface | Ride-up, gap exposure, hook penetration | Maintenance lead, weekly check |
| Bed mat | Landing surface; wear changes rebound and drop-off at the edge | Visible seam stress, loss of rebound, edge fray | Technical inspection, quarterly |
| Enclosure and barriers | Stops a child continuing into the next bay | Loose fixings, torn mesh, gate that no longer self-closes | Duty manager, daily opening check |
| Exposed steel at entries | Hard edge where children step on and off | Any exposed section, however small | Duty manager, daily opening check |
The triggers above are inspection categories, not tolerances. The tolerances come from the manufacturer’s specification and from whatever the local authority enforces, and they differ by market and by product generation. We do not publish a millimetre figure here, because a number copied from one jurisdiction into another is worse than no number at all.
What a buyer can do is insist the schedule exists before opening. Our breakdown of the five-year maintenance bill shows how the line items stack up; the point to take from this paper is narrower. If the pad schedule slips in year three, the exposure that rises first is the exposure this study measured.
7. Session Banding, and What It Costs
Age banding is the intervention the authors push hardest, and it is the cheapest one on paper. It costs nothing to print a session timetable. It costs a great deal to enforce one on a Saturday afternoon when a family with a nine-year-old and a four-year-old arrives at the same counter.
Three models are in common use, and they fail in different ways.
| Model | What the under-six gets | Main failure mode | What to calculate before choosing |
|---|---|---|---|
| Open all-ages jumping | A corner of the main floor with a sign | Mixed-age collisions; a sign is not a barrier | Supervisor hours per open hour; incident log by age band |
| Banded sessions | Dedicated time slots by age | Families split across slots and stop coming | Tickets per slot against tickets per family group |
| Separate under-six court | Own footprint, own ticket, adult required | Area taken from revenue-generating attractions | Area cost per square metre against yield per visit |
The calculation that decides between them is not complicated, and it does not need a forecast. Take the area each model removes from saleable capacity, multiply by the yield per square metre you already achieve, and compare that with the staffing cost of the model plus the exposure you remove. Every term in that sentence is either a number you know or a number you can measure in the first month.
Area cost of the model = square metres removed × yield per square metre Staffing cost of the model = attendant hours per week × loaded hourly cost Net position = area cost + staffing cost - value of exposure removed
The third term is the one nobody puts a number on, which is why the decision usually gets made on the first two. The study cannot price it for you either: it has no cost data, no venue denominator and no equipment field. What it can do is tell you which band the exposure sits in, and that is the under-six band.
One operational point that does not appear in the paper but follows from it: sightlines. A separate under-six court that a parent cannot see from the seating area turns a design control into a staffing cost, because someone has to stand at the barrier. Our article on supervision sightlines covers how that gets designed out.
8. What 101,474 Divided by 842 Parks Does Not Tell You
The paper cites a directory count of more than 842 trampoline parks in the United States as of 2025, alongside the first park opening in 2004. Put that beside the injury estimate and a tempting number appears: 101,474 divided by 842 gives about 120 emergency department visits per park per year.
That number is a ceiling, not an estimate, and it is wrong for at least three reasons. Home injuries were 23.9% of the total, so the park share is necessarily lower. Recreation-location injuries include organised settings that are not trampoline parks. And 46.8% of records carry no location at all, which means the park share cannot be recovered by subtraction either.
The same trap sits inside the age data. Under-sixes were 33.0% of injuries, but the paper publishes no participant count, no visit count and no exposure hours. Without a denominator there is no rate, and without a rate there is no honest way to say one park design is safer than another. What the study supports is a comparison between age bands inside one dataset, and that comparison is the odds ratio in section 2.
Two further limits belong in any citation of this work. The design is cross-sectional, so it establishes association and not causation. And NEISS records only injuries that reached an emergency department, so every minor incident a park handles with an ice pack is invisible in these totals. The authors say as much themselves, listing the missing equipment, supervision and hours-of-use fields as the study’s main weakness.
9. What to Put in the Brief
If the under-six band is part of your model, four lines belong in the tender document rather than in a conversation after the layout is fixed. They cost nothing to write and they change what gets quoted.
- State the age policy in the brief, not the brochure. Say which bands use which court, and ask the supplier to price a layout that makes that policy physical.
- Ask for the standard by number and scope. ASTM F2970 for trampoline courts, ASTM F1918 for soft contained play, EN ISO 23659 for trampoline parks in Europe. One standard per element, no mixing, and the certificate must name the model.
- Put padding into the replacement schedule at signature. Model number, pad part number, price, lead time and the person who inspects. This is the item the study’s five-year note turns into a safety control.
- Ask for the sightline drawing with the furniture in it. Not the equipment plan. The one that shows where a parent sits and what they can see from there.
A supplier who has built for age-banded operation will answer all four from documents. A supplier who has only built open floors will answer the first two and improvise the rest. That difference is visible before you sign, and it is the practical use of a paper that otherwise sits behind a journal paywall.
If you are at the layout stage now, our design and delivery process starts with the age policy for this reason, and the trampoline range page lists the court configurations we build most often.
FAQ
What percentage of trampoline injuries involve the head or face?
21.7% in the 2024 NEISS analysis, which is 22,033 of an estimated 101,474 emergency department visits for children aged 0 to 18 in the United States. Head injuries alone were 9.2%, face injuries 5.5%, neck 4.0% and mouth 2.2%. The study was published in PLOS One on 17 November 2026.
Are trampoline parks worse for head injuries than home trampolines?
This dataset points the other way, with a caveat. Injuries at a recreation location had an adjusted odds ratio of 0.45 for a head, face or neck injury compared with injuries at home, meaning lower odds. But the location field is missing for 46.8% of records, and the recreation code lumps commercial venues with clubs and camps, so it is not a park-versus-home comparison.
At what age should a child be allowed on the main court?
The study gives no cut-off. It reports that under-sixes had 1.66 times the adjusted odds of a head-group injury compared with 12 to 18 year olds, and it cites an earlier recommendation to raise the minimum trampoline age to 10. The age policy is an operator decision, and it belongs in the layout brief rather than on a sign.
Does better edge padding reduce head injuries?
The authors list improved padding at frame edges and mandatory enclosures as event-phase controls that reduce severity in a fall or collision. They do not quantify the effect, and the dataset has no equipment field, so no reduction percentage can be attached to a padding specification. Treat it as a severity control, not a substitute for age banding.
How often should trampoline park pads be replaced?
The paper notes a typical trampoline life of five years, during which the padding, frame, springs, mat and enclosure are expected to be inspected and replaced. The inspection interval and the acceptance criteria come from the manufacturer’s specification and local enforcement, not from this study, and they differ between products.
Sources
- PLOS One, “Age as a determinant of trampoline-related oral-facial-head injuries in pediatric emergency department visits: A NEISS analysis”, published 17 November 2026. journals.plos.org
- Wiener RC, Waters C. PLOS One 21(11): e0362482, DOI 10.1371/journal.pone.0362482. Full text also at pmc.ncbi.nlm.nih.gov
- U.S. Consumer Product Safety Commission, National Electronic Injury Surveillance System, 2024 dataset and query system. cpsc.gov
- ASTM International, F2970, standard practice for trampoline courts, covering design, manufacture, installation, operation, maintenance, inspection and major modification. astm.org
- ASTM International, F1918, standard safety performance specification for soft contained play equipment. astm.org
- Briskin S, LaBotz M. “Trampoline safety in childhood and adolescence”, Pediatrics 2012;130(4):774-779, cited in the study for the five-year service life and the replacement items. publications.aap.org
- ActionParkSource, trampoline park location directory, 2025 count of more than 842 United States parks, as cited in the study. actionparksource.com
