How we estimate pickleball injuries
Where the numbers come from, how we count, and what they can’t tell you.
The data source
Every number comes from the National Electronic Injury Surveillance System (NEISS), run by the U.S. Consumer Product Safety Commission (CPSC). NEISS collects records of injuries treated in the emergency departments of a national probability sample of US hospitals. Each record has a short text description of what happened and a statistical weight: the number of US emergency room visits that one record stands for. Adding up the weights gives a national estimate.
We download the yearly public case files from CPSC and run our own script on them. Our numbers are our own estimates, not official CPSC figures.
What the numbers can’t tell you
- They count emergency room visits only. Injuries treated at urgent care, a doctor’s office or at home are not included, so the true number of injuries is higher.
- They are estimates from a sample of hospitals, so each one has a margin of error.
- A case is found only if the hospital record mentions pickleball. Records that don’t are missed, and a few that do may describe an injury that happened near a court rather than during play. We don’t read and remove cases by hand.
- They don’t show risk per player. More visits can mean more people playing, not that the game got more dangerous.
How we find pickleball cases
NEISS has no product code for pickleball. The coding manual tells hospitals to code it as “other ball sports” (code 3235), which also covers other sports. So we search the case descriptions instead. We look for PICKLEBALL and the spelling variants used in a 2025 JAMA Ophthalmology study, ignoring letter case and spacing between words:
| Search term | Cases matched |
|---|---|
| PICKLEBALL | – |
| PICKLE BALL | – |
| PICKELBALL | – |
| PICKEL BALL | – |
| PICKLE RACKET | – |
| PICKLE RACQUET | – |
Case counts appear here after the first data build.
Which estimates we show
We follow CPSC’s rule: an estimate is unstable if it rests on fewer than 20 cases, if it is below 1,200, or if its coefficient of variation (CV, the standard error divided by the estimate) is above 33%. We mark unstable estimates and don’t show their numbers, but we always show the case count.
Margins of error
The variance method appears here after the first data build.
Age groups and labels
Age groups: 0 to 17; 18 to 39; 40 to 49; 50 to 59; 60 to 69; 70 to 79; 80 and older; age not known. NEISS records children under 2 in months; we count them in the youngest group. Body part, injury type and hospital outcome labels are the NEISS codes from the coding manual, in plain words.
How this site is made
AI tools help write the code and draft the text. A person checks every claim against its source before it is published. All sources are listed on the sources page.