Pickleball injuries rose 17-fold in a decade, but not per player: what the new surveillance study found
Short answer: A medRxiv preprint analyzing NEISS data estimates 101,450 emergency department-treated pickleball injuries from 2014-2023, finds no rise in injury risk per participant, and debunks search-interest proxies with a sourdough control.
The most-quoted number in American pickleball just got its most careful audit, and the result cuts in two directions at once. A preprint posted October 16 on medRxiv estimates that emergency department-treated pickleball injuries in the United States rose 17.5-fold from 2014 to 2023, yet finds no sign the sport grew more dangerous per player, and along the way it demolishes a popular shortcut, using Google search data, with an assist from sourdough bread.
What the study actually did
The paper, titled "Pickleball injuries, search interest, and participation in the United States, 2014-2023: an ecological time series study," pulls emergency department microdata from the National Electronic Injury Surveillance System, the federal injury-surveillance network. The authors adjudicated 1,738 pickleball cases, 921 of them orthopedic, which the NEISS weighting system scales to an estimated 101,450 injuries treated nationally over the decade. Participation estimates from the Sports and Fitness Industry Association supplied the player denominator, while Google Trends, English Wikipedia pageviews and GDELT news volume served as measures of public attention.
The headline numbers
Estimated emergency department-treated pickleball injuries climbed from 1,391 in 2014, a figure the authors themselves flag as highly unstable because of the tiny sample that year, to 24,294 in 2023. Seventy percent of adjudicated cases involved adults aged 60 to 79, matching the older-skewing injury profile clinics have reported for years.
The more useful question is what happened once the boom was underway. Between 2017 and 2023, injury estimates rose 4.0-fold while participation rose about 4.4-fold. The injury-to-participation measure showed no temporal trend across the decade, with a P value of 0.59. In plain terms: more Americans got hurt playing pickleball in absolute terms because so many more Americans were playing, not because each player's hour on court became more hazardous.
The sourdough lesson
The study's second half is a cautionary tale for anyone tracking the sport through search data. In raw annual levels, Google search interest for "pickleball" correlated with total injuries at R equals 0.96, and Wikipedia pageviews for pickleball at R equals 0.98, which looks like an excellent free proxy for injury surveillance. Then the authors ran their negative control: searches for "sourdough," a term with no plausible medical connection to pickleball, correlated with pickleball injuries at R equals 0.76. All three series simply rise together across a decade, and rising-together is not a relationship.
Once trend and seasonality were statistically removed, the apparent signal collapsed: a monthly residual correlation of 0.084, and 0.28 for year-on-year changes across the 96 months from 2015 to 2023.
Seasons that run opposite
The seasonal comparison is the cleanest demonstration. Estimated injuries peaked in February, while search interest peaked in July. On twelve-month mean profiles, the sourdough negative control actually fitted the injury season at R equals 0.96, while pickleball search interest ran against it at R equals negative 0.40. Why injuries peak in February is a genuinely open question the surveillance data cannot answer; winter indoor play and snowbird-season participation in warm-weather states are plausible contributors, but this study tests neither, and any claim about the cause would be speculation.
Search interest and news coverage did track each other in year-on-year changes from 2017 to 2023, at R equals 0.64, but their correlations with actual injuries over the same window were just 0.25 and 0.20. Attention and harm are simply not the same curve.
How the estimate is built
NEISS is not a tally of every emergency room visit in the country; it is a probability sample, collecting detailed case narratives from a network of about 100 hospitals and scaling those cases into national estimates through statistical weights. That is why 1,738 real cases become an estimated 101,450 injuries, and it is also why the authors attach confidence intervals to every annual figure and call the 2014 estimate unstable: when a rare activity produces few sampled cases, the weighted number swings widely. The adjudication step matters too. Coders reviewed each case narrative to confirm the injury actually involved pickleball, separating 921 orthopedic cases, fractures, sprains and the like, from the rest. It is careful, unglamorous work, and it is what makes the 17.5-fold rise worth taking seriously even before peer review.
The fine print
Two disclosures matter. First, this is a preprint: it has not completed peer review, and its numbers could shift in revision. Second, every author disclosed being a consultant for or employee of Meta Platforms, the company whose data-ecosystem tools include some of the attention signals studied. The design is also ecological, meaning it compares national-level curves rather than tracking individual players, and NEISS captures only injuries treated in emergency departments, not urgent care clinics or doctors' offices, so the true injury count is higher than any NEISS-based estimate.
Why players should care
For everyday players the reassuring finding is the flat per-participant trend: the sport did not become more dangerous per player during its fastest growth years, at least not in ways the surveillance system can detect. The concentration of cases among adults in their sixties and seventies reinforces what sports medicine clinicians keep saying about warm-ups, lateral-movement training and gradual returns after time off, especially for the Achilles tendon injuries that dominate pickleball's surgical literature.
What to watch
The next step is peer review, and whether journals accept the authors' argument that search interest cannot substitute for injury surveillance. Watch too for better denominators: several research groups are building exposure-based injury rates using court-booking and app data, which would finally let epidemiologists say whether a Tuesday evening at the local courts is getting safer or riskier, not just busier.
Sources
Sources checked 20 Oct 2026. Header image is illustrative, not a photo of the events described.
Related stories
Five new pickleball facilities open in October: Yucca Valley, Monticello, Maricopa, Palm Springs and Moab
Eight courts here, twelve indoor courts there, 22 in Palm Springs: what opened this month, what it cost, and the access rules players need to know.
Pickleball's club shakeout: chains grew 37% this year, and The Picklr still landed in Chapter 11
Six major pickleball chains grew from 145 to 199 locations between January and September 2026, but indoor court growth is slowing and The Picklr's parent filed for bankruptcy. The buildout continues through a shakeout.
Life Time is building a dedicated pickleball destination in Palm Beach Gardens
Life Time has opened a new 20,500-square-foot pickleball destination with six courts across from its Palm Beach Gardens club, part of an expansion that also adds a 24-reformer studio and 5,600 square feet of exercise floor.