Study limitations include lack of detailed clinical information with each report.
The reporting of ebike news articles should always be treated with a healthy grain of salt, for there is a lot of implicit bias that slips in. Even this fairly tame headline by NPR is falling into the trap, teetering on the suggestion that helmets are the sole preventative measure for avoiding head injuries.
The study -- not very long to read; mostly has tables of data -- details the rate of injuries observed in emergency rooms between 2017 and 2022, showing some monumental increases in that time period, split out by age, sex, hospital type, injury type, and helmet use.
What's not here -- as the study admits -- is the nature of the collision, which is rather important because grievous head injuries involve an impact with someone or something else. There is a substantial difference in how public policy would address head injuries involving: 1) solo bicyclist rider error, 2) stationary motor vehicles, 3) moving motor vehicles, 4) pedestrians, 5) collision with other bicyclists, or 6) stationary object collisions.
At best, this study tells us that bicycle collisions are up, but not the flip side: are pedestrian collisions up? Motor vehicle collisions up? More collisions with stationary objects? We simply don't know.
And in the absence of data collection to even attempt to answer that question, the wrong conclusion is often drawn: that ebikes are inherently more dangerous, which draws further legislative action by confused towns and cities, which eventually prove futile because they're not following any data.
There is reason to believe that these increasing rates of collisions are due in part to popularity of ebikes -- as the NPR as article mentions -- but in larger part by motor vehicles.
A Google Search for "ebike collision causes" sadly turns up mostly ads for lawyers, but I did find this 2017 study of Corvallis, OR data on all bicyclist collisions. Table 2 shows that the top cause is "motorist fails to yield while turning" at 42%. This study notes that their data did not identify ebike vs acoustic bike, but it's hard to see how that would make a difference if it's an error by the motor vehicle driver. The next three causes are faults of the driver as well.
Going down that table, the non-zero collisions attributable to bicyclist behavior are: running red lights, going the wrong way, not yielding while turning, no lights, darting into the road, lane changes, and intoxication. None of those, except maybe that an ebike can dart into the road quicker, are substantially amplified by an ebike compared to an acoustic bike.
IMO, the trend of increasing ebike head injuries is from: 1) motor vehicle collisions where the auto driver is at fault, and 2) insufficient infrastructure to separate bicyclists from pedestrians. Proper infra means bikes and pedestrians are protected from cars, and pedestrian and bike flows are separated.