Helmet statistics are generated by pro-helmet researchers and the studies that support helmet use are flawed.
Response: Many attempts we have seen in the U.S. to measure helmet effectiveness or helmet use nationally based on general statistics of bicycles on the road or rider surveys are indeed not useful. Anyone who has worked with bike riders in this country would not ask the question "How much did you ride last year" and expect a valid answer. Exposure data calculated from such surveys, including the survey by CPSC, are not in our opinion valid. The New York Times published an article in 2001 written by a reporter who missed that point, and reached some startling -- and invalid -- conclusions. We have a page up on that article. But we do consider valid a number of clinically-based studies in the U.S. on helmet effectiveness. We have parts of several of them included in our statistics page, including the landmark study by Thompson, Rivera and Thompson done at Harborview Injury Prevention Center and published in the New England Journal of Medicine in 1989. The same team completed another study that adds helmet analysis to the clinical data. You can find it on the Snell Foundation Web site. The study is old, but helmets sold in the US have actually improved since the data was collected, so we regard the conclusions as still valid and probably conservative. More importantly, the protective effect of helmets has been demonstrated in the field so thoroughly over a period of decades that statistical analysis based on poorly gathered data adds nothing to the knowledge base.