Question:
There were 235 head-injury case patients out of 269 with head injuries and 776 total persons treated. Head injuries were defined as injuries in areas of the head that could reasonably be expected to be protected by a helmet; those with only minor head injuries were not classified as case patients. 116 of those with head injuries had injuries which fell within the upper 2/3 of the 6-level injury scale used. 99 patients had brain injuries (all but 4 unhelmeted). There were three deaths, all unhelmeted. A report is available of a larger study; see http://www.smf.org/report.html. The results of that study were similar, e.g. in 13 out of 14 deaths, the cyclists were unhelmeted (one 6 year-old child wearing a helmet was crushed under a truck). did you happen to learn how many serious head injuries were included in the study and how many deaths?

Answer: I prefer dealing primarily with fatality data. While there still seems to be no commonly accepted definition of precisely what constitutes a 'serious' injury (anything from a broken bone or an overnight hospital stay to a permanently incapitating injury?), at least everyone intuitively understands what 'an injury resulting in death' means. Likewise, fatality data (particularly traffic fatality data) is both far more complete and much easier to come by than data for less serious injuries. There were 235 case patients with serious head injuries, as defined above, and which excludes minor injuries. 116 of those patients had quite serious injuries, including extended periods of unconsciousness, skull fractures, or worse. Statistical significance is shown by the confidence intervals for the results presented by Thompson et al.; there is only a vanishingly tiny chance that helmets were not effective in mitigating serious head and brain injuries.

There were no "measurements" involved; the "85%" reduction in risk, a 6.67:1 higher risk of head injury for unhelmeted cyclists, is based on a statistical analysis of the numbers of serious head-injury case patients who wore helmets as compared to those in the population group, adjusting for age, education, etc. Bear in mind that the risk ratio might be as low as 3.45:1 or as high as 14.3:1; but it is almost certainly greater than 1.0:1 (the lowest of the lower limits that I've seen from any study was 1.20:1, which is still statistically significant evidence of higher risk of head injury by riding unhelmeted).