This data is close to useless because it is not normalized into the number accidents per mile driven. Instead, it is provided as the number of years between accidents. The problem is that those who live near big cities typically have much longer commutes and therefore will put on more miles per year - which increases their exposure and absolute probability of having an accident. The traffic density is also higher - which also increases the probability of accidents. In order to have a truly meaningful comparison, one must measure the number of accidents over an equivalent period of exposure (i.e., per mile driven).
To the extent that accident insurance rates take into account where you live and drive, the data are quite useful for the purpose of setting those rates. For that purpose, there is no need to normalize for miles driven.