banner of Methodology for Determining Motorcycle Operator Crash Risk and Alcohol Impairment

3. Detailed Report of Discussion

Injury Crash Records

One way to determine the effect of BAC on crash likelihood is to compare BACs of crash-involved riders with BACs of the rider population at large. One possibility for obtaining BACs and other pertinent data from riders involved in crashes would be from hospital records. A study that uses injury crash data would need to get the BAC data for the riding population at large from some other source.

In this type of study, researchers would rely on hospital records of BACs for crash-involved riders at the hospital. Not all hospitals collect BAC data as a matter of course. By some counts, only 10 percent of injury cases have a BAC measurement (Liedtke & DeMuth, 1975; Flamm et al., 1977). In hospitals where it is measured only occasionally, measurements tend to be in cases where hospital staff suspect the patient has been drinking. This causes the BACs measured to skew high. For this reason, it is necessary to work with hospitals that can record BACs for all or nearly all of their motorcycle crash-involved patients. Level I Trauma Centers are more likely to have protocols that encourage recording of BACS for all patients, though these procedures may not always be followed strictly. In some cases, hospitals may have been recording sufficient data, for long enough, that archival hospital data can be used for this study. In other cases, arrangements may need to be made with the hospital to get them to start recording some of the data. Because a given hospital may see relatively few motorcycle crash victims, it may be necessary to use several years’ data to get a large enough sample.

Because this methodology only provides data on crashes, consideration must be given to the issues, advantages and disadvantages that apply to the method used to collect comparison data.


There are many more crashes involving injury than there are fatal crashes, therefore more data are available for analysis for a given time period than are available from using fatal crash data alone. Using injury crash data also allows generalization to injury crash population. It may be possible to obtain additional data by linking to larger database (e.g., through the use of the NHTSA’s Crash Outcome Data Evaluation System (CODES) data system).


This methodology provides only half of the data necessary to generate relative risk curves; it would still be necessary to get comparison data from non-crash-involved riders. This methodology would exclude PDO and minor-injury crashes. It may be harder to get BACs on all cases than it is in some other methodologies, depending on the hospital, policies and other factors. Time between crash and BAC measure may be greater than for Contemporary Case Control method, which may make BAC measures less reliable. It may be difficult to determine where the crash took place for the purposes of finding matching comparison case, although in-hospital follow-up may provide sufficient data, as would obtaining a copy of the Traffic Collision Report (TCR), which could be a source of much of the data for the case. There is a range of completeness of BAC data in hospital records: some hospitals collect a lot, some collect very little. Due to the tendency to collect BAC data more often when the evidence of drinking is more obvious, anything less than 100 percent testing has the potential to result in overestimated average BACs.

Not all motorcycle crash victims will go to the hospital, and not all who go will be admitted. Riders with minor injuries may not go to the hospital or be admitted. Single-vehicle crashes in which the rider can still operate the vehicle often lead to emergency room visits, but very often entail no police report, so that information is lost. Fatally injured riders may be pronounced dead at the scene and not go to the hospital. Often, BAC and other data will not be collected until the victim has been admitted. Because riders with minor injuries may not be represented in the sample of crashed riders, this necessarily skews the sample toward more severely injured riders.



Because this method uses crash data collected as part of the hospital’s procedures (assuming the cooperation of the hospital), the costs of obtaining crash data would be relatively low, although the cost of additional BAC tests may have to be considered. The main cost of conducting this study would be obtaining comparison data through roadside surveys (Case Control, Geo-General Comparison Data, or some other method). The cost of obtaining comparison data, then, could be considerable.