Traffic accident analysis using accident database can indicate important causes and useful countermeasures. Japan has an in-depth accident investigation database:​ Institute for Traffic Accident Research and Data Analysis (ITARDA) Micro Data;​ however, the number of investigated accidents is limited. On the contrary, the National Automotive Sampling System/Crashworthiness Data System (NASS/CDS) in the U.S. is the largest and many researchers utilize it. In this study, research questions are clarified by studying similarities and differences between Japan and the U.S. distribution of factors influencing occupant injuries in frontal collisions using these databases, and presenting new utilization of the NASS/CDS for accident analysis.

Accident types compared are head-on collisions and frontal single-vehicle collisions occurred in 2000-2009. Appropriate eleven variables on occupant injuries in frontal collisions are selected, and Mahalanobis distance (MD) of discriminant analysis shows the similarities for each accident database. In this analysis, the variables are classified into four groups:​ injury prediction factors, vehicles factors, occupant factors, and injury outcome factors;​ furthermore, the multidimensional distribution is compared using these groups.

In results, the NASS/CDS data has similar distribution of MD to the ITARDA Micro Data in the injury prediction factors such as delta-V, seat belt use, multiple impact, occupant age etc.;​ however, conditions which include vehicle factor such as vehicle curb weight, and occupant factors such as occupant height have different distribution. Therefore, the NASS/CDS is useful for statistical analysis such as injury prediction for Japan;​ however, it cannot be utilized as it is for small vehicles crashes, or short statured occupants because of the differences in vehicle factors and occupant factors. It is necessary to consider these differences in case of using these factors. Furthermore, new weighting method for the NASS/CDS using the MD can create closer weighted database to traffic accidents in Japan. This method needs further improvements;​ however, it is useful in Japan for analyses using the NASS/CDS.