During 1988-93, the Fatal Accident Reporting System (FARS) coded 4 percent of all fatal large truck-passenger vehiclecrashes as involving underride or override (a passenger vehicle going underneath a large truck). In contrast, the National Accident Sampling System Crashworhtiness Data System (NASS/CDS) coded 27 percent of a sample of 275 faltal large truck-passenger vehicle crashes as underrides during the same years. Severn percent of these 275 fatal crashes are identified as underrides in FASRS. The discrepancy between FARS and NASS coding becomes more pronounced when underrides involving sides of passenger vehicles or trucks are considered. This is because NASS/CDS did not code underrides involving side impacts, and FARS did. When underrides involving side inpacts were added, the total percentage of underrides in NASS/CDS rose from 27 to 50 percent of fatal truck-car crashes. The most likely explanations for the lower incidence of underride coding in FARS are: a) The greater amounts of information available to NASS/CDS analysts enable more complete indentification of underrides, b) FARS analysts sometimes may not recognize that underride has occured, and c) underride was not a separate FARS variable prior to 1994. Based on NASS/CDS data, an estimated 1,108 fatal underride crashes occured each year during 1988-93 (95 percent confidence interval (CI)= 735,1482). Of these 1,108 underrides, 634 involved the fronts (CI=328,248), 248 involved the rears (CI=137,360), and 226 involved the sides (CI=110,341) of large trucks.
Abstract