At the Annual Meeting of the Transportation Research Board in January, 1994, FHWA researcher John G. Viner presented a paper on rollovers and their relationship to slide slopes and ditches.

Viner discussed research that used accident files from the Illinois Highway Safety Information System (HSIS). Motorcycles were excluded from the data files used for the years 1985-1989. The number of accidents used for all other vehicles were 115,858 with 16,453 rollovers and 99,405 non-rollovers. Among the findings were the following:

• Rural road rollovers in Illinois accounted for 47.8% of ran- off-road (ROR) fatalities.
• Rural ROR crashes are 3 « times as likely to result in a rollover as on the urban road.
• Rural interstate and rural 2-lane roads have comparable rollover rates (28.9% and 28%).
• Urban interstates have higher rollover rates than other urban roads (9.1% vs 6.5%).
• Two-thirds of the Illinois ROR rollovers occurred on rural roads.
• Urban fixed object impacts contributed to 3.0% of the rollovers. Several devices contributing to the 3% total were curb/channelizing islands (8.4%), median fence (7.3%), and concrete median barriers (5.1%).
• Rural fixed object impacts contributed to 8.1% of the rural rollovers with culvert headwalls posing the greatest risk of any fixed object in the study.
• Vans, straight trucks, and pickups have the highest rollover rate, while large cars have the lowest rate.
• Rollover crashes on interstate medians are more likely than for ROR right side departures (35.5% vs 29.9%).

It was concluded from the analysis of the data that the common cause of overturn was from tire-soil forces together with ditches/embankment accounting for 82.8% of the rural rollovers and 73% of urban overturns. Also, Viner concluded that rollovers are the dominant rural roadside safety problem, accounting for 48% of the fatalities in ROR crashes. The remaining 52% of the fatalities were attributed to a long list of objects struck on the roadside.

Among Viner's recommendations was the need for studying specific slope and ditch configuration as potential tripping mechanisms and to develop specific cost-effective criteria to reduce rollover risk on slopes and ditches. He points out that field data will have to be collected as current databases do not contain such data. Viner believes certain criteria should be established for needed studies:

1. Identification of areas of roadways such as the outside of horizontal curves that might justify special attention;
2. Revised severity indices for slope and ditch configurations resulting in changes in barrier warrants (need, design, location);
3. Recommendations to maintenance personnel in maintaining relatively flat roadsides; and
4. Defining importance of countermeasures that would reduce the likelihood of loss of control such as anti-lock brakes or higher pavement surface friction.

Viner's research concludes that the criteria for guardrail need and location based on slope steepness must be reevaluated. There must be a better understanding of the roadside and vehicle design factors that contribute to rollover accidents.

Another obvious conclusion that can be drawn from Viner's research is that accident databases simply do not contain the information that is required to do the detailed analysis of causative factors in rollover accidents. This is just another example of a major area of highway fatalities that cannot be properly evaluated, and effective, reliable improvements proposed, without improvements in the current database and accident collection methods.