NHTSA Report Number DOT HS 809 004 | October 1999 | Evaluation of FMVSS 214 - Side Impact Protection: Dynamic Performance Requirement Phase 1: Correlation of TTI(d) with Fatality Risk in Actual Side Impact Collisions of Model Year 1981-1993 Passenger Cars Plan for Phase 2: Effect of FMVSS 214 and Correlation of TTI(d) with Actual Fatality Risk in Model Year 1992-2000 Passenger Cars Charles J. Kahane, Ph.D. Abstract Federal Motor Vehicle Safety Standard (FMVSS) 214, "Side Impact Protection" was amended in 1990 to assure occupant protection in a dynamic test that simulates a severe right-angle collision. It was phased into new passenger cars during model years 1994-97. The test involves a Moving Deformable Barrier hitting the side of a vehicle. Side Impact Dummies are seated adjacent to the impact point. A Thoracic Trauma Index, TTI(d) is measured on the dummies. The standard will be evaluated in two phases. Phase 1, completed in this report, is a statistical analysis of relationships between front-seat TTI(d) and fatality risk in actual side impacts on the highway, in baseline, pre-standard vehicles of model years 1981-93, based on Fatality Analysis Reporting System (FARS) data from late 1980 through early 1998. The report also presents a plan for Phase 2, a proposed statistical comparison of side-impact fatality and injury rates in cars produced immediately after vs. immediately before the implementation of FMVSS 214. The principal finding of Phase 1 is a statistically significant association of TTI(d) with side-impact fatality risk in passenger cars of model years 1981-93. The observed relationship is stronger, however, in 2-door cars than in 4-door cars. Each reduction of TTI(d) by one unit is associated with an estimated 0.927 percent reduction of fatality risk in side impacts of 2-door cars. The association between TTI(d) and fatality risk in the corresponding analysis of baseline 4-door cars was not statistically significant. Executive Summary Federal Motor Vehicle Safety Standard (FMVSS) 214, "Side Impact Protection" was amended in 1990 to assure occupant protection in a dynamic test that simulates a severe right-angle collision. It is one of the most important and promising safety regulations issued by the National Highway Traffic Safety Administration (NHTSA). It was phased into new passenger cars during model years 1994-97. In 1993, side impacts accounted for 33 percent of the fatalities to passenger car occupants. The current FMVSS 214 is the culmination of many years of research to make passenger cars less vulnerable in side impacts, and especially to reduce fatality risk to the nearside occupant when a car is struck in the door area by another vehicle - the configuration responsible for the majority of side-impact fatalities. Interacting with the United States and international safety communities, NHTSA developed: - A test configuration using a Moving Deformable Barrier (MDB) simulating a severe intersection collision between two passenger vehicles.
- Injury criteria, above all a Thoracic Trauma Index (TTI) that predicts the severity of thoracic injuries when occupant's torsos contact the interior side surface of a car.
- A Side Impact Dummy (SID) on which TTI could be reliably measured in side impact tests. The injury score measured on the dummy is called TTI(d).
- Two technologies that, singly or in combination, significantly reduced TTI(d) from its baseline levels in model year 1980-88 production vehicles:
- Structure modifications such as stronger pillars, sills, roof rails, seats or cross-members, to reduce door intrusion into the passenger compartment.
- Padding capable of absorbing significant energy at a force-deflection rate safe for occupants. It is a thick plastic foam - not a soft pad.
- The new FMVSS 214, allowing TTI(d) up to 90 in 2-door cars and 85 in 4-door cars.
The Government Performance and Results Act of 1993 and Executive Order 12866 require agencies to evaluate their existing programs and regulations. The objectives of an evaluation are to determine the actual benefits - lives saved, injuries prevented, damages avoided - and costs of safety equipment installed in production vehicles in connection with a rule. FMVSS 214 will be evaluated in two phases. Phase 1, contained in this report, is a statistical analysis of relationships between TTI(d) and fatality risk in actual side impacts on the highway, in baseline, pre-FMVSS 214 cars of model years 1981-93. It is based on Fatality Analysis Reporting System (FARS) data from late 1980 through early 1998. It will tell us if the cars with lower TTI(d) had lower fatality risk. The analysis is possible because many pre-standard cars were tested during the development of FMVSS 214, and those models have been on the road for a long time and have been involved in many crashes. However, Phase 1, based on pre-standard cars, will not estimate the benefits of FMVSS 214 itself. Phase 2 is a statistical comparison of side-impact fatality and injury rates in cars produced immediately after vs. immediately before the implementation of FMVSS 214. It is designed to measure the actual effects of specific modifications used to achieve compliance with FMVSS 214: structures and padding. Since those changes were only introduced in 1994-97, and since it takes years for crash files to accumulate sufficient data for statistical analyses, Phase 2 is unlikely to be completed before 2001. This report presents an analysis plan for public review and comment. Side air bags have begun to supplement structure and padding in some cars. As of 1999 it does not appear that sufficient crash cases involving side air bags can accumulate within the Phase 2 time frame for meaningful statistical analyses. NHTSA plans to start an evaluation in 2002, but that date could be expedited in response to higher sales volumes or other considerations. In addition, NHTSA's New Car Assessment Program (NCAP) has provided consumers with information since 1997 about side impact performance in a test 5 mph faster than the FMVSS 214 compliance test. By 2002 there may be enough data to study the correlation of Side-NCAP results with fatality risk in real-world side impacts. In most of the Phase 1 analyses, the "side-impact fatality risk" of a make-model is the ratio of the occupant fatalities in side-impact crashes to "purely frontal" crashes (a control group). This definition and various other statistical tools help isolate genuine crashworthiness differences between make-models and minimize possible biases due to models having different types of drivers whose crash involvement rates can vary considerably. The primary finding of Phase 1 is a statistically significant association of TTI(d) with side-impact fatality risk: the lower the TTI(d), the lower the fatality risk. This result is obtained when all baseline-tested make-models of 1981-93 passenger cars are analyzed together. A closer look at the data, however, immediately reveals different relationships in 2-door cars and 4-door cars. The association between TTI(d) and fatality risk in actual side impacts is quite strong in 2-door cars, and the more closely the actual crashes resemble the FMVSS 214 test, the stronger the relationship. But the data show at most a weak relationship between TTI(d) and actual fatality risk in 4-door cars. The difference is unexplained at this time. Data/statistical problems could be masking some of the effect in 4-door cars. And even if the true effect is indeed much stronger in older 2-door cars than 4-door cars, there is no obvious reason why that should be so. A possible factor is that there were stark differences in TTI(d) among pre-standard 2-door cars, including some very poor performers - whereas most 4-door cars, even before FMVSS 214, had fairly similar, fairly adequate performance. In summary, does lower TTI(d) mean lower fatality risk? These analyses generally say "yes," but leave some unanswered questions even about the baseline, pre-FMVSS 214 passenger cars of Phase 1. Needless to say, they should not be used to predict exactly what effectiveness will be found for FMVSS 214 in more recent passenger cars during Phase 2. Nevertheless, they show that FMVSS 214 has, at the very least, already saved lives by mandating redesign of the 2-door models with the poorest performance. The main findings and conclusions of Phase 1 are the following: SIDE IMPACT PERFORMANCE OVER THE YEARS - Average TTI(d) in the FMVSS 214 test configuration, by model year, was approximately:
| 2-Door Cars | 4-Door Cars | FMVSS 214 requirement | 90 | 85 | Model Year | | | 1981-90 (baseline) | 110 | 80 | 1993 (just before FMVSS 214) | 97 | 74 | 1997 (post-FMVSS 214) | 74 | 65 | - In 2-door cars, average performance improved from much worse than the FMVSS 214 requirement to somewhat better.
- In 4-door cars, average performance improved from slightly better than the FMVSS 214 requirement to much better.
- In make-models produced throughout 1993-97, approximately
- 56 percent got substantial structure, usually with padding, during 1994-97
- 27 percent got mainly padding
- 17 percent remained unchanged, and already met FMVSS 214 in 1993
TTI(d) AND SIDE-IMPACT FATALITY RISK: 2-DOOR CARS OF MODEL YEARS 1981-93 - There is a statistically significant association between TTI(d) and side-impact fatality risk in pre-standard 2-door cars. This association was found in regression analyses, correlation analyses, and matched comparisons of make-models with low and high TTI(d).
- In the regression analyses, each reduction of TTI(d) by one unit is associated with an estimated 0.927 percent reduction of fatality risk in side impacts.
- TTI(d) averaged 110 in these cars, and the best score for any model was 82. A reduction of TTI(d) from the "average" 110 to the "best practices" 82 corresponds to a 23 percent1 fatality reduction in side impacts.
- The effect of TTI(d) was strongest in the crashes that most closely resembled the FMVSS 214 test configuration (all effects are statistically significant):
| Fatality Reduction (%) for Reducing TTI(D) | | By One Unit | From 110 to 82 | In all side impacts | 0.927 | 231 | In occupant compartment impacts | 1.080 | 26 | For all nearside occupants | 0.999 | 25 | Nearside occupants in compartment impacts | 1.280 | 30 | Impacts by another passenger car | 1.310 | 31 | Nearside compartment impacts by a pasenger car | 1.730 | 39 | - Correlation analyses, and matched comparisons of make-models with low and high TTI(d) produced nearly the same results as the regression analyses: significantly lower fatality risk in models with low TTI(d).
TTI(d) AND SIDE-IMPACT FATALITY RISK: 4-DOOR CARS OF MODEL YEARS 1981-93 - Regression, correlation and matched-comparison analyses of pre-standard 4-door cars showed at most a weak association between TTI(d) and side-impact fatality risk.
- In one regression analysis, each reduction of TTI(d) by one unit was associated with a nonsignificant 0.168 percent reduction of fatality risk in side impacts, and in another regression analysis, it was associated with a nonsignificant 0.047 percent increase.
- TTI(d) averaged 80 in baseline 4-door cars, and the best score for any small- to mid-sized model was 62. A reduction of TTI(d) from the "average" 80 to the "best practices" 62 corresponds to a 3 percent2 fatality reduction in side impacts, by the first regression analysis, and a 1 percent3 increase, by the second.
- In the regression analyses, the effect of TTI(d) did not show any pattern of getting either stronger or weaker in crashes that more closely resembled the FMVSS 214 test configuration.
- Twelve analyses tested the correlation of TTI(d) with side/frontal fatality risk at the make-model level. They produced one statistically significant positive coefficient [the lower the TTI(d) the lower the risk], four nonsignificant positive coefficients, and seven nonsignificant negative coefficients.
- Twelve analyses compared side/frontal fatality risk in matching make-models with low and high TTI(d). Eight showed fatality reductions up to 16 percent in the models with low TTI(d). Four showed increases. None of the effects was statistically significant.
1. 1 - (1 - .00927)110 - 82 = 23 percent 2. 1 - (1 - .00168)80 - 62 = 3 percent 3. 1 - (1 + .00047)80 - 62 = -1 percent |