Severe injury in multiple impacts: Analysis of 1997–2015 NASS-CDS

ABSTRACT

Purpose: This is a descriptive study of the incidence and risk for severe injury in single-impact and multi-impact crashes by belt use and crash type using NASS-CDS.

Methods: 1997–2015 NASS-CDS data were used to determine the distribution of crashes by the number of impacts and severe injury (Maximum Abbreviated Injury Score [MAIS] 4+F) to >15-year-old nonejected drivers by seat belt use in 1997+ MY vehicles. It compares the risk for severe injury in a single impact and in crashes involving 2, 3, or 4+ impacts in the collision with a focus on a frontal crash followed by other impacts.

Results: Most vehicle crashes involve a single impact (75.4% of 44,889,518 vehicles), followed by 2-impact crashes (19.6%), 3-impact crashes (5.0%) and 4+ impacts (2.6%). For lap–shoulder-belted drivers, the distribution of severe injury was 42.1% in a single impact, 29.3% in 2 impacts, 13.4% in 3 impacts, and 15.1% in 4+ impact crashes. The risk for a belted driver was 0.256 ± 0.031% in a single impact, 0.564 ± 0.079% in 2 impacts, 0.880 ± 0.125% in 3 impacts, and 2.121 ± 0.646% in 4+ impact. The increase in risk from a single crash to multi-impact collisions was statistically significant (P < .001).

In a single impact, 53.8% of belted drivers were in a frontal crashes, 22.4% in side crashes, 20% in rear crashes, and 1.7% in rollover crashes. The risk for severe injury was highest in a rollover at 0.677 ± 0.250%, followed by near-side impact at 0.467 ± 0.084% and far-side impact at 0.237 ± 0.071%. Seat belt use was 82.4% effective in preventing severe injury (MAIS 4+F) in a rollover, 47.9% in a near-side impact, and 74.8% in a far-side impact.

In 2-impact crashes with a belted driver, the most common sequence was a rear impact followed by a frontal crash at 1,843,506 (21.5%) with a risk for severe injury of 0.100 ± 0.058%. The second most common was a frontal impact followed by another frontal crash at 1,257,264 (14.7%) with a risk of 0.401 ± 0.057%. The risk was 0.658 ± 0.271% in a frontal impact followed by a rear impact. A near-side impact followed by a rear crash had the highest risk for severe injury at 2.073 ± 1.322%.

Conclusions: Restraint systems are generally developed for a single crash or sled test. The risk for severe injury was significantly higher in 2-, 3-, and 4+-impact crashes than a single impact. The majority (57.9%) of severe injuries occurred in multi-impact crashes with belted drivers. The evaluation of restraint performance warrants additional study in multi-impact crashes.

KEYWORDS:

• Single impacts
• multiple impacts
• injury risk
• frontal crashes

Introduction

Fay et al. (2001), Digges and Bahouth (2003), and Bahouth and Digges (2005) assessed the distribution of crashes and injury risks in single- and multi-impact collisions. Fay et al. (2001) compared the frequency of single and multiple impacts by injury severity using German and UK data. They reported that multiple impacts accounted for about 30% of all crashes. About 60% of multiple impacts involve 2 impacts in the collision.

Digges and Bahouth (2003) and Bahouth and Digges (2005) found that injury risks were higher in multi-impact crashes than a single impact. For example, Digges and Bahouth (2003) determined the risk for serious injury (Maximum Abbreviated Injury Scale [MAIS] 3+) by number of impacts in the crash. Table A1 (see online supplement) provides the data. The risk for serious injury was 1.9 times higher in crashes with 2 impacts than single frontal crashes (3.21 versus 1.65%) and 3.8 times higher in 3+-impact crashes than in single frontal impacts (6.25 versus 1.65%). The risk was 7.6 times higher in 2-impact crashes than in a single rear crash (3.21 versus 0.42%) and 14.9 times higher in 3+ impacts than single rear impacts (6.25 versus 0.42%).

Bahouth and Digges (2003) also evaluated the risk of serious injury by belt use. Table A2 (see online supplement) summarizes the results. The risk for serious injury was 2.2 times higher in 2-impact crashes than single frontal crashes for belted occupants (2.08 versus 0.93%) and 3.5 times higher in 3+ impacts than single frontal impacts for belted occupants (3.52 versus 0.93%). The risk was 6.1 times higher in 2 impacts than single rear impacts with belt use (2.08 versus 0.34%) and 10.4 times higher in 3+ impacts than single rear crashes with belt use (3.52 versus 0.34%). Seat belt use was effective in reducing injury risks in multi-impact crashes. It was 74% effective in preventing serious injury (MAIS 3+) in 2-impact crashes and 79% effective in 3+ impacts.

Temming and Zobel (1998), Fay et al. (2001), and Bahouth and Digges (2005) found an increase in serious injury risk in multiple-impact collisions. Temming and Zobel (1998) identified multiple impacts as second only to frontal impacts causing injury to belted occupants. Viano and Parenteau (2010) studied individual electronic cases from NASS-CDS with serious head injury in frontal crashes with <24 km/h (<15 mph) delta V. They identified impacts on the A-pillar, side roof rail, B-pillar, windshield, steering wheel, and airbag. Many of the head impacts occurred with a belted occupant and airbag deployment. The crashes often involved multiple impacts in the collision sequence that started at a high vehicle speed and frontal impact followed by other collisions causing injury. They suggested that safety systems consider prolonged collision events that may require sustained inflation of front airbags or backup chambers that sustain pressure for 5 or more seconds to address front interior contacts in multiple impacts.

Restraint systems are generally developed and assessed in a single-impact crash or sled test. Fay et al. (2003) concluded that though analysis of accident data and crash tests focused on a single impact, multiple impacts accounted for a large proportion of serious injuries. They pointed out that safety countermeasures are typically developed for a single impact and that multiple impacts should be considered because their characteristics may have implications for occupant protection. Adams et al. (1990) described how multiple impacts were considered during development of airbag sensing systems. They found that 32% of collisions involved multiple impacts.

The development of safety systems for multi-impact collisions requires information on the frequency of multi-impact crashes, collision sequences, and injury risks. There is currently limited information in the literature. The current study updates Bahouth and Digges (2003) using more recent NASS-CDS data. It provides additional information on the risk for severe injury to belted drivers when the first impact was a frontal crash followed by other impacts in the collision sequence.

Methods

NASS-CDS

NASS-CDS (www.nhtsa.dot.gov) is a stratified, multiphase, unequal selection probability sample of motor vehicle crashes that are prospectively selected for in-depth investigation. Most of the vehicles were towed from the scene because of damage. The data include information from crash investigation teams that gather information from the crash site, vehicle, medical records, police accident reports, and personal interviews. NASS-CDS data for calendar years 1997–2015 was evaluated for light vehicles with 1997+ MY (model year) with 15+-year-old nonejected drivers. The analysis was conducted by crash type, number of impacts, injury severity, and seat belt use.

Number of impacts

Number of impacts was assessed by determining the number of impacts associated per vehicle using variable ACCSEQ in the event file.

Crash types

Impacts were defined with GADEV1 and GADEV2 variables:

•	Frontal impact (F) and no rollover (rollover ≤ 0).
•	Side impact (L or R) and no rollover (rollover ≤ 0).
•	Rear impact (B) and no rollover (rollover ≤ 0).
•	Rollover: >one-quarter turn (rollover > 0).

Crash severity

Crash severity was assessed with the DVTOTAL variable using the following categories:

•	<24 km/h (0 ≤ DVTOTAL < 24).
•	24–48 km/h (24 ≤ DVTOTAL < 48).
•	48 + km/h (48 ≤ DVTOTAL).
•	Unknown/missing (DVTOTAL = other).

The unknown or missing crash severity data were assessed using the DVEST variable.

Belt use

Belt use was defined by the NASS-CDS investigator's variables MANUSE (manual belt use) and ABELTUSE (automatic belt use). Unbelted was defined as MANUSE ≤ 1 and ABELTUSE < 1 or = 2 for calendar years less than 2010 and as MANUSE = 0 or 1 for calendar years greater than 2009. Belted was defined as MANUSE = 4.

Maximum injury severity

Maximum injury severity was assessed using the Abbreviated Injury Scale (AIS) and body region (region90). Injuries were classified using AIS 98 coding (Association for the Advancement of Automotive Medicine 1998). The MAIS “TREATMNT” and “INJSEV” variables were analyzed. MAIS represents maximum injury severity. It is the assessment of injury at the time of first medical evaluation and not long-term consequences. It ranges from MAIS 0 to 6 and 9, where MAIS 9 is an injury with unknown severity. Fatality was recorded if the occupant died within 30 days of the crash. Fatality (F) was defined by the following:

•	TREATMNT = 1, which means that the occupant was fatally injured and not transported to the hospital.
•	INJSEV = 4 represents a fatality according to police rating.

Exposed occupants were defined as those with known MAIS (MAIS 0–6) or with a fatality. The shorthand notation is MAIS 0+F. Severely injured occupants were defined as those with MAIS 4–6 or fatality, because fatalities can occur at any MAIS level. The shorthand notation is MAIS 4+F.

Analysis

The risk that an occupant experienced^{Table 2} a severe injury was determined by dividing the number of occupants with severe injury (MAIS 4+F) by the number of occupants with known injury status, MAIS 0–6 or F (MAIS 0+F). Risks are provided plus or minus one standard error. The^{Table 1}significance of differences in risk was determined using the Rao-Scott chi-square test in SAS with PROC SURVEYFREQ. P < .10 was significant. Belt effectiveness was assessed by subtracting the severe injury risk for belted occupants from the risk for unbelted occupants and dividing by the risk for unbelted occupants.

Weighted data

National estimates for the number of crashes and occupant injuries were made using the ratio weight (RATWGT) variable in the NASS-CDS. All calculations were based on weighted values. Cases with a RATWGT equal to 0 or with a negative RATWGT were excluded from the analysis.

The analysis presented in this article is based on weighted data. The unweighted data are provided in the Appendices (see online supplement). Some numbers in the tables are highlighted in grey. The highlight was used to identify weighted numbers that were based on an unweighted sample ≤ 10. Though these samples are small and there should be caution, the numbers provided represent the best national estimates from the NASS-CDS.

Results

Table 1 shows the number of vehicles involved in single impacts and multiple impacts. It provides the number of >15-year-old drivers who were not ejected by the number of impacts and seat belt use. There were 44,889,519 vehicles with MY 1997+ involved in the 19 years of NASS-CDS data. More than 75% (75.4%) were involved in a single impact, 19.6% in 2 impacts, 5.0% in 3 impacts, and 2.6% in 4+ impacts. There were 25,049,962 drivers involved with known injury (MAIS 0+F); 79.6% were lap–shoulder belted, 15.7% were unbelted, and 4.7% had unknown or other restraint type.

Table 1. Nonejected driver distribution and injury risks by belt use and number of impacts using 1997–2015 NASS-CDS (MY 1997+ vehicles, age 15–104).

	Number of impacts
	Single	2	3	4+	Total
Vehicles	32,998,791	8,566,067	2,195,676	1,128,984	44,889,518
	MAIS 0+F
All	17,268,455	5,375,856	1,627,028	778,624	25,049,962
Belted	13,720,725	4,339,101	1,274,456	594,630	19,928,913
Unbelted	2,605,916	874,768	285,159	167,718	3,933,561
	MAIS 4+F
All	62,364	53,954	27,063	23,793	167,174
Belted	35,153	24,473	11,209	12,615	83,449
Unbelted	25,127	28,028	14,951	10,309	78,415
	MAIS 4+F risk
All	0.361%	1.004%	1.663%	3.056%	0.667%
SE	0.064%	0.271%	0.389%	0.778%	0.145%
Belted	0.256%	0.564%	0.880%	2.121%	0.419%
SE	0.031%	0.079%	0.125%	0.646%	0.059%
Unbelted	0.964%	3.204%	5.243%	6.147%	1.993%
SE	0.265%	1.316%	1.776%	1.559%	0.621%
Belt effectiveness	73.4%	82.4%	83.2%	65.5%	79.0%

There were 83,449 belted drivers with severe injury (MAIS 4+F); 42.1% were in a single impact, 29.3% were in 2 impacts, 13.3% were in 3 impacts, and 15.1% were in 4+ impacts. The majority (57.9%) of severe injuries occurred in multi-impact crashes. Overall, the risk for a belted driver was 0.256 ± 0.031% in a single impact, 0.564 ± 0.079% in 2 impacts, 0.880 ± 0.125% in 3 impacts, and 2.121 ± 0.646% in 4+ impact crashes. The increase in risk with multi-impacts was statistically significant (P < .001). The risk in all multi-impact crashes was 0.778% for belted drivers. It was significantly higher than in a single impact with a belted driver (P < .001). There were 78,415 unbelted drivers. The injury risks were higher than for those who were lap–shoulder belted. The risk was highest for unbelted in 4+-impact crashes at 6.147 ± 1.559%. The belt effectiveness in preventing MAIS 4+F was 73.4% in a single impact and 82.4% in 2-impact crashes. The risk in all multi-impact crashes was 4.014% for unbelted drivers. It was significantly higher than for a single impact with an unbelted driver (P < .001).

Table 2 shows the data for single-impact crashes. There were 32,998,791 vehicles, with 53.8% involved in a frontal impact, 22. 4% in a side impact, and 20% in a rear impact. There were 35,153 lap–shoulder-belted drivers with severe injury (MAIS 4+F); 62.7% were in a frontal impact, 24.4% were in a side impact, and 7.4% were in a rear impact. The risk for severe injury to belted drivers was highest in a rollover at 0.677 ± 0.250%, followed by near-side impact at 0.467 ± 0.084% and far-side impact at 0.237 ± 0.071%. The risk in a frontal impact was significantly lower in a left-side or rollover crash and higher in a rear impact (P < .001). Belt use was 82.4% effective in a rollover, 47.9% in a near-side impact, and 74.8% in a far-side impact.

Table 2. Single impact distribution and nonejected driver injury risks by crash type and belt use using NASS-CDS 1997–2015 (MY 1997+ vehicles, age 15–104).^a

	Impact type in single-event crashes
Single impact	Front	Left (near)	Right (far)	Rear	Rollover	Other/unknown	Total
Vehicles	17,738,325	4,071,507	3347872	6,585,710	546,055	709,322	32,998,791
	MAIS 0+F
All	11,011,178	2,346,339	1,958,022	1,218,553	392,106	342,257	17,268,455
Belted	8,606,905	1,884,556	1,607,027	1,020,837	345,154	256,246	13,720,725
Unbelted	1,772,687	331,111	262,100	137,671	45,409	56,938	2,605,916
	MAIS 4+F
All	37,009	12,119	6,520	1,142	4,078	1,496	62,364
Belted	18,426	8,800	3,802	1,012	2,336	777	35,153
Unbelted	17,279	2,968	2,457	81	1,742	600	25,127
	Risk MAIS 4+F
All	0.336%	0.517%	0.333%	0.094%	1.040%	0.437%	0.361%
SE	0.057%	0.091%	0.087%	0.020%	0.480%	0.127%	0.064%
Belted	0.214%	0.467%	0.237%	0.099%	0.677%	0.303%	0.256%
SE	0.018%	0.084%	0.071%	0.024%	0.250%	0.114%	0.031%
Unbelted	0.975%	0.896%	0.937%	0.058%	3.836%	1.054%	0.964%
SE	0.272%	0.206%	0.273%	0.023%	2.733%	0.230%	0.265%
Belt effectiveness	78.0%	47.9%	74.8%	−69.5%	82.4%	71.2%	73.4%

^a Highlighted cells have unwgt samples < 10.

Table 3 shows the vehicle data for 2-impact crashes, which involved 8,566,067 vehicles with a lap–shoulder-belted driver. The data are presented for various combinations of impacts. For example, the most common 2-impact crash was rear impact, followed by a frontal crash at 1,843,506 (21.5%). The second most common was a frontal impact followed by another frontal crash at 1,257,264 (14.7%). The risk of severe injury to a belted driver was 0.100 ± 0.058% in a rear crash followed by a frontal crash. It was 0.980 ± 0.802% in a rear impact followed by another rear impact. The risk was 0.401 ± 0.057% in a front impact followed by another frontal crash. It was 0.658 ± 0.271% in a front impact followed by a rear impact. A near-side impact followed by a rear crash had the highest risk for severe injury at 2.073 ± 1.322%.

Table 3. Two-impact crash distribution and injury risks by crash type for belted, nonejected drivers using NASS-CDS 1997–2015 (MY 1997+ vehicles, age 15–104).^a

		Front				Left (near)				Right (far)
First impact
Second impact	Front	Left (near)	Right (far)	Rear	Front	Left (near)	Right (far)	Rear	Front	Left (near)	Right (far)	Rear
Vehicles	1,257,264	695,866	645,941	300,027	264,265	538,385	179,559	46,567	337,850	241,647	446,615	85,187
MAIS 0+F	618,020	423,235	370,264	131,071	138,404	318,351	108,637	24,670	249,797	173,623	14,328	149,568
MAIS 4+F	2,478	1,353	1,301	863	1,779	3,521	1,001	511	1,009	330	225	538
Risk MAIS 4+F	0.401%	0.320%	0.351%	0.658%	1.285%	1.106%	0.921%	2.073%	0.404%	0.190%	1.568%	0.360%
se	0.057%	0.110%	0.060%	0.271%	0.351%	0.440%	0.369%	1.322%	0.102%	0.070%	0.931%	0.147%
	Rear
First impact
Second impact	Front	Left (near)	Right (far)	Rear	Rollover	Total
Vehicles	1,843,506	74,931	51,788	28,915	1,106,665	8,566,067
MAIS 0+F	565,247	21,114	33,490	7,011	700,385	4,339,101
MAIS 4+F	566	154	307	69	5,301	24,473
Risk MAIS 4+F	0.100%	0.731%	0.916%	0.980%	0.757%	0.564%
Second impact	0.058%	0.383%	0.804%	0.802%	0.101%	0.079%

^a Highlighted cells have unwgt samples < 10.

Figure 1 shows the risk of severe injury (MAIS 4+F) with a frontal crash as the first impact followed by other collisions with a lap–shoulder-belted driver. The risk was highest in 3+-impact crashes with a rollover at 1.304 ± 0.360%. The next highest risk was a frontal impact followed by a rear crash at 0.658 ± 0.271%. It was lowest in a single frontal impact at 0.214 ± 0.018%. The risk in a single frontal impact was significantly lower than in frontal impacts followed by another collision (P < .001), except for the front impact followed by a left-side impact (NS). Figure A1 (see online supplement) shows the risk of severe injury (MAIS 4+) for lap–shoulder-belted and nonejected drivers involved in a rear crash as the first impact followed by other collisions. The risk was similar for a single rear impact compared to a rear impact followed by a frontal crash (0.099 ± 0.024% versus 0.100 ± 0.058%, NS). A rear impact followed by any other type of crash had higher risks; however, the sample size was too small to make statistical comparisons.

Figure 1. Risk of severe injury (MAIS 4+) for belted drivers in a frontal crash as the first impact.

Table A3 (see online supplement) shows the delta V for the multi-impact crashes using the DVTOTAL variable. The data were analyzed in 3 categories of severity for the vehicle and belted drivers (MAIS 0+F and MAIS 4+F) who were not ejected. About 50% of the crash severity was unknown. For collisions classified as >48 km/h, the percentage of vehicles involved increased with the number of impacts in the collision from 1.0% for single, 1.4% for 2, 2.5% for 3, and 2.5% for 4+ impacts in the collision. Table A4 (see online supplement) shows the estimated delta V for multi-impact crashes with the DVEST variable. At the vehicle level, 3.1% of single-vehicle crashes were listed as severe. The percentage increased to 6.5% for 2, 8.2% for 3, and 16.5% for 4+ impacts in the collision sequence. There was a corresponding decrease in the percentage of minor impacts. There were not enough data to see trends with MAIS 4+F injury by delta V.

Discussion

In this study, the incidence and risk for severe injury (MAIS 4+F) was determined for single-impact and multi-impact collisions. The scheme used was similar to that of Digges and Bahouth (2003), but the type of crashes and injury severity were different. Digges and Bahouth (2003) reported on serious injury (MAIS 3+). Severe injury was chosen because AIS 3 injuries often include extremity fractures and other non-life-threatening injuries. The risk of death for lap–shoulder-belted drivers was 2.99 ± 0.50% for MAIS 3 injury, but it was 17.4 ± 3.40% for MAIS 4 and 39.6 ± 9.50% for MAIS 5 injuries (Viano and Parenteau 2018). Another reason for selecting MAIS 4+F over MAIS 3+F is that when MAIS 3+F is used, the MAIS 3 injuries tend to overwhelm the data cells. This obscured the population of crashes that are most likely to cause a fatality. On the other hand, using MAIS 3+F produces fewer cells with less than 10 unweighted cases. The development of safety systems focuses on the prevention of life-threatening injuries, which typically excludes MAIS 1–3 injury. The exception is low-speed whiplash testing.

Table A5 (see online supplement) compares the current study results to those of Digges and Bahouth (2003). The risk for severe injury (MAIS 4+F) was 3.0 times higher in 2-impact crashes than in single frontal crashes (1.00 versus 0.336%) and 4.9 times higher in 3-impact crashes and 9.1 times higher in 4+-impact crashes than in single frontal impacts. The risk was 10.6 times higher in 2-impact crashes than in single rear impacts (1.00 versus 0.094%). It was 17.7 times higher in 3-impact crashes and 32.5 times higher in 4+-impact crashes than single rear impacts.

The risk was higher for belted drivers in multi-impact crashes. Table A6 (see online supplement) compares the current study results to those of Digges and Bahouth (2003) by belt use. The risk of severe injury was 2.6 times higher in 2-impact crashes than in single frontal crashes for belted drivers (0.564 versus 0.214%) and 4.1 times higher in 3-impact crashes and 9.9 times higher in 4+-impact crashes than in single frontal impacts. The risk was 5.7 times higher in 2-impact crashes than in single rear impacts with belt use (0.564 versus 0.099%) and 8.9 times higher in 3-impact crashes and 21.4 times higher in 4+-impact crashes than in single rear crashes with belt use. Seat belt use was effective in reducing injury risks in multi-impact crashes. It was 82% effective in preventing severe injury (MAIS 4+) in 2-impact, 83% in 3-impact, and 66% in 4+-impact crashes. Despite belt effectiveness, about half of the MAIS 4+ injuries occurred to belted occupants. About 31.2% of belted occupants were exposed to multiple-event crashes. However, the majority (57.9%) of MAIS 4+ injuries occurred in multi-impact crashes, highlighting the need for further studies.

Restraint systems are generally designed and developed using single-impact crash or sled tests (Bahouth and Digges 2005; Fay et al. 2001; Temming and Zobel 1998). This study found that 75.4% of crashes were a single impact; however, the risk of severe injury increased with the number of impacts. For example, the risk for a belted driver was 0.256 ± 0.031% in a single impact, 0.564 ± 0.079% in 2 impacts, 0.880 ± 0.125% in 3 impacts, and 2.121 ± 0.646% in 4+ impacts. The increase in risk was statistically significant (P < .001). Though 24.6% of crashes involve multiple impacts, the higher risks warrant further study of means to improve occupant protection systems for multiple impacts. There are questions about what to deploy and when in multi-impact crashes. In addition, there are questions about the proper mix of single-action restraints, such as front airbags and pretensioners, and prolonged restraints, such as curtain airbags and electric belt retractors.

Acknowledgment

The authors appreciate the help of Henry Scott, Ford Motor Company, in setting up the SAS routine to determine the multi-impact crashes and injuries in NASS-CDS. His help was invaluable.