Fetal deaths and maternal injury in motor-vehicle crashes using NASS-CDS and CISS field data

Abstract Objective This study determined the risk for fetal death and maternal injury in the same sample of motor-vehicle crashes. The frequency and risk of serious injury (MAIS 3 + F) were also assessed by sex, pregnancy, seating position and crash type. Method The 2008–2015 NASS-CDS and 2017–2020 CISS are representative samples that were analyzed for the risk of fetal death and the risk of maternal injury grouped by MAIS 0–2, MAIS 3 + F and death (F) in 2000+ model year (MY) light vehicles. All electronic cases involving fetal mortality were reviewed for mechanism of injury. Separately, the 2000–2015 NASS-CDS and 2017–2020 CISS data was analyzed for the risk of serious injury for male, female and pregnant female occupants by seating position and crash type in 2000+ MY light vehicles. All calculations are made with weighted data. The significance of differences in risk was determined by the Rao-Scott chi-square test in SAS and z-test for differences in proportions. Results There were 2,467 ± 1,407 fetal deaths in light vehicle crashes from 2008–2015 and 2016–2020 with an annual occurrence of 206/yr. The risk for fetal death was 1.25% ± 0.74% of exposed pregnant females. There were 127 ± 67 deaths of pregnant females, or 11/yr in the same sample. The fatality risk was 0.065% ± 0.035%. The difference in proportions was statistically significant (z = 46.1, p < 0.0001). Fetal deaths occurred 19.4-times more often than deaths of pregnant females. In 82.9% of the crashes with a fetal death, the pregnant female was not seriously injured (MAIS 3 + F). The most common mechanism of fetal death was a minor crash, in 80.1% of the weighted cases based on review of photos of the case vehicle and observing very minor structural damage to the vehicle. The minor crash involved either yaw of the occupant compartment with side loading of the pregnant female or her displacement into the restraint system and side interior in 71.7% of the crashes. A severe crash with intrusion at the seating position of the pregnant female occurred in only 11.5% of cases. It usually caused serious injury to the pregnant female and fetal death. Conclusions Fetal deaths occurred 19.4-times more often than deaths of pregnant females in a 12-year sample of motor-vehicle crashes. The most common mechanism was a minor crash that resulted in a fetal death without serious injury to the pregnant female and involved side or oblique loading of the pregnant female.


Introduction
The protection of pregnant females and fetuses is an important issue of public health and motor-vehicle safety. There have been a number of studies on pregnant females using NASS-CDS field data, since pregnancy and trimester status was coded starting in 1995 (Weiss and Strotmeyer 2002;Manoogian 2015;Hattori et al. 2021;Koh et al. 2021). Weiss and Strotmeyer (2002) were the first to analyze NASS-CDS data on pregnant females for age, driver status, seatbelt use and treatment. There have been other studies on pregnant females in motor-vehicle crashes using state and health data (Hyde et al. 2003;Weiss 2006;Vivian-Taylor et al. 2012;Vladutiu et al. 2013). Amezcua-Prieto et al. (2020) conducted a systematic review and meta-analysis of maternal trauma and pregnancy outcome in motor-vehicle crashes. Maternal death occurred in 0.036% (95% CI 0.025%-1.04%) and fetal death or stillbirth in 0.066% (95% CI 0.38%-1.01%) of females involved in motor-vehicle crashes from various sources of data. Females involved in crashes had increased odds of placental abruption (OR 1.43, 95% CI 1.27-1.63) and maternal death (OR 202.3; 95% CI 110.6-367.0) compared with females not involved in a crash. They concluded that pregnant females involved in motor-vehicle crashes had higher risks of maternal death and complications than those not involved. Kvarnstrand et al. (2008) studied maternal fatalities, fetal and neonatal deaths in Swedish motor-vehicle crashes. They found crashes caused 1.4 maternal fatalities per 100,000 pregnancies and a fetus/neonate mortality rate of least 3.7 per 100,000 pregnancies. The incidence of major injury to the pregnant female was 23/100,000 pregnancies with a crash involvement of 207/100,000 pregnancies. Weiss et al. (2001) studied 240 traumatic fetal deaths and determined 3.7 fetal deaths per 100,000 live births. Motor vehicle crashes were the leading mechanism in 82% of fetal deaths giving 2.3 fetal deaths per 100,000 live births, followed by firearm injuries (6% of cases) and falls (3% of cases). Placental injury was mentioned in 100 cases (42%) and maternal death was noted in 27 cases (11%). Mendez-Figueroa et al. (2013) found the risk for an adverse outcome increased in the second trimester of pregnancy if the pregnant woman was a driver, but this was not the case for pregnant passengers or pedestrians. Vivian-Taylor et al. (2012) found 3.5% of pregnant females were admitted to hospital from a motor-vehicle crashes. 96% of females remained undelivered after a crash and had similar pregnancy outcomes as females not involved in a crash. Friese and Wojciehoski (2005) noted that placental abruption was common and lifethreatening to the fetus in motor vehicle crashes.
Various factors influence the injury to pregnant females and their fetus in a motor vehicle crashes. Klinich et al. (2008) studied the effects of crash characteristics and maternal restraint on fetal outcome in a Michigan study. They found that proper restraint of pregnant occupants resulted in an 84% reduction in adverse fetal outcomes, including pregnancy loss, compared with unrestrained pregnant females. They reported the risk of adverse fetal outcome was 80-90% with improper restraint in crashes of 39-44 km/h (24-27 mph) delta V. The odds ratio was 4.5 with improper belt use compared to proper use. The risk of adverse fetal outcome was 22-38% in 39-44 km/h (24-27 mph) delta V crashes with proper belt use. Klinich et al. (1999) estimated 300 to 3,800 fetal losses per year from maternal involvement in crashes with females over 20 weeks gestation using several methods and databases. They also investigated the causes and circumstances for loss by analysis of real-world crashes involving pregnant occupants. Eight crashes were severe (30þ mph delta V), 11 were moderate (15-30 mph delta V) and 24 were minor (0-15 mph delta V) severity. 69.2% were lap-shoulder belted. Twenty-seven of the crashes were frontal, 10 were side and 6 were rear impacts. They did not include rollovers in the study. Eight crashes resulted in fetal loss, 8 crashes resulted in a placental abruption.
Aitokallio-Tallberg and Halmesmaki (1997) analyzed 35 pregnant females involved in crashes. 15 were in frontal impacts, 13 in side impacts, 5 in multiple impacts and two were cyclists. Twenty-eight (80%) of the pregnant females wore lap-shoulder belts. They found more medical complications for the side impact cases compared to the frontal impact cases. Fetal loss occurred by placental abruption in 5 severe crashes. Manoogian (2015) evaluated 2000-2012 NASS-CDS and identified 324,535 pregnant occupants and 640,804 injuries. The data was evaluated by trimester of pregnancy, restraint use and crash type. 65% of pregnant females were drivers. The average delta V was 18.7 km/h (11.6 mph) for the frontal crashes, which was more than 50% of the crashes. Less than 7% of pregnant females had MAIS 2þ injuries.
The risk for MAIS 2þ injury for pregnant occupants was similar to the risk for non-pregnant occupants based on delta V. Jernigan (2002) evaluated 1995-2000 and found approximately 230,000 pregnant females in motor vehicle crashes. 78% were drivers, 49% were in frontal impacts and 83.6% were belted. The 42.5% of the pregnant females were in the 1 st trimester, 28.4% 2 nd trimester and 29.1% in the 3 rd trimester. Seatbelt use was 60.4% in the 1 st trimester, 92.5% in the 2 nd and 85.4% in the 3 rd trimester.
In this study, risks of fetal death and the risk for injury to pregnant females MAIS 0-2, MAIS 3 þ F and death F were determined in a nationally representative sample of motor vehicle crashes. This type of study was only possible with NHTSA field data from calendar years 2008-2020, because a fetal death variable was added in 2008. All electronic cases with a fetal death were summarized and reviewed for mechanism of injury. A larger sample of crashes from 2000 to 2020 was used to determine the frequency and risk of serious injury (MAIS 3 þ F) by sex, pregnancy, seating position and crash type.

Tow away crashes
The crashes were from the 2000-2015 NASS-CDS (National Automotive Sampling System-Crashworthiness Data System) and 2017-2020 CISS (Crash Investigation Sampling System) available from NHTSA (www.NHTSA.org). The Crashworthiness Data System is a national, statistically sampled vehicle crash database consisting of about 5,000 yearly crashes, with the final year of data available in calendar year (CY) 2015. The Crash Investigation Sampling System replaced NASS-CDS in 2016, with the first publicly available data in CY 2017. The data from both samples were merged according to recommendations from NHTSA to estimate the national incidence of injury of pregnant occupants by seating position and crash type using weighting and sampling factors provided by NHTSA (Zhang et al. 2019). Crashes from the 2008-2015 NASS-CDS and 2017-2020 CISS were analyzed for fetal death and maternal injury. The crash data started reporting fetal death in 2008. All calculations, risks and proportions are reported using weighted data.

Sex and pregnancy
Non-pregnant females were identified with SEX variable, SEX ¼ 2. Pregnant females were identified with SEX ¼3-6 in NASS-CDS/CISS. SEX ¼ 3 represents a pregnant female in the 1 st trimester, 4 in the 2 nd trimester, 5 in the 3 rd trimester and 6, pregnant with unknown trimester. Males were identified with SEX ¼ 1. Non-pregnant females were identified as females in the study.

Injury severity
The Maximum Abbreviated Injury Scale (MAIS) and fatality (F) of the occupant was grouped by MAIS 0-2, MAIS 3-6 plus fatality (F) abbreviated MAIS 3 þ F and fatality (F). Injury coding varied over the calendar years of the field data. During 2000-09, AIS98, based on AIS95, was used. During 2010-15, AIS08, an update of AIS05, was used. During CISS 2017-20, AIS15 was used. The MAIS coded in the calendar year of the crash was used.

Fetus mortality
Death of a fetus was identified using FETALDOA ¼1 in NASS-CDS and FETALMORT ¼1 in CISS. This variable became available in 2008 and indicates a fetal death occurring within 30 days of the crash and as a consequence of the crash.

Injury risk
The risk for serious injury was determined by dividing the number of occupants with serious injury or greater (MAIS 3 þ F) by the number of occupants with known injury status (MAIS 0 þ F). All calculations, proportions and risks were based on the weighted sample. Risks are shown with ± one standard error and 95 th confidence intervals (95 th CI) are reported. The risk represents the average risk determined from the NASS-CDS and CISS occupants, whose injuries were weighted to a nationally representative sample. These counts were used to estimate the frequency or risk for serious injury (MAIS 3 þ F) among all occupants exposed to tow away crashes in the database. Cases with a RATWGT in NASS-CDS and CASEWGT in CISS equal to 0 or negative were excluded from the analysis.

Statistics
SURVEYFREQ procedure from SAS (Release 9.4) was used for the analysis of complex survey data. The multi-years of field data were merged according to NHTSA recommendations. The Taylor series method was used from Zhang et al.
(2019). The statistical significance of the difference in serious injury risks was determined by the SURVEYFREQ procedure using the Rao-Scott chi-square test of association for contingency tables. The analysis accounts for the sampling design and adjusts the observed and expected frequencies.
The difference in proportions was determined by the z-test.

Electronic cases
All available electronical cases were summarized and reviewed for injury mechanisms in a fetus mortality. Photographs of the case vehicle damage and interior showed four different mechanisms of fetal death: I: intrusion at the seating position of the pregnant female, M: minor impact damage to vehicle structures, R: rollovers and D: occupant displacement into the restraints and interior with deformation of the occupant compartment. The coding of mechanism was subjective based on the photographs of damage to vehicle structures and the interior based on the experience of the authors.

Re-Analysis
Appendix B (Supplementary material) shows the results with a modified methodology that excluded vehicles over 10 years old and used a different coded variable for injury. AIS08/ MAIS08 was used for calendar years 2010-2015. AIS15 coding was used in the CISS data for calendar years 2017-2020. The additional data is provided for completeness. Table 1 summarizes the 2008-2015 NASS-CDS and 2017-2020 CISS field data on exposure and injury of pregnant females by MAIS 0-2, MAIS 3 þ F and death (F). There were 195,722 ± 24,410 pregnant females with MAIS 0-2 injury, 1,327 ± 403 with MAIS 3 þ F and 127 ± 67 deaths in 12 years of field data. The annual occurrence was 16,310 ± 2,722 pregnant females with MAIS 0-2 injury and 111 ± 34 with MAIS 3 þ F in 2000þ MY light vehicle crashes. The risk for serious injury was 0.67% ± 0.20% and the risk for death was 0.065% ± 0.035% for pregnant females based on the weighted data. Table 1 shows fetal deaths in the same sample of crashes. There were 2,467 ± 1,407 fetal deaths in 2000þ MY light vehicles since 2008. This gives an annual occurrence of 206 fetal deaths/yr. The risk for fetal death was 1.25% ± 0.74% of exposed pregnant females. In the same time period, there were 127 ± 67 deaths of pregnant females. This gives an annual occurrence of 11 pregnant females deaths/yr with 2000þ MY light vehicle crashes. The difference in proportions was statistically significant (z ¼ 46.1, p < 0.0001). Fetal deaths occurred 19.4-times more often than deaths of pregnant females. All of the fatal pregnant females had a fatal fetus. Table 1 also summarizes the 2000-2015 NASS-CDS and 2017-2020 CISS field data on exposure and serious injury by sex and injury severity groups by MAIS 0-2 and MAIS 3 þ F. Pregnant females were identified using the SEX variable available in all calendar years analyzed. There were 269,878 pregnant females with MAIS 0-2 injury and 2,270 with serious to fatal injury (MAIS 3 þ F) in 20 years of field data. The annual occurrence was 13,494 pregnant females with MAIS 0-2 injury and 114 with MAIS 3 þ F. 1.37% of females were pregnant in crashes. 0.54% of females with serious injury (MAIS 3 þ F) were pregnant. The risk for serious injury was 0.83 ± 0.19% for pregnant females. The risk was 60% lower than for non-pregnant females, who had a risk of 2.08 ± 0.14%, although it is noted that there are differences in ages between pregnant and non-pregnant females. The lower risk for pregnant females was statistically significant (Chi-Square ¼18.65, p < 0.001). The risk was similar between males and non-pregnant females. Figure 1 shows the risk for serious injury (MAIS 3 þ F) by seating position. Non-pregnant female drivers had a similar injury risk as male drivers. In the right-front passenger seat, non-pregnant females had a statistically significant higher risk than men (Chi-Square ¼ 23.06, p < 0.001). The risk was lowest for pregnant females in any seating position. The highest risk was for pregnant females in the 2 nd row but the sample size is small and the standard errors are large. Figure 2 shows the risk for serious injury by crash type. The highest risk for pregnant females was in rollover crashes and the lowest risk was in rear impacts. The differences were statistically significant (Chi-Square ¼ 27.16, p < 0.001).

Results
Also, non-pregnant females had a higher risk of serious injury in side impacts than males. Table A1 in the Appendix A (Supplementary material) summarizes the NASS-CDS and CISS cases with fetal death from 2008-2015 and 2017-2020 that make up the unweighted sample. There were 17 crashes investigated by NASS-CDS and CISS. The average age of the pregnant females with fetal death was 32.5 years old based on the weighted data. Their average weight was 76.7 kg (168.7 lb) and height was 159.4 cm (62.8"). The majority of females with a fetal death were in passenger cars (71.9%) based on the weighted count or in minivans (23.1%) with a small number in pickups (0.9%) and SUVs (0.6%). The most common type of collision with fetal death was a side impact (64.2%). Frontal impacts were the next highest at 28.7% followed by rollovers (5.5%) and rear impacts (0.6%).
The fetal deaths occurred with a range of injury severities to the pregnant females. In 82.9% of the crashes in Table A1 the pregnant female was not seriously injured (MAIS 0-2) based on the weighted count. In only 17.1% of the cases, the pregnant female was seriously injured or killed (MAIS 3 þ F) with a fetal death. There was one pregnant female who died with MAIS 2 injury. The pregnant females were lap-shoulder belted 94.1% of the time based on the weighted data. Only 5.9% were unbelted and their movement in the interior with impacts was related to the fetal death. The fetal death occurred in the 2 nd trimester 64.6% of the time. This was followed by the 3 rd trimester (19.7%) and 1 st (15.7%).   A review of each electronic case and photos indicated several mechanisms of fetal death. The most common mechanism was a minor crash in 80.1% of the weighted cases. The minor crash either involved yaw of the occupant compartment and side interior impact on the abdomen of the pregnant females or displacement of the occupant into the restraint system or interior without deformation of the occupant compartment. In 11.5% of the crashes, the mechanism was a very severe impact with intrusion at the seating position of the pregnant woman and direct loading on the occupant. Rollovers (5.5%) formed another group because of the complex dynamics of the vehicle and occupant contacts in the interior. In 3.0% of the crashes, the occupant moved into the restraint system or interior in a severe crash with essentially no intrusion at the seating position. Figure A1 in the Appendix A summarizes photos of the five electronic cases of fetal death in a minor impact (cases #2, 4, 6, 7, and 13). In all of the cases, there was essentially no deformation of the occupant compartment and minimal or no deformation of vehicle structures. Lateral or oblique loading of the side interior on the torso of the pregnant woman occurred in 71.7% of the crashes (4 of cases #2, 4, 7 and 13). The PDOF of the collision or yaw motion of the occupant compartment caused lateral loads resulting in fetal death with only minor or no injury to the pregnant female.

Discussion
The 2008-15 NASS-CDS and 2017-2020 CISS field data provides a nationally representative sample of motor-vehicle crashes where fetal death and injury to the pregnant females were reported. The combination of these variables only started in 2008 field data. Earlier NASS-CDS studies on injury to pregnant females did not include information on fetal death, so the relative risk to the fetus and pregnant female could not be determined. The 17 electronic cases of fetal death have a weighted incidence of 2,467 ± 1,407 and provide insights on mechanisms of injury based on a nationally representative sample using weighting factors from NHTSA. The additional focuses include 1) fetal deaths occurring 19.4-times more often than deaths of pregnant females significant (z ¼ 46.1, p < 0.0001), 2) fetal death occurring with the pregnant female not seriously injured in 82.9% of the crashes, 3) the primary mechanism of fetal death was a minor crash in 80.1% of the cases and 4) the pregnant females used lap-shoulder belts in 94.1% of the crashes.
The finding that fetal deaths occur 19.4-times more often than deaths of pregnant females is based on injury in the same sample of motor-vehicle crashes with recorded fetal death and injury to the pregnant female. Most other studies have inferred or not determined relative risks for fetal death compared to the pregnant female. There is caution about the specific value of relative risk because the death of pregnant females is based on five unweighted cases and has a large standard error. The 19.4 relative risk represents the best estimate of national incidence based on the NHTSA sampling procedures and weighting factors. Manoogian (2015) evaluated 2000-2012 NASS-CDS field data and reported pregnant females had over 50% frontal crashes with an average 18.7 km/h (11.6 mph) delta V. This study found 23.1% of fetal deaths occurred in frontal crashes, indicating a different mechanism for fetal deaths than maternal crash exposure. Delta V was not evaluated in this study. Klinich et al. (1999Klinich et al. ( , 2008 noted that 56% (24 of 43 crashes) of special investigation of pregnant females involved minor impacts (0-15 mph delta V). They did not have weighting factors to estimate national incidence as is possible with NASS-CDS and CISS field data. There was no special attention to minor crashes. This study gives importance to minor severity crashes where the pregnant female is not seriously injured but there is a fetal death. The case reviews points to side or oblique loading of the pregnant female by the crash delta V or yaw of the vehicle in 71.7% of the crashes. Several of the cases involved side airbag and curtain deployments raising the question of what type of counter-measures may be needed to protect the fetus in low severity crashes, where the crash does not pose a risk for serious injury to the adult occupants. Klinich et al. (1999Klinich et al. ( , 2008 found 23% of cases were side impacts compared to 64.2% in this study based on weighted field data. Use of the weighting factors in this study may help explain difference between the nationally representative sample reported here and the individual crash frequencies from the convenience sample of Michigan crashes reported by Klinich et al. (1999Klinich et al. ( , 2008. This study found 206 fetal deaths/yr in motor vehicle crashes. Klinich et al. (1999Klinich et al. ( , 2008) estimated 300-3,800 fetal deaths per year from maternal involvement in crashes for females over 20 weeks gestation using several methods and databases. Weiss (2006)  The risk of death to pregnant females was 0.065% ± 0.035% (95 th CI 0.000%-0.135%) in this study. It was higher than the risk of 0.036% (95% CI 0.025%-1.04%) found by Amezcua-Prieto et al. (2020). The average risk was within the 95 th CI. The risk for fetal death was 1.25% ± 0.74% (95 th CI 0.00%-2.75%) in this study. It was significantly higher than the risk of 0.066% (95% CI 0.38%-1.01%) found by Amezcua-Prieto et al. (2020). The average risk was higher than their 95 th CI. The lower risk for pregnant females may involve different crash experiences than non-pregnant female. The 2008-15 NASS-CDS and 2017-20 CISS field data and electronic case reviewer provides insights on fetal deaths and injury to pregnant females using the same sample of crashes. The crashes involved a vehicle being towed from the scene and only light vehicles with 2000þ MY (model year) were included. This study points to areas where counter-measures are needed to protect the fetus in low severity crashes.

Limitations
There are a number of limitations to the study of injury and death of pregnant females and death of fetuses using field accident data. First, fetal deaths were determined by the accident investigator's review of medical records, police reports and interviews. They were not derived from official fetal death certificates. Second, there can be confusion about what constitutes a fetal death, as most states do not report fetal deaths up to 20 weeks or so gestation. The fetal deaths in NASS-CDS and CISS are based on the field accident investigator's work in the case. Third, the NASS-CDS and CISS methods for ascertaining fetal death have never been externally validated. Weiss (2006) noted there is undercounting. Fourth, NASS-CDS and CISS define "A fetal fatality is indicated when fetal death occurs within 30 days of the crash. The death must have occurred as a consequence of the crash." A fetal death was determined by an interview question: was there any indication that this pregnant occupant lost the fetus and other facts about the crash and occupants (NHTSA 2009). The level of training and the procedures and protocols used by the accident investigator have not been evaluated. Similar types of limitations apply to determining pregnancy. Fifth, the trimester of pregnancy was not considered in the evaluation of seating position and crash type. There have been studies that the trimester is not a major determinant of crash risk from linked crash birth data (Hyde et al. 2003). Sixth, the NASS-CDS coding rules state that when pregnancy status is unknown cases are to be assigned to the "female not-reported pregnant" category. This can result in some misclassification, which have not been investigated. Seventh, the NASS-CDS and CISS crashes investigated in the study included only vehicles towed from the scene to match the two samples of field data. NASS-CDS included vehicles towed due to damage, while CISS included vehicles towed for any reason. This study did not differentiate the reason the vehicle was towed. The sample of crashes may miss fetal deaths in minor crashes that did not involve a tow away of the vehicle. Eighth, the unweighted sample of fetal deaths was 17. In our experience, this is a reasonably large number of cases involving a death in NASS-CDS and CISS. Our practice has been not to report trends on fewer than five cases, to add a caution for situations with <10 cases and to consider the sample reasonably robust with more than 10 cases. We typically judge the adequacy of the sample size based on in-depth review of the cases. We determined the 17 fetal death cases provided a wealth of information. Ninth, we determined that many of the crashes with fetal death involved minor damage to the case vehicle. We classified one of the injury mechanisms as a minor impact, based on review of photographs of the case vehicles. There is no specific definition of "minor" that everyone agrees upon. Klinich et al. (1999) defined minor by a delta V < 24 km/h (<15 mph). We did not use a delta V classification, but reviewed the photographs of the vehicle.
Appendix A shows photos of the damage to the vehicles that were classified as minor impacts. Tenth, the main finding of this study was the identification of all cases of fetal death in the NASS-CDS and CISS field data and analysis of the individual case information for the underlying mechanism of death. We also analyzed the risk for MAIS 3 þ F by seating position and crash type for males and females, including pregnant females. The number of years of field data included calendar years 2000-20. Various AIS coding versions were used by the crash investigators, including AIS98, based on AIS95, AIS08, based on AIS05, and AIS15. The effect of AIS coding and limitations of vehicles <10 years old was analyzed in Appendix B.

Funding
The author(s) reported there is no funding associated with the work featured in this article.