Schiff, Melissa A. MD, MPH; Mack, Christopher D. MS; Kaufman, Robert P. BS; Holt, Victoria L. PhD; Grossman, David C. MD, MPH
Motor vehicle crashes affect an estimated 32,800 pregnant women annually in the United States,1 with approximately 2.8% of pregnant women experiencing a crash during pregnancy.2 A crash during pregnancy can affect not only the pregnant woman but also her fetus, and can result in adverse pregnancy outcomes. Prior studies have noted that crashes are one of the leading causes of injury-related maternal3,4 and fetal deaths.5 In addition to fatalities, a crash is the most common cause of injury-related hospitalization among pregnant women.6
Although it has been well documented that seatbelts decrease the risk of maternal and fetal death and adverse pregnancy outcomes after motor vehicle crashes,2,7–9 relatively little is known about the effects of air bags on pregnancy. Since the introduction of federal legislation in 1998 requiring air bags as standard safety equipment in all automobiles and light trucks, studies10–13 have documented that air bags reduce fatalities from crashes in nonpregnant occupants. However, reports of death among infants, children,14–17 and women of short stature18,19 have increased awareness of the potential for adverse effects of air bags in special populations. Pregnant women represent an additional special population that may be adversely affected by air bags because of the proximity of the pregnant abdomen to the air bag in later pregnancy.
The effect of air bags on pregnancy outcomes has been investigated in only a limited number of studies. Several case reports of pregnant crash occupants have noted uterine rupture20 and placental abruption21 resulting in fetal deaths, and rupture of membranes22 occurring in crashes in which an air bag deployed. In contrast, Klinich et al23 used a non–population-based case series of 42 pregnant occupants in motor vehicle crashes and reported that air bags in conjunction with three-point seatbelt use were associated with fewer adverse fetal outcomes compared with three-point seatbelt use only. Because of the lack of population-based studies of this important motor vehicle safety issue among pregnant women, we performed a cohort study of all pregnant crash occupants in Washington State to evaluate the effects of air-bag availability and deployment on adverse pregnancy outcomes. We were interested in assessing effects of air-bag availability because this type of study evaluates simultaneously the effect of activation or no activation of an air bag and the effect the air bag has when it is activated.24 Evaluation of this exposure has been performed in prior studies,11,13,15 and from a public health perspective, it is essential in developing recommendations for pregnant women regarding the safety of driving or riding in a vehicle with an air bag. We hypothesized that air-bag deployment would be associated with increased risk of adverse pregnancy outcomes.
MATERIALS AND METHODS
This retrospective cohort study assessed the effect of air-bag availability and deployment on the risk of adverse pregnancy outcomes among women who were front seat occupants in motor vehicle crashes occurring during pregnancy in Washington State from 2002 to 2005. We identified pregnant women who were occupants in a motor vehicle crash by linking state birth and fetal death certificate data with Washington State Patrol crash data. The Washington State Patrol collects crash data on all motor vehicle traffic collisions occurring in Washington State that result in injury or death of any person or $500 or more in property damage. Information collected on Washington State Patrol crash reports includes personal identifiers, date of crash, air-bag availability, air-bag deployment, occupant seat position, seatbelt use, type and condition of roadway, posted speed limit, weather and lighting conditions at the crash scene, vehicle identification number, and vehicle make, model, and year. Because our identification of pregnant occupants was based on the presence of a birth or fetal death certificate, our study population was restricted to women with a live birth or fetal death at 20 weeks of gestation or greater. From this group, we excluded women who were in rollover crashes because these crashes involve different types of crash forces than nonrollover crashes. We also excluded crashes involving large trucks, buses, and motorcycles. This study was approved by the Washington State Institutional Review Board.
The Washington State Patrol crash report codes the air-bag availability of each crash vehicle as 1) not equipped with an air bag, 2) equipped with an air bag but not deployed, or 3) deployed air bag. We evaluated two different exposures in our cohort: air bag available compared with no air bag available and air bag deployed compared with no air bag deployed (restricted to crashes in which an air bag likely would have deployed). We classified all crash occupants by the availability (equipped not deployed and deployed) or nonavailability (not equipped) of an air bag in the vehicle. For the 536 vehicles (16.0%) for which we did not have air-bag information, we classified air-bag availability by processing the vehicle identification number through the VINDICATOR program (Insurance Institute for Highway Safety, Arlington VA) that decodes the vehicle identification number and identifies the vehicle year, make, and model and determines whether the vehicle was equipped with air bags. We excluded 162 women in vehicles for which we were not able to determine air-bag availability from either the Washington State Patrol crash report or the VINDICATOR program. We were interested in the availability of an air bag because from a public health perspective it is essential in developing recommendations for pregnant women regarding the safety of driving or riding in a vehicle with an air bag.
We also evaluated the effects of air-bag deployment on pregnant women. An air bag typically deploys in the event of a moderate or severe frontal crash and at a specific threshold determined by the change in velocity. We classified crash occupants as having a deployed air bag if this was coded on the Washington State Patrol crash report. We compared these occupants with those in a vehicle without an air bag in a type of crash where an air bag would have deployed if it had been present. This was determined by having an experienced crash investigator (author R.P.K.) who was blinded to the perinatal outcomes review the crash and vehicle data contained in the Washington State Patrol crash report for each individual crash to determine the probable direction of force for the crash based on vehicle action codes and the likelihood of an air-bag deployment. We initially evaluated the specific codes documenting the type of collisions in which an air bag deployed and then restricted our comparison group (no air bag available) to only these types of collisions as well. The Washington State Patrol crash data did not include crash investigation information on the change in velocity. This comparison group of occupants was classified as in a crash without an air-bag deployment.
To evaluate maternal and perinatal outcomes, we used the Washington State Comprehensive Hospitalization and Recording System data, which contain the maternal and infant International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) diagnosis and procedure codes for the delivery hospitalization, linked to the birth or fetal death certificate data. A prior study evaluating the linkage procedure for creating the Birth Event Records Database file reported that the linkage was 96% complete.25 Since 1980, the Washington State birth and fetal death certificates have included a checklist for obstetric procedures, method of delivery, complications of labor and delivery, and abnormal conditions of the newborn. Maternal outcomes evaluated after a crash were the following: preterm labor less than 37 weeks of gestation (ICD-9-CM codes 644.0 and 644.2 or use of tocolysis on the birth or fetal death certificate), placental abruption (ICD-9-CM code 641.2 or as listed on the birth or fetal death certificate), induction of labor (ICD-9-CM diagnosis codes 659.0 and 659.1; procedure codes 73.1, 73.4, and 96.49; or as listed on the birth or fetal death certificate), and cesarean delivery (ICD-9-CM codes 74–74.2 or as listed on the birth or fetal death certificate).
Perinatal outcomes investigated were the following: less than 37 weeks of gestation (as recorded on the birth or fetal death certificate), low birth weight (less than 2,500 g on the birth or fetal death certificate), small for gestational age, moderate to heavy meconium at delivery (birth or fetal death certificate), fetal distress (ICD-9-CM codes 656.3 and 768.2–768.4 or as listed on the birth or fetal death certificate), respiratory distress syndrome (ICD-9-CM code 769 or as listed on the birth certificate), and fetal death. We used an algorithm by Alexander et al26 to define small for gestational age as infant birth weight below the 10th percentile for a given gestational age, stratified on sex and race/ethnicity. We derived the gestational age at the time of crash by subtracting the number of weeks between the crash as listed on the Washington State Patrol crash report and the birth or fetal death. We defined fetal death as an intrauterine death identified by a fetal death certificate.
We evaluated the frequency of nonrollover crashes per 1,000 deliveries using the number of police-reported crashes as the numerator and the sum of the number of live births and fetal deaths during the study period 2002–2005 as the denominator. We also evaluated the proportion of crashes by pregnancy trimester. We compared crash occupants with an air bag available, regardless or deployment (n=2,207), and those without an air bag available (n=1,141) on demographic, obstetric, vehicle, and crash characteristics.
We used Poisson regression analysis to calculate relative risks and 95% confidence intervals for associations between availability of air bag and maternal and perinatal outcomes. In our initial analyses, we built a regression model for each pregnancy outcome comparing pregnant women in vehicles with an air bag and pregnant women in vehicles with no air bag. We performed subanalyses to evaluate adverse pregnancy outcomes among crash occupants in vehicles in which an air bag deployed and compared them with those in vehicles with no air bag in the type of crash in which an air bag would have deployed if present as described previously. We assessed possible interactions between air-bag availability by seatbelt use, air-bag availability by trimester of crash, and air-bag availability by seat position of driver or passenger using the likelihood ratio (P<.05) and found no significant interactions. We assessed possible confounding of the associations of air-bag availability and air-bag deployment and adverse pregnancy outcomes by age, race, income, parity, prenatal smoking, vehicle type, vehicle model year, seatbelt use, seat position, and crash severity. Relative risks were adjusted for age, seatbelt use, and vehicle model year. For selected maternal and perinatal outcomes with fewer than 10 observations, we present unadjusted relative risks. All statistical analyses were performed using Stata 9.0 (Stata Corporation LP, College Station, TX).
We found 3,348 police-reported, nonrollover crashes among pregnant front seat occupants during the study time period of 2002–2005. The frequency of motor vehicle crashes was 10.3 per 1,000 deliveries with 26.1% of crashes in the first trimester, 42.9% in the second trimester, and 31.0% in the third trimester. An air bag was present in 65.9% of the vehicles. Among occupants for whom seatbelt use was known, 96.1% wore a seatbelt. In comparing demographic and obstetric characteristics among pregnant occupants with and without an air bag available in their vehicle (Table 1), we found that pregnant occupants in vehicles with an air bag available were older, were more likely to be Asian, were more likely to be married, and had greater educational levels compared with occupants in vehicles without an air bag available. Occupants in vehicles with an air bag also had greater income and were more likely to have commercial insurance and prenatal care in the first trimester. The trimester in which the crash occurred was similar for the two groups of pregnant occupants.
We also compared vehicle and crash characteristics for the occupants with and without an air bag available in the vehicle (Table 2). Crash conditions, including weather, lighting, and road conditions, were similar for the two groups of pregnant occupants. Crash severities as recorded in police reports were similar, with most crashes resulting in property damage or injury. Pregnant occupants in air-bag–equipped vehicles were more likely to be in a light truck or van, to be belted, to wear a lap and shoulder belt, and to be the driver compared with occupants in vehicles without an air bag available.
Controlling for maternal age, seatbelt use, and vehicle model year, we found that maternal and perinatal outcomes were similar among the pregnant occupants in crashes of air-bag–equipped and air-bag–unequipped vehicles (Table 3). In our subanalyses of crashes in which air-bag deployment would be likely, for the majority of maternal and perinatal outcomes, we found no increased or decreased risk associated with air-bag deployment. We did find a nonsignificant 70% increased risk of preterm labor and a nonsignificant threefold increase in fetal death among pregnant occupants in crashes in which an air bag deployed compared with those with no air-bag deployment (Table 4), although fetal death results were limited by small numbers (2/198 [1.0%] in pregnant women whose air bags deployed; 2/622 [0.3%] in pregnant women whose air bags did not deploy). Our findings did not vary for air-bag availability or air-bag deployment by seatbelt use, trimester of the crash, or seat position.
Our finding of 10.3 crashes per 1,000 deliveries is similar to the rate of nine crashes per 1,000 live births reported by Weiss et al1 using a National Highway Traffic Safety Administration data set that includes crash information on a probability sample of all crashes reported to police in the United States, but smaller than the 2.8% reported by Hyde et al using Utah crash data. Similar to our study, Weiss et al1 reported that more crashes among pregnant occupants occurred in the second (36.4%) and third (33.8%) trimesters than in the first trimester (29.8%).
The evaluation of adverse pregnancy outcomes associated with air-bag availability is essential from a public health perspective in developing recommendations for pregnant women regarding the safety of driving or riding in a vehicle with an air bag in the event of a crash. The presence of an air bag incorporates the availability of the safety device in the vehicle as well as the deployment of the device in the event of a crash. Lack of deployment can be explained mainly by the direction of force of the crash (nonfrontal crashes) or a low-speed frontal crash that does not result in deployment of the air bag. We found no association between the presence of an air bag in the vehicle and risk of adverse maternal or perinatal outcomes in the event of a crash, possibly because only a small proportion of the crashes that we evaluated resulted in air-bag deployment.
All prior studies of the effects of air bags during pregnancy have evaluated crashes in which an air bag deployed. Motor vehicle crash case reports are described in which the pregnant occupant had rupture of membranes,22 uterine rupture,20 or placental abruption21 resulting in fetal death, and documentation of extensive brain injury by autopsy findings that resulted in fetal death.27 In addition to case reports, several case series of pregnant occupants in crashes have been reported. Klinich et al23 evaluated 42 crashes involving pregnant occupants using a non–population-based case series and suggested that air bags in conjunction with three-point seatbelt use may be associated with fewer adverse fetal outcomes compared with three-point seatbelt use only. Metz et al28 reported adverse pregnancy outcomes among a case series from two hospitals of 30 pregnant occupants in crashes with air-bag deployment, with 1 occupant experiencing a placental abruption resulting in fetal death, 73% experiencing uterine contractions, and 20% having abnormal fetal heart tones. These studies were limited by small sample sizes and the potential for selection bias as the cases were not selected from a defined study population and were likely to overrepresent adverse maternal or fetal outcomes because of reporting bias. The study by Metz et al was also limited by the absence of an unexposed (no air bag) comparison group of pregnant occupants in crashes. Our main findings should provide clinicians with evidence to advise women that air bags do not seem to increase risk for most potential adverse outcomes during pregnancy. Our finding of a nonsignificant 70% increased risk of preterm labor and a nonsignificant threefold increased risk of fetal death among the occupants in a vehicle in which an air bag deployed highlights the need for additional, larger studies to identify whether air-bag deployment increases the risk of these outcomes. In addition, we found that air bags did not result in a decreased risk of adverse pregnancy outcomes. This may be explained by the fact that air bags originally were designed to protect a male occupant of average height and weight. Pregnant occupants undergo continuous anatomical and physiologic changes during pregnancy, and these changes were not considered when air bags were designed. In addition, prior studies10,13 have documented that air bags are not a major additional source of protection for otherwise belted occupants. Clinicians should continue to advise pregnant women to wear seatbelts because they decrease the risk of maternal and fetal death and adverse pregnancy outcomes in the event of a crash.2,7–9
Our study had several limitations that may affect the interpretation of our results. Our comparison that used the no-air-bag-available group for evaluating the effects of air-bag availability and air-bag deployment was not ideal. The no-air-bag-available and no-air-bag-deployment groups both consisted of vehicles that had no air bag. These vehicles without air bags differed from those with an air bag in terms of age and associated safety features in the vehicle, with no air-bag vehicles being older and having fewer safety features. This may have confounded our findings, although we attempted to control for these differences by adjusting for model year in the analyses. In addition, our assessment of which vehicles would have had an air bag deploy in a crash if an air bag had been available may have resulted in misclassification of the potential for air-bag deployment because we did not have information such as the change in velocity (important for air-bag deployment) and had limited Washington State Patrol information to determine the specific direction of force of the crashes. This differential misclassification would have resulted in a possible overestimate of the risks associated with air-bag deployment if crashes that were nonfrontal or had a lower change in velocity were included in the no deployment group. We attempted to mitigate this possible misclassification by using an experienced crash investigator to review the crash report data to best categorize these based on direction of movements and vehicle actions as well as selecting only crash types similar to the air-bag deployment group. We also were unable to control completely for crash severity because we did not have information on the change in velocity, and crash severity is likely a confounder of the relationship between air bags and adverse outcomes.
We did not attempt to differentiate the effect of second-generation air bags in the vehicles that are depowered and deploy with less force and may be associated with a different risk of adverse outcomes. We also did not have information on side air bags because this information was not collected by Washington State Patrol during the years of this study. Our study may also have been limited by the completeness and accuracy of personal identifiers in the Washington State Patrol crash data, resulting in our data linkage procedures not identifying all pregnant women in crashes during the study time period. In addition, our analysis was restricted to pregnant women who had a delivery at 20 weeks of gestation or greater. We are unable to determine the effect of air bags on risk of spontaneous abortion at less than 20 weeks of gestation. Last, our findings of the effects of air bags on fetal death should be interpreted with caution because our study was limited by small numbers of pregnant women experiencing these outcomes. Our study power to detect a threefold difference in the risk of fetal death was approximately 30%. To have sufficient power to detect this large of a difference with 80% power, we would need approximately 1,630 pregnant women in crashes with air-bag deployment and a similar number in crashes without air-bag deployment.
In conclusion, we found that pregnant occupants are not at increased risk of adverse pregnancy outcomes while traveling in a vehicle that is equipped with an air bag and crashes. The American College of Obstetricians and Gynecologists29 and the National Highway Traffic and Safety Administration30 both recommend that pregnant women who are occupants in motor vehicles should wear lap and shoulder seatbelts and should not turn off air bags. Our findings are in agreement with these recommendations. Although we found a nonsignificant increased risk of fetal death among pregnant occupants in motor vehicle crashes with air bag deployment, our results should be interpreted with caution because of the limited sample size and study power. Future studies of the effects of air bags in pregnancy with larger samples of pregnant occupants in crashes should evaluate the risk of fetal death with greater precision. Future studies should also evaluate the effect of air bags on pregnancy outcomes that occur at less than 20 weeks of gestation.
1.Weiss HB, Strotmeyer A. Characteristics of pregnant women in motor vehicle crashes. Inj Prev 2002;8:207–10.
2.Hyde LK, Cook LJ, Olson LM, Weiss HB, Dean JM. Effect of motor vehicle crashes on adverse fetal outcomes. Obstet Gynecol 2003;102:279–86.
3.Schiff M, Albers L, McFeeley P. Motor vehicle crashes and maternal mortality in New Mexico: the significance of seat belt use. West J Med 1997;167:19–22.
4.Harper M, Parsons L. Maternal deaths due to homicide and other injuries in North Carolina: 1992–1994. Obstet Gynecol 1997;90:920–3.
5.Weiss HB, Songer TJ, Fabio A. Fetal deaths related to maternal injury. JAMA 2001;286:1863–8.
6.Weiss HB. Pregnancy-associated injury hospitalizations in Pennsylvania, 1995. Ann Emerg Med 1999;34:626–36.
7.Crosby WM, Costiloe JP. Safety of lap-belt restraint for pregnant victims of automobile collisions. N Engl J Med 1971;284:632–6.
8.Wolf ME, Alexander BH, Rivara FP, Hickok DE, Maier RV, Starzyk PM. A retrospective cohort study of seatbelt use and pregnancy outcome after motor vehicle crash. J Trauma 1993;34:116–9.
9.Pearlman MD, Klinich KD, Schneider LW, Rupp J, Moss S, Ashton-Miller J. A comprehensive program to improve safety for pregnant women and fetuses in motor vehicle crashes: a preliminary report. Am J Obstet Gynecol 2000;182:1554–64.
10.Zador PL, Ciccone MA. Automobile driver fatalities in frontal impacts: air bags compared with manual belts. Am J Public Health 1993;83:661–6.
11.Lund AK, Ferguson SA. Driver fatalities in 1985–1993 cars with airbags. J Trauma 1995;38:469–75.
12.Crandall CS, Olson LM, Sklar DP. Mortality reduction with air bag and seat belt use in head-on passenger car collisions. Am J Epidemiol 2001;153:219–24.
13.Cummings P, McKnight B, Rivara FP, Grossman DC. Association of driver air bags with driver fatality: a matched cohort study. BMJ 2002;324:1119–22.
14.National Highway Traffic Safety Administration. Fifth/sixth report to Congress: effectiveness of occupant protection systems and their use. Washington, DC: US Department of Transportation; 2001.
15.Cummings P, Koepsell TD, Rivara FP, McKnight B, Mack C. Airbags and passenger fatality according to passenger age and restraint use [published erratum appears in Epidemiology 2002;13:749]. Epidemiology 2002;13:525–32.
16.Cooper JT, Balding LE, Jordan FB. Airbag mediated death of a two-year old child wearing a shoulder/lap belt. J Forensic Sci 1998;43:1077–81.
17.Willis BK, Smith JL, Falkner LD, Vernon DD, Walker ML. Fatal air bag mediated craniocervical trauma in a child. Pediatr Neurosurg 1996;24:323–7.
18.Kleinberger M, Summers L. Mechanisms of injuries for adults and children resulting from airbag interaction. In: Proceedings: 41st Annual Proceedings of the Association for the Advancement of Automotive Medicine; November 10–12, 1997; Orlando, Florida. Barrington (IL): Association for the Advancement of Automotive Medicine; 1997. p. 405–20.
19.Maxeiner H, Hahn M. Airbag-induced lethal cervical trauma. J Trauma 1997;42:1148–51.
20.Fusco A, Kelly K, Winslow J. Uterine rupture in a motor vehicle crash with airbag deployment. J Trauma 2001;51:1192–4.
21.Schultze PM, Stamm CA, Roger J. Placental abruption and fetal death with airbag deployment in a motor vehicle accident. Obstet Gynecol 1998;92(pt 2):719.
22.Gimovsky ML, Nunez G, Beck P. Fetal heart rate monitoring casebook. Airbag-associated rupture of membranes: evaluation of trauma in pregnancy. J Perinatol 2000;4:270–3.
23.Klinich KD, Schneider LW, Moore JL. Investigations of crashes involving pregnant occupants. In: 44th Annual Proceedings of the Association for the Advancement of Automotive Medicine; October 2–4, 2000; Chicago, Illinois. Barrington (IL): Association for the Advancement of Automotive Medicine; 2000. p. 37–55.
24.Cummings P, Weiss N. Re: “Mortality reduction with air bag and seat belt use in head-on passenger car collisions.” Am J Epidemiol 2001;154:387–9.
25.Hohner V, Armstrong R. Birth event records project: developing a permanent linked database. Olympia (WA): Washington State Department of Health; 1991.
26.Alexander GR, Kogan MD, Himes JH. 1994–1996 U.S. singleton birth weight percentiles for gestational age by race, Hispanic origin, and gender. Matern Child Health J 1999;3:225–31.
27.Karimi P, Ramus R, Urban J, Perlman JM. Extensive brain injury in a premature infant following a relatively minor maternal motor vehicle accident with airbag deployment. J Perinatol 2004;24:454–7.
28.Metz TD, Abbott JT. Uterine trauma in pregnancy after motor vehicle crashes with airbag deployment: a 30-case series. J Trauma 2006;61:658–61.