a b s t r a c t

Background: Literature predating routine availability of airbags reported an association between seatbelt signs and internal injuries. We measured this association among patients involved in motor vehicle crashes (MVCs) with airbag deployment.

Methods: We conducted a retrospective cohort study by chart review of all MVC patients presenting to our Emer- gency Department (ED) during 1 January 2008-30 September 2015. We included all adult MVC patients in the driver or front passenger seats with both shoulder and lap seatbelts and airbag deployment. Two trained chart abstractors recorded data regarding restraints and airbag deployment. We obtained all other data via electronic medical record abstraction including demographics, injuries, and survival. We compared the prevalence of cervicothoracic and intra-abdominopelvic injuries between patients with a documented seatbelt sign versus no seatbelt sign using a logistic regression model.

Results: Of 1379 MVC patients, 350 met inclusion criteria. Of these, 138 (39.4%) had a seatbelt sign. The preva- lence of cervicothoracic injury was higher among subjects with a documented seatbelt sign (54.3% versus 42.9%, p = 0.036) Seatbelt sign predicted cervicothoracic injury with a positive likelihood ratio of 1.3 (95% CI 1.0-1.7) and negative likelihood ratio of 0.8 (95% CI 0.7-1.0). The odds ratio of cervicothoracic injury among pa- tients with a seatbelt sign versus no seatbelt sign was 1.58 (95% confidence interval 1.02-2.46) in the logistic re- gression model. There was no association between seatbelt sign and intra-abdominopelvic injury (p = 0.418). Conclusions: In the setting of airbag deployment, there is an association between seatbelt sign and cervicothoracic injury but not intra-abdominopelvic injury.

Introduction

Background

Motor vehicle crashes (MVCs) result in a substantial burden of mor- bidity and mortality in the United States [1-3]. MVCs accounted for over 35,000 deaths nationwide in 2015 [4]. Year-to-year variations notwith- standing, there has been a general trend towards fewer annual traffic- related deaths among motor vehicle passengers over the past decade [5,6]. This reduction in mortality is likely due in large part to the rising use of seatbelts [7,8].

* Corresponding author at: MCHE-EMR, 3551 Roger Brooke Dr., Fort Sam Houston, TX 78234, United States.

E-mail address: [email protected] (M.D. April).

The advent of seatbelt restraints and airbag technology has led to dramatically lower mortality related to MVCs [7-10]. Yet, these devices result in distinct injury patterns related to their use [11]. Seatbelt use during high impact MVCs may result in the transmission of significant forces to the abdomen and chest. These forces may generate skin abra- sions and ecchymosis termed the “seatbelt sign,” a physical examination finding long reported to portend a higher risk of Significant injuries to internal organs and vasculature [12-14].

The applicability of these data in the modern era are unclear given that many of these studies predated the widespread availability of airbags. It is possible that airbag deployment may mitigate and redirect some of these forces. The existing literature suggests that airbag use re- sults in alternative injury patterns distinct from those associated with seatbelt use. The diversion by airbags of blunt force energy away from the abdomen appears to result in lower mortality [9,10,15] and a

http://dx.doi.org/10.1016/j.ajem.2017.09.011 0735-6757/

distribution of injuries more frequently involving the face [16,17] and extremities [11,18,19] rather than internal organs.

Study objective

To our knowledge, no studies have re-examined the association be- tween seatbelt sign and internal injuries in the setting of concomitant airbag deployment. The objective of this study was to compare the prev- alence of internal injuries among restrained front passengers involved in an MVC with airbag deployment presenting with a seatbelt sign ver- sus no seatbelt sign. We hypothesized that those patients with a seatbelt sign would continue to demonstrate a higher prevalence of internal in- juries compared to patients without a seatbelt sign.

Methods

Study design and setting

We conducted a retrospective cohort analysis by chart review of adults treated at our hospital following an MVC with seatbelt use and airbag deployment. Our hospital is an urban tertiary care hospital and level 1 trauma center. Our institutional review board approved the study (Brooke Army Medical Center protocol number C.2016.005d). We conducted the study in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) state- ment [20].

Selection of subjects

We defined all inclusion and exclusion criteria a priori. Inclusion criteria comprised adult (N 17 years) drivers or front seat passengers in- volved in an MVC with documentation of both seatbelt use and airbag deployment. Exclusion criteria comprised patients experiencing MVCs involving vehicles other than cars or light trucks and patients who had unknown information with regards to (1) collision type, (2) seat-belt or airbag availability, (3) seat-belt sign, and (4) diagnosis (ICD code). The study period spanned from 1 January 2008 (inception of our institu- tional trauma registry) through 30 September 2015 (based on sample size estimate).

Data collection

Our study utilized chart review methodology [21]. We first used a computer-based query of our institutional trauma registry based upon our inclusion and exclusion criteria to retrieve eligible subjects. We spe- cifically queried for the International Classification of Diseases (ICD-9) related to MVCs (Supplementary Appendix Table A1). The initial query yielded 1379 consecutive subjects potentially eligible for study inclusion. Investigators with emergency medicine training not blinded to the study hypothesis manually reviewed the electronic medical re- cord for each of these subjects to ensure eligibility for study inclusion. Of the 1379 charts reviewed, 400 underwent review by a second inves- tigator to assess inter-rater reliability for purposes of quality assurance. Investigators resolved discordant eligibility decisions by mutual agreement.

Upon identification of the study sample, we electronically abstracted all study variables from the trauma registry. These data included subject demographics, subject’s position in the car (driver’s seat versus front passenger seat), and presence or absence of seatbelt sign. Regarding the latter variable, we counted only those patients whose documenta- tion of physical examination used the specific term “seatbelt sign;” we did not include patients with documentation of chest wall or abdominal abrasions or ecchymosis. We also electronically abstracted data regard- ing the primary outcome of internal injury based on ICD-9 codes. We defined internal cervicothoracic injuries (Supplementary Appendix Table A2) and intra-abdominopelvic injuries (Supplementary Appendix

Table A3) based on distinct ICD-9 codes. Finally, we determined wheth- er each subject survived to hospital discharge.

Analysis

Our study sample size estimate assumed ? = 0.05 and ? = 0.20 with two-sided testing. We anticipated approximately 27% of subjects would have documentation of seatbelt sign [22]. Among subjects with a seatbelt sign, we anticipated a prevalence of internal injury of 50% [23]. We sought to power our study to detect a 20% difference in the prevalence of internal injury between patients with documentation of seatbelt sign versus documentation of no seatbelt sign [24]. This yielded a sample size of 344. Based on preliminary analyses of our trauma reg- istry, we determined we would need to retrieve subjects from 1 January 2008 through 30 September 2015 to achieve this sample size.

We consolidated all data into a secured database (Excel version 14, Microsoft, Redmond, WA). We then exported all data into SPSS for sta- tistical analysis (version 21, IBM, Armonk, NY). We calculated a kappa coefficient to assess inter-rater reliability for the determination of pa- tient eligibility for the study among the 400 charts screened by two in- vestigators. We used descriptive statistics to describe subject demographics and car position: means for continuous variables and proportions for categorical variables with 95% confidence intervals (CIs). We stratified these descriptive statistics according to the presence or absence of documented seatbelt sign.

We then compared the prevalence of internal injuries between sub- jects with documented seatbelt sign versus subjects without document- ed seatbelt sign using a chi-squared test. Our primary analysis stratified this comparison according to prevalence of cervicothoracic or intra- abdominopelvic internal injuries. We calculated diagnostic accuracy characteristics for documented seatbelt sign for either cervicothoracic or intra-abdominopelvic injuries to include sensitivity, specificity, posi- tive likelihood ratio (LR+), negative likelihood ratio with 95% CIs. We further stratified comparisons for individual injuries using Fisher’s exact test.

We also compared in-hospital mortality between patients without versus patients with documented seatbelt sign. Finally, we calculated the odds ratios (OR) of cervicothoracic and intra-abdominopelvic inju- ries based upon documentation of a seatbelt sign using logistic regres- sion models to control for confounders. These logistic regression models controlled for age, sex, and patient position in the vehicle (driver’s seat versus front passenger seat).

Results

Study sample characteristics

We reviewed the records of 1379 MVC patients presenting to our hospital during the study period. Of these, 350 ultimately met all criteria for inclusion in the study (Fig. 1). The kappa statistic for the determina- tion of patient eligibility between the two investigators screening charts was 0.932. There was documentation of seatbelt sign in 138 (39.4%) pa- tients. There were no significant differences in the characteristics of pa- tients without documented seatbelt sign versus those patients with documented seatbelt sign (Table 1).

Main results

There was an association between documentation of seatbelt sign and the presence of cervicothoracic injury (Table 2). The prevalence of cervicothoracic injury was 42.9% among patients without documented seatbelt sign versus 54.3% among patients with seatbelt sign (p = 0.036). The diagnostic accuracy of documented seatbelt sign for intra- thoracic injury was sensitivity 45.2% (95% CI 37.5-53.1%), specificity

65.8% (95% CI 58.4-72.5%), LR+ 1.3 (95% CI 1.0-1.7), and LR- 0.8 (95%

CI 0.7-1.0). There was no association between documentation of

Fig. 1. Flow diagram of study subject inclusion.

seatbelt sign and the presence of intra-abdominopelvic injury. Similarly, there was no association between documentation of seatbelt sign and in-hospital mortality: 1.9% among patients without seatbelt sign versus 0.0% among patients with seatbelt sign.

The distributions of cervicothoracic and intra-abdominopelvic inju- ries were broadly comparable between patients with documented seatbelt sign versus patients without documented seatbelt sign. Regard- ing cervicothoracic injuries, the most common injuries sustained in- cluded rib fractures and pulmonary contusions (Table 3). In stratified analyses, the only specific thoracic injury predicted by a seatbelt sign was retrosternal hematoma: 2.8% in patients without a seatbelt sign versus 10.1% in patients with a seatbelt sign (p = 0.008). A higher pro- portion of patients without a seatbelt sign had cervical spine fractures as compared to patients with a seatbelt sign: 14.6% versus 7.2% (p = 0.041). We did not observe any associations between seatbelt sign and other specific cervicothoracic injuries in the remaining stratified analyses.

Regarding intra-abdominopelvic injuries, the most common injuries sustained included lumbosacral fractures and pelvic fractures. There were no individual intra-abdominopelvic injuries for which stratified analyses demonstrated an association with seatbelt sign (Table 4). The proportion of patients with small bowel injuries specifically was 2.8% among patients without a seatbelt sign versus 0.7% among patients with a seatbelt sign (p = 0.252).

Our main findings from the primary analysis held in logistic regres- sion analyses (Table 5). There was an association between documented seatbelt sign and cervicothoracic injury (OR 1.63, 95% CI 1.05-2.54). Conversely, there was no such association between documented

Table 1

SubjeCT characteristics

seatbelt sign and intra-abdominopelvic injury (OR 1.21, 95% CI 0.76- 1.92).

Discussion

Overview

The advent of seatbelt restraints and airbag technology has led to dramatically lower mortality related to MVCs [7-10]. Yet, these devices result in distinct injury patterns related to their use [11]. To our knowl- edge, this study represents the first examination of the prognostic indi- cations of the seatbelt sign specifically among patients experiencing airbag deployment. Our study suggests that among these patients, an association between seatbelt sign and cervicothoracic injury persists. This finding is noteworthy given that previous studies suggest that the concomitant use of seatbelts and airbag deployment may be protective against thoracic pathology including blunt thoracic aortic injury [25-27]. Yet, when a seatbelt sign is present, our data support a heightened index of suspicion for thoracic injuries in the setting of airbag deployment. Conversely, there appears to be no such association between seatbelt sign and intra-abdominopelvic injury. These results may help to guide judicious and cost-effective imaging and disposition strategies among MVC patients [28].

While our findings support an association between seatbelt sign and

cervicothoracic injuries in the setting of airbag deployment, the strength of this association is much less than that reported in studies preceding the airbag era. A systematic review of 12 studies involving 10,757 patients found LR+ for Intra-abdominal injuries among patients with seatbelt signs range 5.6-9.9 [23]. Less data exists with regards to the accuracy of seatbelt sign for thoracic injury but indicates a compara- ble LR+ of 5.0 [29]. The lack of association between seatbelt sign and intra-abdominopelvic injuries and a more modest association between seatbelt sign and cervicothoracic injuries (LR+ 1.3) perhaps reflect

Variable Patients without seatbelt sign (n = 212)

Patients with seatbelt sign (n = 138)

the protection conferred by airbag deployment in our unique patient

cohort.

Mean age, years (95% CI) 45.4 (42.8-47.9) 45.1 (41.9-48.3)

Male sex, % (95% CI) 52.4 (45.3-59.4) 51.4 (43.5-59.4)

Driver seat, % (95% CI) 84.4 (79.2-89.2) 81.9 (75.4-88.4)

CI-confidence interval.

Indeed, our study examined a very specific patient population. By in- cluding only those patients with documentation of both seatbelt use and airbag deployment while in the driver or front passenger seats, we cannot extrapolate these findings to all patients involved in MVCs.

Table 2

Subject outcomes

Variable Patients without seatbelt sign (n = 212)^a Patients with seatbelt sign (n = 138)^b p^c

Cervicothoracic injury, n (%)	91 (42.9%)	75 (54.3%)	0.036
Intra-abdominopelvic injury, n (%)	62 (29.2%)	46 (33.3%)	0.418
Any internal injury, n (%)	117 (55.2%)	89 (64.5%)	0.084
Mortality, n (%)	4 (1.9%)	0 (0.0%)	0.105

^a Percentages reflect the proportion of patients without documented seatbelt sign (denominator 212).

^b Percentages reflect the proportion of patients with documented seatbelt sign (denominator 138).

^c Statistical tests are chi-squared tests.

Only 25.4% of the charts we reviewed met all of these inclusion criteria. There are several reasons why other patients may have not met all in- clusion criteria. First, some patients may have, in fact, met all inclusion criteria in reality, but if providers did not properly document seatbelt use and airbag deployment, they lacked eligibility for our study. Second, many patients may have experienced MVCs in vehicles not equipped with airbag deployment; we expect the proportion of patients to which this category applies will continue to decrease in the future as airbags become increasingly ubiquitous. Third, our requirement of airbag deployment for study inclusion makes it likely that our cohort ex- perienced high-velocity MVCs.

Regarding the external validity of our study, we observed a higher proportion of patients with documentation of seatbelt sign than many previous studies. We found 39.4% of patients had documentation of this physical examination finding versus 12-36% in previous research [30-32]. We suspect much of this discrepancy reflects our requirement of documentation of seatbelt use by all study participants. These previ- ous studies did not require such documentation.

The distribution of cervicothoracic injuries we observed is broadly consistent with that reported elsewhere in the literature [33,34]. Specif- ically, the most common cervicothoracic injuries included rib fractures, pulmonary contusion, cervical spine fractures, and pneumothorax. The distribution of intra-abdominopelvic injuries in our study was also sim- ilar to that reported by other authors [35]. The most common abdominal organs injured included the spleen, kidney, and liver. We also noted high proportions of injuries attributable to lumbosacral or Pelvic fractures.

Our analyses stratifying the incidence of individual injuries by the presence of seatbelt sign are limited by small numbers and multiple

Table 3

Distribution of cervicothoracic injuries

inferential testing. Consequently, readers must interpret findings from these subgroup analyses with caution. Nevertheless, the finding of asso- ciation between seatbelt sign and retrosternal hematoma was particu- larly robust (p = 0.008). Less robust but nevertheless interesting was our finding that patients with a seatbelt sign were less likely to have a cervical spine fracture (p = 0.041). It is possible that patients with seatbelt sign may experience less neck mobility during impact; it would be interesting for future studies to examine this phenomenon to see if this association persists in other settings.

Limitations

There are several important limitations to our study. First, our chart review methodology depends upon the accuracy of the medical records for all study variables [21]. To the extent that data inaccuracies occurred with more or less frequency based upon the presence or absence of ei- ther our main independent variable (seatbelt sign) or dependent vari- ables (cervicothoracic and intra-abdominopelvic injuries) our results may reflect bias. Data inaccuracies could potentially have also resulted in a selection bias as we had no mechanism to objectively and indepen- dently verify documentation of subject eligibility criteria (e.g., re- straints, airbag deployment, position in vehicle). Furthermore, we were unable to collect data on many relevant data variables not routine- ly reported in medical charts. Examples include specific location of the seatbelt sign, car make and model, and velocity and vector of vehicle collisions.

A related point is that documentation of seatbelt sign is potentially subjective. The seatbelt sign as historically described and generally un- derstood represents a specific syndrome representing pain with ab- dominal or chest wall ecchymosis in the setting of recent MVC. It is possible that instances of medical record documentation of abdominal

Injuries Patients

without

Patients with p^c seatbelt sign

Table 4

Distribution of intra-abdominopelvic injuries

seatbelt sign (n = 138)

(n = 212)

n %^a n %^b

Rib fractures 59 27.8% 43 31.2% 0.548

Pulmonary contusion Cervical spine fracture	24 31	11.3% 14.6%	19 10	13.8% 7.2%	0.509 0.041	n %a n %b
Pneumothorax	21	9.9%	17	12.3%	0.481	Lumbosacral fractures	57	26.9%	29	21.0%	0.253
sternal fracture	17	8.0%	17	12.3%	0.194	Pelvic fractures	40	18.9%	16	11.6%	0.075
Retrosternal/chest hematoma	6	2.8%	14	10.1%	0.008	Splenic laceration	10	4.7%	11	8.0%	0.251
Other fractures	10	4.7%	10	7.2%	0.351	Renal lacerations/hematomas	10	4.7%	3	2.2%	0.261
Clavicle fracture	10	4.7%	6	4.3%	1.000	Liver laceration/hematomas	5	2.4%	8	5.8%	0.146
thoracic spine fracture	8	3.8%	7	5.1%	0.596	vascular injuries	5	2.4%	7	5.1%	0.230
Other Pulmonary injuries	8	3.8%	3	2.2%	0.537	Small bowel injuries	6	2.8%	1	0.7%	0.252
Vascular injury	3	1.4%	2	1.4%	1.000	Mesenteric injuries	1	0.5%	2	1.4%	0.564
Otherd	4	1.9%	4	2.9%	0.717	Otherd	2	0.9%	6	4.3%	0.062

Injuries Patients

without seatbelt sign (n = 212)

Patients with p^c seatbelt sign

(n = 138)

^a Percentages are of total number of patients without seatbelt sign (n = 212). Per- centages do not sum to total percentage of these patients sustaining cervicothoracic inju- ries (42.9%) due to some patients sustaining multiple injuries.

^b Percentages are of total number of patients with seatbelt sign (n = 138). Percentages do not sum to total percentage of these patients sustaining cervicothoracic injuries (54.3%) due to some patients sustaining multiple injuries.

^c Statistical tests are Fisher’s exact tests.

^d Other category includes injuries to the trachea, pericardium, nerves, and internal hematomas.

^a Percentages are of total number of patients without seatbelt sign (n = 212). Per- centages do not sum to total percentage of these patients sustaining intra-abdominopelvic injuries (29.2%) due to some patients sustaining multiple injuries.

^b Percentages are of total number of patients with seatbelt sign (n = 138). Percentages do not sum to total percentage of these patients sustaining intra-abdominopelvic injuries (33.3%) due to some patients sustaining multiple injuries.

^c Statistical tests are Fisher’s exact test.

^d Other category includes injuries to the pancreas, bladder, diaphragm and internal hematomas.

Table 5

Logistic regression analyses

Odds Ratio (95% Confidence Interval)

	Cervicothoracic injury	Intra-abdominopelvic injury
Age, years Male sex Driver’s seat position Seatbelt sign	1.02 (1.01-1.03) 1.60 (1.00-2.43) 0.90 (0.50-1.63) 1.63 (1.05-2.54)	1.01 (0.99-1.02) 1.17 (0.73-1.87) 0.71 (0.39-1.30) 1.21 (0.76-1.92)

General, the Department of the Army and Department of Defense or the U.S. Government.

Funding

None.

Conflicts of interest

None.

or thoracic “abrasion” or “ecchymosis” may have been consistent with seatbelt sign, but for the purposes of our study we required the specific use of this term to ascribe its presence to a patient. Our study simply re- lied upon medical record documentation and did not apply any pro- spective standardized mechanism for establishing the presence or absence of this examination finding. Furthermore, we were unable to specify the location of the seatbelt sign in our cohort (e.g., abdomen, chest, neck).

Our documentation of seatbelt sign had a further limitation in that we were unable to specify its location. Logically, we expect that most pa- tients with cervicothoracic injuries had seatbelt sign over their thorax or neck as opposed to the abdomen. However, we are unable to establish whether this is true given these data. Pending future studies of the im- portance of seatbelt sign anatomic location, we can only advise providers to have a high index suspicion for thoracic injuries among these patients with seatbelt sign in any location, to include the abdomen.

Another important limitation relates to our outcome measures. We documented any internal injuries in the thorax or abdomen for our out- come measures. We did not collect any data regarding the actions or medical interventions required as a result of these findings (e.g., hospi- talization, surgery). Our only patient-focused clinical outcome is that of mortality and we observed too few deaths to make any meaningful in- ferences regarding the prognostic implications of a seatbelt sign among patients who are status post MVC with airbag deployment. We furthermore did not measure any outcomes to quantify injury severity (e.g., injury severity score or acute injury score, etc.).

A final limitation of note is that we did not collect data on many po- tential confounders. Examples of potentially unmeasured confounders include airbag location (side versus frontal), passenger comorbidities, and body habitus. Our finding of an association between documented seatbelt sign and cervicothoracic injury held in a logistic regression analysis controlling for those study variables which we did collect in- cluding demographics and position within the vehicle during the MVC. It is impossible to know whether this finding would continue to hold in logistic regression controlling for additional confounders.

Conclusions

In conclusion, we find that among patients in the driver or front pas- senger seats during MVCs with airbag deployment and restraint by seatbelt, there was an association between documentation of a seatbelt sign and the subsequent diagnosis of a cervicothoracic injury. Of note, this association did not hold for intra-abdominopelvic injuries. Given these findings, we believe trauma care providers should have a high index of suspicion for thoracic injuries in these patients presenting with this physical examination finding.

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ajem.2017.09.011.

Disclaimers

The view(s) expressed herein are those of the author(s) and do not reflect the official policy or position of Brooke Army Medical Center, the

U.S. Army Medical Department, the U.S. Army Office of the Surgeon

References

1. Fantus RJ, Nance ML. NTDB data points: annual report 2015: regional trauma. Bull Am Coll Surg 2016;101:74-5.
2. Johnson NB, Hayes LD, Brown K, Hoo EC, Ethier KA, Centers for Disease C, et al. CDC National Health Report: leading causes of morbidity and mortality and associated behavioral risk and protective factors-United States, 2005-2013. MMWR Suppl 2014;63:3-27.
3. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United

   States, 2000. JAMA 2004;291:1238-45.

   McCarthy M. US traffic deaths show sharp increase. BMJ 2016;355:i5473.

4. McDonald NC. Trends in automobile travel, motor vehicle fatalities, and physical ac- tivity: 2003-2015. Am J Prev Med 2017;52:598-605.
5. National Highway Traffic Safety A. An analysis of the significant decline in motor ve- hicle traffic fatalities in 2008. Ann Emerg Med 2011;58:377-8.
6. Beck LF, Shults RA. seat belt use in States and territories with primary and secondary laws-United States, 2006. J Safety Res 2009;40:469-72.
7. Dinh-Zarr TB, Sleet DA, Shults RA, Zaza S, Elder RW, Nichols JL, et al. Reviews of ev- idence regarding interventions to increase the use of safety belts. Am J Prev Med 2001;21:48-65.
8. 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.
9. Cummins JS, Koval KJ, Cantu RV, Spratt KF. Do seat belts and Air bags reduce mortal- ity and injury severity after car accidents? Am J Orthop (Belle Mead NJ) 2011;40: E26-.
10. Cummins JS, Koval KJ, Cantu RV, Spratt KF. Risk of injury associated with the use of seat belts and air bags in motor vehicle crashes. Bull NYU Hosp Jt Dis 2008;66:290-6.
11. Doersch KB, Dozier WE. The seat belt syndrome. The seat belt sign, intestinal and mesenteric injuries. Am J Surg 1968;116:831-3.
12. Garrett JW, Braunstein PW. The seat belt syndrome. J Trauma 1962;2:220-38.
13. Hayes CW, Conway WF, Walsh JW, Coppage L, Gervin AS. Seat belt injuries: radiolog- ic findings and clinical correlation. Radiographics 1991;11:23-36.
14. Williams RF, Fabian TC, Fischer PE, Zarzaur BL, Magnotti LJ, Croce MA. Impact of

    airbags on a Level I trauma center: injury patterns, infectious morbidity, and hospital costs. J Am Coll Surg 2008;206:962-8 [discussion 968-9].

    Koisaari T, Leivo T, Sahraravand A, Haavisto AK, Sulander P, Tervo TMT. Airbag de- ployment-related eye injuries. Traffic Inj Prev 2017;18:493-9.

15. Shields RA, Rachitskaya A. Ocular trauma and airbag deployment. JAMA Ophthalmol

    2014;132:1245-6.

    Atkinson P, Hariharan P, Mari-Gowda S, Telehowski P, Martin S, Van Hoof J, et al. An under-hand steering wheel grasp produces significant injury risk to the upper ex- tremity during airbag deployment. Annu Proc Assoc Adv Automot Med 2002;46: 45-62.

16. Jernigan MV, Duma SM. The effects of airbag deployment on severe upper extremity injuries in Frontal automobile crashes. Am J Emerg Med 2003;21:100-5.
17. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med 2007; 147:573-7.
18. Kaji AH, Schriger D, Green S. Looking through the retrospectoscope: reducing bias in emergency medicine chart review studies. Ann Emerg Med 2014;64:292-8.
19. Bansal V, Conroy C, Tominaga GT, Coimbra R. The utility of seat belt signs to predict intra-Abdominal injury following motor vehicle crashes. Traffic Inj Prev 2009;10: 567-72.
20. Nishijima DK, Simel DL, Wisner DH, Holmes JF. Does this adult patient have a blunt intra-abdominal injury? JAMA 2012;307:1517-27.
21. Velmahos GC, Tatevossian R, Demetriades D. The “seat belt mark” sign: a call for in- creased vigilance among physicians treating victims of motor vehicle accidents. Am Surg 1999;65:181-5.
22. Laurence G, Grim R, Bell T, Carney D, Ahuja V. The impact of seatbelt use and airbag deployment on blunt thoracic aortic injury. Am Surg 2013;79:E335-.
23. Matthes G, Schmucker U, Lignitz E, Huth M, Ekkernkamp A, Seifert J. Does the frontal airbag avoid thoracic injury? Arch Orthop Trauma Surg 2006;126:541-4.
24. Monkhouse SJ, Kelly MD. Airbag-related chest wall burn as a marker of underlying injury: a case report. J Med Case Reports 2008;2:91.
25. April MD, Murray BP. cost-effectiveness analysis appraisal and application: an emer- gency medicine perspective. Acad Emerg Med 2017.
26. Sharma OP, Oswanski MF, Kaminski BP, Issa NM, Duffy B, Stringfellow K, et al. Clin- ical implications of the seat belt sign in blunt trauma. Am Surg 2009;75:822-7.
27. Abbas AK, Hefny AF, Abu-Zidan FM. Seatbelts and road traffic collision injuries. World J Emerg Surg 2011;6:18.
28. Chase J, Donaldson L, Duflou J, Gorrie C. Safety restraint injuries in fatal motor vehi- cle collisions. Forensic Sci Med Pathol 2007;3:258-63.
29. Markogiannakis H, Sanidas E, Messaris E, Koutentakis D, Alpantaki K, Kafetzakis A, et al. Motor vehicle trauma: analysis of Injury profiles by road-user category. Emerg Med J 2006;23:27-31.
30. Carter PR, Maker VK. Changing paradigms of seat belt and air bag injuries: what we have learned in the past 3 decades. J Am Coll Surg 2010;210:240-52.
31. Howes MK, Gregory TS, Hardy WN, Beillas PD. Kinematics of the thoracoabdominal contents under various loading scenarios. Stapp car crash J 2012;56:1-48.
32. Yoganandan N, Pintar FA, Gennarelli TA, Maltese MR. Patterns of abdominal injuries in frontal and side impacts. Annu Proc Assoc Adv Automot Med 2000;44:17-36.