An assessment of the impact of

pregnancy on trauma mortality

Preeti R. John, MD, MPH, Aki Shiozawa, MPH, MBA, Elliott R. Haut, MD, David T. Efron, MD, Adil
Haider, MD, MPH, Edward E. Cornwell III, MD, and David Chang, MPH, MBA, PhD, Baltimore, MD

Background. In the United States, trauma is the leading cause of maternal mortality and an important
source of maternal morbidity. Few studies have compared outcomes in injured pregnant women to their
nonpregnant counterparts. Some clinical literature regarding hormonal influences on outcomes after
trauma suggests a survival advantage in premenopausal women with higher estrogen levels. Given this,
as well as possible outcome differences as a result of physiologic changes that occur during pregnancy, we
tested the hypothesis that pregnant women have different outcomes after trauma compared with similarly
injured nonpregnant women in the same age groups.
Methods. We used data derived from 1.46 million patients listed in The National Trauma Data Bank
from 2001 to 2005, to query all injured patients between ages 12 and 49 years inclusive, and divided
them into 2 comparison groups: nonpregnant and pregnant women. We compared differences in outcome
after trauma between pregnant and nonpregnant women. Because the number of pregnant women was
small in comparison to the number of nonpregnant women, multivariate analysis after 1:3
(pregnant:nonpregnant) matching was attempted.
Results. Crude mortality rate comparisons and unadjusted logistic regression analyses both before and
after matching data reveal lower mortality rates in pregnant women. Multivariate logistic regression
analyses both before and after matching data also reveal lower mortality rates in pregnant women; but
this is statistically significant (P = .01) only after matching data.
Conclusion. Among women of similar age groups who are equivalently injured, those who are pregnant
exhibit lower mortality. These findings suggest that hormonal and physiologic differences during the gestation
period may play a role in outcomes following trauma in pregnant women. (Surgery 2011;149:94-8.)
From The Johns Hopkins Hospital, Baltimore, MD

INJURY remains the leading cause of mortality during

the first 3 decades of life in both males and females
in the United States, and the incidence of injuries in
women is increasing. In the U.S., trauma is the leading cause of maternal mortality and an important
source of maternal morbidity,1 but the true incidence and prevalence of trauma during pregnancy
is unknown. It has been estimated that trauma complicates 6--7% of pregnancies2 and that 0.3--0.4% of
pregnant women require hospitalization following
trauma.3 The more active lifestyles led by pregnant
women in todays society may put them at increased
risk of injury.4 Because the association between
trauma and pregnancy is relatively uncommon, it
is well suited for analysis using large, secondary
data sources.
P.R.J. and A.S. contributed equally to this manuscript.
Accepted for publication April 16, 2010.
Reprint requests: Preeti R. John, MD, Department of Surgery,
5C-125, VA Medical Center, 10 North Greene Street, Baltimore,
MD 21201. E-mail: preeti.john@va.gov.
0039-6060/$ - see front matter
2011 Mosby, Inc. All rights reserved.
doi:10.1016/j.surg.2010.04.019

94 SURGERY

Pregnancy is associated with anatomic and

physiologic changes, as well as with endocrinologic
changes (estrogen and progesterone are produced
by the placenta throughout the gestational
period). We thought it would be interesting to
assess how these changes affect outcomes after
injury and hypothesized that there might be a
difference in outcome/mortality between injured
pregnant women and nonpregnant women. The
effect of pregnancy on trauma outcomes in the
female population is largely unknown, as few
previous studies have compared outcomes in injured pregnant women to their nonpregnant
counterparts.
Literature regarding hormonal influences on
outcomes after trauma is abundant, yet findings
have been contradictory. In experimental models,
relatively high estrogen (and progesterone) levels
have been beneficial with respect to immunomodulatory and vasodilatory effects and ultimate outcome (survival) after traumatic injuries.5-8
Findings in clinical publications vary widely,
with some studies showing survival advantage in
premenopausal women.9-11 If a hormone dependent survival benefit does exist, then pregnant

Surgery
Volume 149, Number 1

women who have higher estrogen and progesterone levels might be expected to exhibit lower
mortality compared with similarly injured nonpregnant women.
Given the potential for sex hormone-based
outcome differences, as well as outcome differences as a result of physiologic changes that occur
during pregnancy, the hypothesis that pregnant
women have different outcomes after trauma compared with similarly injured nonpregnant women
in the same age groups was tested in this study.
The purpose of this study was to examine the
largest available trauma database, the National
Trauma Data Bank (NTDB), to compare outcomes
between pregnant and nonpregnant women sustaining equivalent injuries.
METHODS
The NTDB is the largest aggregation of trauma
registry data ever assembled, consisting of cases
from over 600 U.S. trauma centers in 45 states,
Puerto Rico and the District of Columbia. We used
data derived from 1.46 million patients listed in
the NTDB from 2001 to 2005, to query all injured
women between ages 12 and 49 years inclusive, and
divided them into 2 comparison groups: nonpregnant women versus pregnant women, based on
NTDB definition (urine human chorionic gonadotropin [HCG] positive). Burn patients were
excluded.
The primary outcome variable was in-hospital
death. Multivariate analysis adjusted for the following covariates: age (categorical), race, Injury
Severity Scale (ISS), mechanism of injury (penetrating versus blunt), insurance status, calendar
year, presence of shock (systolic blood pressure
[SBP] <90 mm Hg) and comorbidity. Subset analyses were performed to assess patients in the
following categories: patients in shock (SBP less
than 90 mm Hg), patients with severe injuries
overall (ISS >15), patients with severe (Abbreviated
Injury Scale [AIS] $3) head trauma, patients with
severe (AIS $3) abdominal trauma and younger
patients (<25 years).
Following multivariable logistic regression, we
then performed a matched analysis (1 case matched
with 3 controls) given the disproportionately small
number of pregnant women compared to nonpregnant women. Patients were matched on age category,
race, ISS, mechanism of injury, and insurance status.
Statistical analysis was performed using STATA
10.0 (STATA Corp., College Station, TX). A computerized matching process was used to enable the
matched cohort study design, using 3 controls

John et al 95

(nonpregnant women) for every case (pregnant

woman).
RESULTS
A total of 218,157 patients were identified in
this particular age group (12--49 years). Nonpregnant women (N = 214,394) accounted for 98.28%
of the patients and pregnant women (N = 3,763)
accounted for 1.72% of the patients. The mean age
of all patients was 29.8 years. There were 132,190
(66.2%) white, 37,069 (18.6%) black, and 18,548
(9.3%) Hispanic patients. A total of 116,461
(56.6%) patients had an ISS <9.
Comparison of characteristics of pregnant
versus nonpregnant women is presented in
Table I. Pregnant women were significantly younger than nonpregnant women, with mean
(median) ages of 25.6 (24) versus 29.9 (29), respectively. There was a higher proportion of minority groups among the pregnant women. Pregnant
patients had less severe injuries (ISS <9) than nonpregnant women. A total of 116,461 (56.6%) patients had an ISS <9, out of which 2,514 (77.0%)
were pregnant and 113,947 (56.2%) were nonpregnant women. Among a total of 36,897 patients
with ISS >15, 36,582 (18.1%) were nonpregnant
women, and only 315 (9.7%) were pregnant. Patients in the pregnant group had a significantly
lower portion of severe head injury, severe abdominal injury, and penetrating injury, compared with
those in the nonpregnant group. There was no significant difference between pregnant and nonpregnant women in terms of proportion of
patients with shock. Within the pregnant group
there were fewer patients with private insurance
and more with Medicaid or no insurance, versus
the nonpregnant group.
As depicted in Table II, the unadjusted mortality
for pregnant women was 1.2% whereas that for
nonpregnant women was 2.7% before matching
data. This is statistically significant (P < .001).
After matching data, the unadjusted mortality for
pregnant women was 1.1% whereas that for nonpregnant women was 2.0%. This trend (lower mortality for pregnant women) remains statistically
significant (P = .001).
On unadjusted logistic regression analysis, pregnant women are significantly less likely to die than
nonpregnant women (odds ratio [OR], 0.43;
P < .001), as shown in Table III. However, on multivariate analysis, controlling for the set of covariates described in the methods section, there was
no statistically significant difference between pregnant and nonpregnant women (OR, 0.80; P = .32).

96 John et al

Surgery
January 2011

Table I. Baseline characteristics, nonpregnant women versus pregnant women (before matching)
Nonpregnant (%)
Age
Systolic blood pressure
Patients with severe head injury
Patients with severe abdominal injury
Patients with penetrating injury
Patients with shock
Age (years)
12--17
18--24
25--29
30--34
35--39
40--44
45--49
Race
White, not of Hispanic origin
Black
Hispanic
Asian or Pacific Islander
Native American or Alaskan Native
Other
ISS category
<9
9--14
15--24
>25
Insurance
Private insurance
Medicaid/no charge/charity/self
Medicare
Others
Year
2001
2002
2003
2004
2005

Pregnant (%)

Total (%)

P value

29.9 10.1
126.2 25.6
11,454 (5.3)
6,267 (2.9)
12,350 (5.8)
7,933 (3.7)

25.6
122.3
93
80
121
144

6.3
25.4
(2.5)
(2.1)
(3.2)
(3.8)

29.8 10.1
126.1 25.6
11,547 (5.3)
6,347 (2.9)
12,471 (5.7)
8,077 (3.7)

<.001
<.001
<.001
.004
<.001
.68

31,204
53,532
25,700
23,923
25,918
27,851
26,266

(14.6)
(25.0)
(12.0)
(11.2)
(12.1)
(13.0)
(12.3)

196
1,706
903
567
279
97
15

(5.2)
(45.3)
(24.0)
(15.1)
(7.4)
(2.6)
(0.4)

31,400
55,238
26,603
24,490
26,197
27,948
26,281

(14.4)
(25.3)
(12.2)
(11.2)
(12.0)
(12.8)
(12.1)

<.001

130,623
36,186
17,792
3,982
1,688
5,916

(66.6)
(18.4)
(9.1)
(2.0)
(0.9)
(3.0)

1,567
883
756
112
50
142

(44.6)
(25.7)
(21.5)
(3.2)
(1.2)
(4.1)

132,190
37,069
18,548
4,094
1,738
6,058

(66.2)
(18.7)
(9.3)
(2.1)
(0.9)
(3.0)

<.001

113,947
52,089
21,323
15,259

(56.2)
(25.7)
(10.5)
(7.5)

2,514
438
175
140

(77.0)
(13.4)
(5.4)
(4.3)

116,461
52,527
21,498
15,399

(56.6)
(25.5)
(10.4)
(7.5)

<.001

75,989
64,429
4,072
19,222

(46.4)
(39.4)
(2.5)
(11.7)

1,075
1,517
42
276

(37.0)
(52.1)
(1.4)
(9.5)

77,064
65,946
4,114
19,498

(46.2)
(39.6)
(2.5)
(11.7)

<.001

38,076
43,636
43,949
42,684
46,049

(17.8)
(20.4)
(20.5)
(19.9)
(21.5)

469
565
755
792
1,182

(12.5)
(15.0)
(20.1)
(21.1)
(31.4)

38,545
44,201
44,704
43,476
47,231

(17.7)
(20.3)
(20.5)
(19.9)
(21.7)

<.001

Repeating the multivariate analysis on different

subgroups of patients, we found no statistically
significant difference in in-hospital deaths between
pregnant and nonpregnant women among
severely injured patients (ISS >15), patients with
severe head injuries, and patients in shock. However, among younger patients, there was a significantly decreased risk of death for pregnant
women compared to nonpregnant women (OR,
0.46; P = .04). There was a trend towards increased
risk of death for pregnant women with severe
abdominal trauma compared to nonpregnant
women (OR, 1.96; P = .12).
After matching cases and controls, however,
pregnant women are significantly less likely to die
than nonpregnant women, in both unadjusted
(OR, 0.57; P = .001) as well as adjusted analysis

(OR, 0.59; P = .01), as shown in Table IV. When

the multivariate analysis was repeated in different
patient subgroups after matching, the following
were noted: younger pregnant women and pregnant women sustaining blunt injuries were significantly less likely to die than their nonpregnant
counterparts. There was no statistically significant
difference among the following subgroups:
severely injured patients (ISS >15), patients with
severe head injuries, patients in shock, and
patients with severe abdominal injury.
DISCUSSION
This is the largest study we are aware of to date
that compares outcomes between pregnant and
nonpregnant women and the only one that uses
multivariate analyses (as well as the matched

John et al 97

Surgery
Volume 149, Number 1

Table II. Unadjusted mortality, nonpregnant women versus pregnant women (before and after matching)
Before matching
Alive
Dead
Total
After matching
Alive
Dead
Total

Nonpregnant (%)

Pregnant (%)

Total (%)

P value

208,700 (97.3)
5,694 (2.7)
214,394 (100)

3,719 (98.8)
44 (1.2)
3,763 (100)

212,419 (97.4)
5,738 (2.6)
218,157 (100)

<.001

10,989 (98.1)
219 (2.0)
11,208 (100)

3,694 (98.9)
42 (1.1)
3,736 (100)

14,683 (98.3)
261 (1.8)
14,944 (100)

.001

Table III. Odds of death for pregnant women compared with nonpregnant women (before matching)
All patients
Unadjusted logistic regression
Multivariate logistic regression*
Patient subgroups
Severe injury (ISS $15)
Severe head injury
Severe abdominal injury
Low blood pressure (systolic <90)
Younger patients (age <25)
Blunt injury

OR

CI

P value

0.43
0.80

0.320.58
0.511.24

<.001
.32

1.12
0.80
1.96
1.20
0.46
0.85

0.691.81
0.292.24
0.854.52
0.652.22
0.220.98
0.541.35

.63
.67
.12
.56
.04
.50

*Multivariate logistic regression: pregnant, race, age category, ISS category, insurance category, shock, comorbidity, mechanism and type of injury, year of
admission.

Table IV. Odds of death for pregnant women compared with nonpregnant women (after matching)
All patients
Unadjusted logistic regression
Multivariate logistic regression*
Patient subgroups
Severe injury (ISS $15)
Severe head injury
Severe abdominal injury
Low blood pressure (systolic <90)
Younger patients (age <25)
Blunt injury

OR

CI

P value

0.57
0.59

0.410.80
0.390.89

.001
.01

0.89
0.65
1.94
1.04
0.42
0.56

0.561.43
0.241.75
0.665.66
0.541.98
0.210.82
0.370.86

.65
.40
.23
.91
.01
.008

*Variables: pregnant, insurance category, shock, comorbidity, mechanism and type of injury, year of admission.

cohort study design) to compare survival differences between the 2 groups. Using the matching
process, we found that pregnant trauma patients
are approximately 40% less likely to die than their
nonpregnant counterparts.
On subgroup analysis, this survival benefit was
evident in younger women, suggesting a possible
additive beneficial effect of youth and pregnancy. Of
interest, there was no survival benefit in pregnant
women when severely injured patient subgroups
were compared (ISS >15, severe head injury, severe
abdominal injury, or patients in hypotensive shock),
suggesting that whatever advantage that pregnancy
may confer may be limited. Also of note is the trend
towards increased likelihood of death in pregnant

patients with severe abdominal injury, a finding

which may be related to placental abruption contributing to internal hemorrhage in this subgroup.
A previous study used the NTDB for the period
1994--2001 and analyzed outcomes in 1,195 pregnant trauma patients.12 The crude mortality rate
for pregnant patients who were injured in their
study was 1.4%, compared to 3.8% for nonpregnant patients (P < .001). Another study using a
smaller dataset found that pregnancy does not
increase maternal mortality from trauma, and
that the most frequent cause of death in injured
pregnant patients was head injury.13
Our study demonstrates a statistically significant
survival advantage among pregnant women. The

98 John et al

results raise more questions than can be answered,

and we can only speculate about reasons for this
apparent advantage. The physiologic adaptations
and hormonal changes during the gestational
period may contribute. During pregnancy there
is an increase in plasma volume and red blood cell
mass, and cardiac output increases by as much as
50% during the first trimester. These factors may
confer better organ perfusion, improved maternal
tolerance to hemorrhage and increased resistance
to the shock state in pregnant women.14
Estrogen and progesterone levels are higher in
pregnant women during all 3 trimesters as compared to nonpregnant women. Both estrogen and
progesterone have been shown, in animal models,
to confer a survival advantage in trauma and
hemorrhagic shock.5-8 We speculate that higher estrogen and progesterone levels may also play a role
in the apparent survival advantage.
We acknowledge several limitations to our study.
This is a retrospective study using data from a
population-based databank. We have no data about
actual hormone status of individual women; oral
contraceptives, hormonal supplementation, hysterectomy, and early menopause could have affected
hormone levels in the nonpregnant group. Nor do
we have data regarding trimester of pregnancy
(which affects hormone level). In the NTDB,
women are listed as pregnant if their urine HCG
was positive at the time of admission, which may
include molar (nonembryonic) pregnancies as well.
However, there are strengths to be noted. This is
the largest study (number of pregnant women:
3,763) from the most recent trauma database to
date, assessing outcomes in injured pregnant
women. Additionally, data were analyzed in our
study using the matched cohort study design in
addition to the more common multiple logistic
regression analysis. The computerized matching
process made a difference to the statistical results.
In studies such as this one involving large numbers
of patients where there is a disproportionate ratio
between cases and controls, an imbalance in sample sizes may mean thousands of controls for every
positive case, possibly introducing too much noise
in the regression models and reducing their power
to detect a difference. The matched cohort study
design takes this into account and reduces this
effect.

Surgery
January 2011

These results may further contribute to the

clinical gender dimorphism literature suggesting
a hormonal basis for improved survival in injured
women compared to men. We hope these data will
stimulate more detailed analyses of outcomes in
injured pregnant women. Future prospective
clinical studies should attempt to elucidate the
hormonal milieu of women (pregnant and nonpregnant) at the time of injury.
REFERENCES
1. Fildes J, Reed L, Jones N, Martin M, Barrett J. Trauma: The
leading cause of maternal death. J Trauma 1992;32:643-5.
2. Peckham CH, King RW. A study of intercurrent conditions
observed during pregnancy. Am J Obstet Gynecol 1963;87:
609-24.
3. Lavin JP, Polsky SS. Abdominal trauma during pregnancy.
Clin Perinatol 1983;10:423-38.
4. Shah KH, Simons RK, Holbrook T, Fortlage D, Winchell R,
Hoyt DB. Trauma in pregnancy: maternal and fetal outcomes. J Trauma 1998;45:83-6.
5. Mendelsohn ME, Karas RH. The protective effects of estrogen on the cardiovascular system. N Engl J Med 1999;340:
1801-11.
6. Knoferl MW, Diodato MD, Angele MK, Ayala A, Cioffi WG,
Bland KI, et al. Do female sex steroids adversely or beneficially affect the depressed immune responses in males after
trauma-hemorrhage? Arch Surg 2000;135:425-33.
7. Angele MK, Schwacha MG, Ayala A, Chaudry IH. Effect of
gender and sex hormones on immune responses following
shock. Shock 2000;14:81-90.
8. Kuebler JF, Jarrar D, Bland KI, Rue L III, Wang P, Chaudry
IH. Progesterone administration after trauma and hemorrhagic shock improves cardiovascular responses. Crit Care
Med 2003;31:1786-93.
9. Deitch EA, Livingston DH, Lavery RF, Monaghan SF, Bongu
A, Machiedo GW. Hormonally active women tolerate shocktrauma better than do men. A prospective study of over
4000 trauma patients. Ann Surg 2007;246:447-55.
10. George RL, McGwin G, Metzger J, Chaudry IH, Rue LW III.
The association between gender and mortality among
trauma patients as modified by age. J Trauma 2003;54:464-71.
11. Haider AH, Efron DT, Haut ER, Chang DC, Paidas CN,
Cornwell EE. Mortality in adolescent girls vs boys following
traumatic shock. An analysis of the National Pediatric
Trauma Registry. Arch Surg 2007;142:875-80.
12. Ikossi DG, Lazar AA, Morabito D, Fildes J, Knudson MM.
Profile of mothers at risk: an analysis of injury and pregnancy loss in 1,195 trauma patients. J Am Coll Surg 2005;
200:49-56.
13. Esposito TJ, Gens DR, Smith LG, Scorpio R, Buchman T.
Trauma during pregnancy. A review of 79 cases. Arch Surg
1991;126:1073-8.
14. Kissinger DP, Rozycki GS, Morris JA Jr, Knudson M, Copes
WS, Bass SM, et al. Trauma in pregnancy. Predicting pregnancy outcome. Arch Surg 1991;126:1079-86.