Motor vehicle accidents are common yet unexpected traumatic events. In 1997 there were more than 13 million motor vehicle accidents in the United States, resulting in 1.5 million serious injuries and more than 38,000 deaths (1). Recent studieshave shown substantial psychiatric morbidity after serious motor vehicle accidents, with rates of acute posttraumatic stress disorder (PTSD) ranging from 34% to 39% (2, 3). The risk for acute PTSD after a motor vehicle accident is 4.64 times greater for women than men (3). Epidemiological studies in the general population also indicate higher rates of PTSD in women than in men (4–7). In general, women are 2.38–2.49 times more likely to develop lifetime PTSD than men after exposure to similar traumas (7, 8).
Several studies have suggested risk factors for the development of PTSD that might explain the different rates of PTSD in men and women. People with a history of major depression are at greater risk for PTSD after all types of trauma (5) and after a serious motor vehicle accident (2). Prior trauma, prior PTSD, and prior major anxiety disorder other than PTSD are also significant risk factors for the development of PTSD after a motor vehicle accident (2, 3). In addition, passenger injury has been suggested as a predictor of PTSD after a motor vehicle accident (3). Those who experience peritraumatic dissociation at the time of a motor vehicle accident are 4.12 times more likely to develop acute PTSD and 4.86 times more likely to go on to chronic PTSD (3). Each of these risk factors may lead to gender differences in the rate of PTSD either because the risk factors occur at different rates in men and women or because of fundamentally different neurobiological or psychological factors related to these risk factors in women and men. Recent findings indicating improvement in PTSD symptoms in women taking sertraline but not in men taking the medication also suggest fundamental gender differences in PTSD (9, 10).
Few empirical studies have specifically examined these risk factors as possible explanations for the different rates of PTSD in men and women. In a large study of a health maintenance organization population, preexisting major depression or any anxiety disorder not including PTSD (examined as a single variable) increased vulnerability for lifetime PTSD in women but not in men after control for the type of trauma. However, this difference was related to the higher frequency of major depression or anxiety disorder among the women in this study (8). In contrast, in the National Comorbidity Study, after control for the type of trauma, a history of affective disorder predicted lifetime PTSD in women but not in men, whereas a history of anxiety disorder (not including PTSD) predicted lifetime PTSD in men but not in women (11). To our knowledge, no one has examined prior PTSD, peritraumatic dissociation, or passenger injury as possible explanations of the gender difference in PTSD.
To better understand gender differences in PTSD, we assessed a group of 122 subjects who had been in motor vehicle accidents. We report here on the following variables as possible explanations for the different rates of PTSD in men and women: prior trauma, PTSD, major depression, anxiety disorder, and peritraumatic dissociation; current peritraumatic dissociation; and passenger injury.
The subjects were drivers or passengers in serious motor vehicle accidents involving another passenger car, motorcycle, or truck within the past 2 weeks. Potential participants were at least 18 and not older than 65 years old and resided within a 40-mile radius of the regional trauma center of a major suburban hospital in a large metropolitan area. Two physicians systematically reviewed hospital records and excluded individuals with evidence of head trauma, coma, or organic brain syndrome identified by a neurological examination, mental status examination, or computerized tomography scan.
A full description of the study methodology has been previously published (3). The majority of the subjects were recruited from the regional trauma center. Twenty-eight percent were recruited through local police reports of severe motor vehicle accidents. Recruits from these sources did not differ from the trauma center recruits in demographic or dependent variables. Approximately half of the hospital recruits and 25% of recruits from the police records agreed to participate. The recruits and the refusers did not differ demographically except that the refusers were less educated. Of the total of 122 participants, 64 (52.5%) were men and 58 (47.5%) were women. The majority were white (N=93, 76.2%), were not married (N=76, 62.3%), had attended college (N=89, 73.0%), and reported an income greater than $20,000 per year (N=69, 56.6%). The mean age at the time of the accident was 35.6 years (SD=13.1). The men and women did not differ in age, education, marital status, or race. After completely describing the study and procedures to the subjects, we obtained written informed consent.
At 1 month postaccident we used the Structured Clinical Interview for DSM-III-R (SCID) (12), including the PTSD supplement (13), to assess lifetime and current PTSD, and the Peritraumatic Dissociative Experiences Questionnaire—Rater Version (14) to measure both present (immediately after the motor vehicle accident) and past peritraumatic dissociation. The Peritraumatic Dissociative Experiences Questionnaire is an eight-item scale that assesses eight peritraumatic symptoms. Clinicians rated subjects’ responses to symptom probes on a scale from 0 to 3 (0=inadequate information, 1=absent, 2=subthreshold, and 3=threshold). The presence of accident-related passenger injuries was determined (0=no, 1=yes) (3).
The SCID, PTSD supplement, and Peritraumatic Dissociative Experiences Questionnaire were administered (and tape-recorded) at the subject’s home by one of two senior licensed clinical social workers experienced in the use of the SCID. Half of the tapes were reviewed by a senior clinician (R.J.U.) on an ongoing basis to ensure consistent interview procedures and reliability. The average agreement between the two interviewers and the senior clinician on PTSD diagnoses and each of the PTSD symptoms was excellent (kappa>0.95, p<0.001). Kappas for the other diagnoses and for peritraumatic dissociative symptoms were similarly high. Diagnostic disagreements were resolved by consensus. To assess prior trauma and prior (lifetime) PTSD, individuals were asked if they had experienced as a child or an adult any events that met the DSM-III-R stressor criterion (criterion A: events outside the range of usual human experience) before their accident. Subjects were asked specifically about major earthquakes or floods, serious accidents, fires, physical assault, rape, seeing another person killed or dead, war and combat experience, and other types of disaster. The interviewer both probed and inquired specifically about different periods of the subject’s life. For positive answers, data on the type of event and time of exposure were recorded. Adjusting the symptoms in criteria B and D for DSM-III-R PTSD by using the DSM-IV criteria did not affect our results. We report DSM-IV symptoms in t1 to allow for convenience of comparison.
The significance level was set at 5% for all analyses. The effect of potential risk factors on the relationship of gender and PTSD at 1 month after the motor vehicle accident was evaluated by using multiple logistic regression and chi-square analyses. Odds ratio was defined as the ratio of the odds of developing the disease for persons with a risk factor versus those without the risk factor. Estimates of the odds ratios and their 95% confidence intervals (CIs) are reported. The Wald test was used to evaluate significance of any difference in the odds of developing the disease for persons with a risk factor versus those without the risk factor (or if odds ratio is different from 1). For analyses of differences in data on PTSD criteria sets (criteria B, C, and D symptom totals), we set significance at 5%; for analysis of data on individual PTSD symptoms, we used the Bonferroni correction (p<0.008) to control for experiment-wise error. All models examining risk factors included the primary variable and its gender interaction to identify possible gender differences. Statistical Analysis Software (SAS) (15) was used for statistical analysis.
Gender Differences in PTSD Symptoms
Overall, 42 subjects (34.4%) developed PTSD by 1 month after the motor vehicle accident (18.8% of the men [N=12] and 51.7% of the women [N=30]). The risk of developing PTSD at 1 month was 4.39 times greater in the women than in the men (95% CI=1.69–11.38, p=0.002), after control for other variables. Women did not differ from men in meeting the overall reexperiencing criterion (criterion B); however, women were at greater risk for two of the B criteria symptoms (t1). Women were 3.79 times more likely than men to report intense feelings of distress when in a situation similar to the motor vehicle accident and 5.16 times more likely to report a physical reaction to memories of the motor vehicle accident. Women were 4.71 times more likely than men to meet the overall avoidance/numbing criterion (criterion C). Examination of individual criterion C symptoms indicated that women were 3.75 times more likely than men to report avoiding thoughts, feelings, activities, or places associated with the motor vehicle accident; 3.65 times more likely to report loss of interest in significant activities; and 2.90 times more likely to experience a sense of a foreshortened future. Women were 3.83 times more likely than men to meet the overall arousal criterion (criterion D) and also more often reported trouble sleeping (odds ratio=2.52), difficulty concentrating (odds ratio=2.90), and being easily startled (odds ratio=4.63).
Gender Differences in Predictors of PTSD
To further examine the relationship of gender and PTSD 1 month after a motor vehicle accident, we used separate multiple logistic regression analyses to examine peritraumatic dissociation related to the motor vehicle accident, prior trauma, prior peritraumatic dissociation, prior PTSD, prior major depression, prior anxiety disorder not including PTSD (i.e., generalized anxiety disorder, panic disorder, social phobia, agoraphobia, and obsessive-compulsive disorder), and passenger injury. Because peritraumatic dissociation was related to race (3), we examined all models by adjusting for race. Women and men did not differ in the frequency of any of the risk factors except for passenger injury. Passenger injury was more frequent in women than in men (23 of 58 women [39.7%] and 13 of 64 men [20.3%]) (χ2=4.98, df=1, p<0.03).
Ninety-six subjects (78.7%) reported at least one peritraumatic dissociative symptom (mean=2.10 symptoms, SD=1.83, range=0–8) after the motor vehicle accident. Men and women had similar frequencies of peritraumatic dissociation (49 of 64 men [76.6%] and 47 of 58 women [81.0%]) (χ2=0.36, df=1, p=0.55). In a multiple logistic regression analysis, after adjustment for peritraumatic dissociation, gender was no longer a significant predictor of PTSD, but the interaction of gender and peritraumatic dissociation was a significant predictor (t2). We examined this interaction further by stratifying the gender groups to form subgroups with and without peritraumatic dissociation. Women with peritraumatic dissociation were 7.55 times more likely than men with peritraumatic dissociation to develop PTSD (Wald χ2=17.19, df=1, p=0.0001, 95% CI=2.90–19.64) (F1). Of those with peritraumatic dissociation, 59.6% of women but only 16.3% of men developed PTSD (χ2=19.14, df=1, p<0.001).
To further assess peritraumatic dissociation, we also examined past episodes of peritraumatic dissociation. Twenty-six subjects reported peritraumatic dissociative symptoms related to a trauma that occurred before the motor vehicle accident. After adjustment for prior peritraumatic dissociation, gender remained a significant predictor of acute PTSD. Women were 7.86 times more likely than men to develop PTSD at 1 month (Wald χ2=13.95, df=1, p=0.0002, 95% CI=2.66–23.2). However the interaction of gender and prior peritraumatic dissociative symptoms was not significant (Wald χ2=0.03, df=1, p=0.87, odds ratio=0.84, 95% CI=0.11–6.74), suggesting that the gender difference in 1-month PTSD was not due to prior peritraumatic dissociation.
Prior trauma and prior PTSD
Fifty-two individuals (42.6%) reported having experienced a traumatic event sometime during their lifetime, before the motor vehicle accident. After adjusting for the effects of prior trauma, gender remained a significant predictor of 1-month PTSD (Wald χ2=11.41, df=1, p=0.0007, odds ratio=10.61, 95% CI=2.7–41.77). However the interaction of gender and prior trauma was not significant (Wald χ2=0.37, df=1, p=0.54, odds ratio=0.57, 95% CI=0.09–3.51), indicating that the gender difference in 1-month PTSD was not due to prior trauma.
Thirteen subjects (10.7%) had a history of PTSD before their motor vehicle accident. Even after adjusting for prior PTSD, women had a 9.24-fold greater risk than men of 1-month PTSD (Wald χ2=15.80, df=1, p=0.0001, 95% CI=3.09–27.68). The interaction of gender and prior PTSD was not significant (Wald χ2=1.56, df=1, p=0.21, odds ratio=0.14, 95% CI=0.007–3.02), suggesting that the gender difference in 1-month PTSD was not due to prior PTSD.
Since women have been shown to be at high risk for major depression, which is a risk factor for PTSD, we also examined gender differences in 1-month PTSD after adjusting for prior major depression. Even after adjustment for prior major depression, women were at an 11.72 times greater risk of 1-month PTSD than men (Wald χ2=16.77, df=1, p=0.0001, 95% CI=3.61–38.08). The interaction of gender and prior major depression was not significant (Wald χ2=3.23, df=1, p<0.08, odds ratio=0.15, 95% CI=0.02–1.19), indicating that the gender difference was not due to prior major depression.
Prior anxiety disorder (not including PTSD)
We also examined gender differences in 1-month PTSD after adjusting for prior anxiety disorder not including PTSD. Even after adjusting for prior anxiety disorder, women were at an 8.22 times greater risk of 1-month PTSD than men (Wald χ2=15.83, df=1, p=0.0001, 95% CI=2.91–23.21). The interaction of gender and anxiety disorder was not significant (Wald χ2=1.10, df=1, p=0.29, odds ratio=0.2, 95% CI=0.01–4.05), indicating that the gender difference was not due to prior anxiety disorder.
We examined passenger injury as an explanation of gender differences in 1-month PTSD. Even after adjustment for passenger injury, women had a 5.01 times greater risk for PTSD than men (Wald χ2=7.24, df=1, p=0.007, 95% CI=1.55–16.22). The interaction of gender and passenger injury was not significant (Wald χ2=2.63, df=1, p=0.11, odds ratio=6.07, 95% CI=0.69–53.67), indicating that the gender difference was not due to passenger injury.
Previous studies of motor vehicle accidents have indicated that women are at greater risk of motor vehicle accident-related PTSD (2, 3). We examined the extent to which this gender difference might be explained by known risk factors for the development of PTSD. Few studies have examined the reasons why women are at greater risk of PTSD without the confound of type of trauma, which is often different in women and men and is known to affect rates of PTSD (6, 16). In this study we controlled for the type of trauma by examining only motor vehicle accident victims. We used standardized clinician assessments, another methodological strength of the study. Our results indicate that prior trauma, prior PTSD, prior major depression, prior anxiety disorder, and passenger injury do not explain gender differences in PTSD. However, peritraumatic dissociation occurring at the time of the motor vehicle accident does appear to substantially explain the higher rates of PTSD in women after a motor vehicle accident.
Kessler et al. (7) suggested that prior trauma or age at exposure to trauma might explain the gender differences in PTSD. Since the women and men in our study did not differ in age at the time of their motor vehicle accident, and age did not predict PTSD (3), age is not a likely explanation of the gender differences in PTSD in our study group. Our findings also indicate that the gender difference in acute PTSD related to a motor vehicle accident was not due to passenger injury or prior trauma. Bromet et al. (11), after controlling for the type of trauma, found that prior trauma did not explain the gender differences in PTSD. To our knowledge, no one has previously examined whether prior PTSD or prior peritraumatic dissociation could explain the gender difference. Our data indicate that neither of these trauma-related factors contribute to the gender difference in PTSD. However, these results must be interpreted cautiously because of the low rate of prior PTSD and prior peritraumatic dissociation in our study group.
Rates of major depression were similar for the men and women in our study group. Most often women in the general population have been found to be at greater risk for depression than men (7, 17). Since we found no difference, it may be that depressed women are less likely to drive or to be in a serious motor vehicle accident or, alternatively, that depressed men are more likely to be driving or in a serious accident. Regardless of these possibilities, our analyses indicate that neither prior major depression nor prior anxiety disorder not including PTSD explained the gender difference in rates of PTSD. This is similar to the finding of Breslau et al. (8) that the interaction between gender and preexisting major depressive disorder or anxiety disorder not including PTSD had no effect on gender differences in PTSD. In contrast, Bromet et al. (11), using data from the National Comorbidity Study, found that women with depression were at increased risk of PTSD but men were not, while a history of anxiety disorder predicted lifetime PTSD in men but not in women. Both of these studies were large-scale epidemiologic studies, rather than a more focused investigation of motor vehicle accidents, and differ greatly in design from the present study. The present study used prospective, current assessments with standardized clinician ratings, in contrast to the large-scale studies, which relied on subjects’ recall of past traumatic events and used trained nonclinician raters. These factors may contribute to these differences in results. In summary, two studies found prior depression to be unrelated to the gender differences in PTSD, and one study supported prior depression as part of the explanation for this difference. All three studies indicate that prior anxiety disorder cannot explain the higher rate of PTSD in women.
No previous studies have examined peritraumatic dissociation as a factor in gender differences in PTSD. Women and men in our study reported the same frequency of peritraumatic dissociative symptoms. It is important to note, however, that peritraumatic dissociative symptoms appeared to carry a different risk for PTSD in women and men, increasing the risk for PTSD significantly more in women. This finding may indicate that there are fundamental neurobiological differences in peritraumatic dissociation in women and men.
Gender differences in the specific symptoms of PTSD can indicate avenues for further study and can suggest specific targets for therapeutic interventions that may vary between women and men. For example, we found that although women did not differ from men in meeting the overall reexperiencing criterion for PTSD (criterion B), women were at greater risk of intense feelings of distress when in a situation similar to the motor vehicle accident and were more likely to report a physical reaction to memories of the motor vehicle accident. In one model, these might be the primary symptom differences in PTSD in men and women. The gender difference in meeting the overall avoidance criterion (criterion C) and the overall arousal criterion (criterion D) might be a result of a greater sensitivity of women to contextual and memory-linked arousal, which could sustain arousal states after an motor vehicle accident and lead to chronic avoidance. Replication of findings of gender differences in rates of specific PTSD symptoms would aid in developing individualized treatments. The relationship of peritraumatic dissociation to these specific symptom differences requires further study. Further studies of gender and PTSD may help define the gender differences in event-related neurobiology (e.g., how the outside world changes neurobiological function) and genetic contributions to brain function. Such studies are important to our understanding of fundamental neurobiology and to clinical care.
Received June 20, 2000; revision received Dec. 12, 2000; accepted Feb. 21, 2001. From the Department of Psychiatry and the Department of Preventive Medicine and Biometrics, Uniformed Services University of the Health Sciences; and the Department of Behavioral Medicine and Oncology, University of Pittsburgh Cancer Institute. Address reprint requests to Dr. Fullerton, Department of Psychiatry, Uniformed Services University of the Health Sciences, 4301 Jones Bridge Rd., Bethesda, MD 20814-4799; email@example.com (e-mail). Supported by NIMH grant MH-40106.
Percentage of Women and Men With and Without Peritraumatic Dissociation After a Motor Vehicle Accident Who Met DSM-III-R Criteria for PTSD 1 Month After the Accidenta
aGoodness of fit χ2=0.60, df=4, p=0.96. Model adjusted for race.