Journal of Traumatic Stress, Vol. 20, No. 4, August 2007, pp. 643 649 ( C 2007) Circumstances of Service and Gender Differences in War-Related PTSD: Findings From the National Vietnam Veteran Readjustment Study J. Blake Turner Department of Psychiatry, College of Physicians and Surgeons, Columbia University and New York State Psychiatric Institute, New York, NY Nicholas A. Turse Department of Sociomedical Sciences, Mailman School of Public Health, Columbia University, New York, NY BruceP.Dohrenwend Department of Psychiatry, College of Physicians and Surgeons, Columbia University, New York State Psychiatric Institute, and Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY Data from the National Vietnam Veteran Readjustment Study (NVVRS) revealed a prevalence of current posttraumatic stress disorder (PTSD) in female Vietnam Theater veterans half the size of the prevalence in their male counterparts. This stands in contrast to the elevated prevalence of PTSD in women obtained in general population surveys. This study undertakes further analyses of gender differences in the NVVRS and how these differences might be specified by the amount and type of exposure to war-zone stress. The findings indicate that male elevations in PTSD are limited to men who served under circumstances of high probable severity of war-zone stress exposure. When prewar demographic differences are controlled, male veterans in low-exposure circumstances display a level of PTSD prevalence substantially lower than female veterans. Epidemiological research has found that women are twice as likely as men in the general population to have a lifetime diagnosis of post-traumatic stress disorder (PTSD; Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). The prevalence of PTSD in male and female veterans who served in Vietnam shows a very different pattern. In describing findings from the National Vietnam Veterans Readjustment Study (NVVRS), Kulka and his col- leagues (1990) reported that male veterans who served in the Vietnam theater have nearly twice the prevalence of current PTSD as female Vietnam theater veterans (15.2% vs. 8.5%) precisely the opposite of the general population studies. One possible explanation for this striking difference is the marked contrast in the demographic character of the male and female service personnel who entered Vietnam. This research was supported by grant MH59309 from the National Institute of Mental Health and by grants from the Spunk Fund, Inc. The authors acknowledge Joan Furey for her valuable insights during the early stages of this work. William Schlenger and Tom Yager provided helpful comments on earlier drafts and David Schatzkamer provided valuable assistance in the preparation of the manuscript. Correspondence concerning this article should be addressed to: J. Blake Turner, New York State Psychiatric Institute, 1051 Riverside Drive, Box 78, New York, NY 10032. E-mail: jbt12@columbia.edu. C 2007 International Society for Traumatic Stress Studies. Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/jts.20245 643
644 Turner, Turse, and Dohrenwend Because they were mostly nurses (Kulka et al., 1990), female veterans were substantially older at the time they entered Vietnam compared to their male counterparts. Younger age is associated with increased risk for the development of PTSD following exposure to violent terrorist acts (Schlenger et al., 2002; Trautman et al., 2002), so it is reasonable to surmise that this risk will extend to late adolescents and young adults exposed to combat. The nurses were also substantially more educated than the male veterans, on average, as most would have completed 3 years in hospital schools of nursing (Norman, 1990). Higher levels of education have been found to be inversely associated with the probability of PTSD in situations of high probable exposure to severe stressors (Dirkzwager, Bramsen, & van der Ploeg, 2005; Parslow, Jorm, & Christensen, 2006). Another quite intuitive explanation centers on the differences between male and female veterans exposure to the type of experiences likely to elicit PTSD. Because they were not directly involved in combat operations, female veterans were comparatively unlikely to experience life-threatening events that satisfied Criterion A for PTSD in the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (American Psychiatric Association, 1987). On the other hand, a substantial proportion of the male veterans served in service support roles that were at least equally removed from combat circumstances. Furthermore, these rear echelon soldiers would not have had the extensive exposure to the bloody and grotesque consequences of combat characteristic of the experience of nurses. In this article, we attempt to interpret the gender differences in PTSD in the NVVRS in the context of variation in the circumstances of service of the male veterans who served in Vietnam. The NVVRS supplemented the selfreport data from its survey instrument with a substantial amount of information abstracted from the veterans personnel files. Using this information, the analyses in this study address the following questions: First, to what extent are the differences between male and female veterans in PTSD prevalence explained by gender differences in prewar demographic risk factors? Second, is the elevated PTSD prevalence relative to female veterans limited to men who served under circumstances of high probable exposure to war-zone stress? Finally, do female veterans, primarily nurses, have a higher prevalence of PTSD than men who served in service-support, low-exposure circumstances? METHOD Sample The NVVRS was conducted by the Research Triangle Institute (RTI; Research Triangle Park, NC) in response to a Congressional mandate to investigate PTSD and other psychological problems in U.S. Vietnam veterans. This study included intensive interviews with a probability sample of 1,200 men and 432 women who served in the Vietnam theater during the war years of 1964 1975. This sample was drawn from military records with strong efforts to get good completion rates and with careful checks on and adjustments for nonresponse. The completion rate for the Vietnam theater veterans was 83%. Female veterans were substantially oversampled, as were Blacks and Hispanics among the men. Measures Posttraumatic stress disorder. All veterans in the study were administered the Mississippi Scale for Combat- Related PTSD (Keane, Caddell, & Taylor, 1988) as part of the survey. The original investigators conducting the NVVRS wanted to use this self-report information to make PTSD prevalence estimates. To this end, they constructed a variable for the predicted probability of being a PTSD case, generating these predicted values from regression models that accurately predicted clinical diagnoses (as a function of Mississippi Scale scores and a number of other variables) in a subsample of veterans who received in-depth psychiatric assessments. These predicted probabilities provided the basis for the prevalence estimates published from the NVVRS (Kulka et al., 1990), and serve as the dependent variable in this study. Prewar risk factors. The RTI investigators abstracted extensive information from the personnel files of the veterans
Gender Differences in War-Related PTSD 645 selected into the NVVRS sample. This information included the veterans racial background, their dates of birth, and the dates of their tour(s) in Vietnam. The race/ethnic categorizations were modified, where appropriate, based on self-identification and interviewer observation. For the analyses in this article, age at entry to Vietnam was determined by comparing each veteran s date of birth to the date of entry to Vietnam as given in the military records, and then was collapsed into five categories, ranging from 19 or younger at the low end to up to 25 and older. Level of premilitary education was also provided in the military records, but it was inconsistently recorded. Thus, analysis of premilitary education in this study relied on self-report and separated veterans into four groups: those who failed to complete high school (no diploma); those who completed high school and went no further; those who had some college education; and those who graduated from college with a bachelor s degree or higher. Probable severity of war-zone stress exposure among the male veterans. In addition to demographic variables, the military record information available to us included a number of military service variables that are related to the probable severity of war-zone stress exposure. These include the Military Occupational Specialty (MOS) of the veteran, the starting and end dates of Vietnam service, and the unit (from division down to the company-level) in which the veteran served. Military occupational specialties vary substantially in terms of likelihood of combat involvement; some military divisions are explicitly combat or non-combat; and the likelihood of exposure to combat and to combat consequences is associated with the time of service since the amount of fighting and the casualty rates varied substantially over the 11-year history of the war. Unlike measures of exposure based on self-report, these Military Historical Measures (MHMs) are contemporaneous and independent of respondent recall. This is important because there is substantial evidence indicating that retrospective reports of exposure to war-zone stress are strongly influenced by PTSD symptom levels (Roemer, Litz, Orsillo, Ehlich, & Friedman, 1998; Southwick, Morgan, & Nicolau, 1997). These three measures were combined into a composite MHM that differentiated male theater veterans into groups with low (25.9%), moderate (63.4%), and high (10.7%) probable severity of exposure to war-zone stress in a manner described in an earlier study (Dohrenwend et al., 2004). We have been able to validate these groupings using information compiled on most of the approximately 58,000 U.S. veterans killed in action (KIA) during the Vietnam War (Coffelt, Arnold, & Argabright, 2002). Over 30% of the veterans in the high-exposure category had 10 or more KIA in their company during their tour in Vietnam. The corresponding figures for the moderate and low categories were 7.3% and 2.9%. In contrast, 71.4% of the veterans in the low-exposure category had no KIA at all in their company during the period they were in Vietnam. The veterans in the original high-exposure category who also had 10 or more KIA in their company during their period of service were placed into a fourth category identified as having a very high probable severity of war-zone stress. The resulting four-category MHM composite is the measure of severity of war-zone stress exposure used in this study. Previous analyses have shown this measure to be strongly related to self-reported war-zone stress exposure, and have demonstrated a very strong dose-response relationship between this MHM and PTSD (Dohrenwend et al., 2006). Data Analysis Initial analyses compared linear regression models in which gender differences in PTSD were estimated with and without controls for race/ethnicity, age at entry to Vietnam, and premilitary education. A separate set of models similarly assessed the role of race and ethnicity, premilitary education, and age of entry to Vietnam on gender differences in PTSD, but separated the male veterans into probable warzone exposure levels based on the four-category composite MHM described above. The only missing data for any of these analyses were on the dependent variable 35 of the 1,632 cases were missing on the algorithm for the predicted
646 Turner,Turse,andDohrenwend Table 1. Premilitary Demographics and Rates of Current Posttraumatic Stress Disorder (PTSD; Total Vietnam Theater Sample) Male % Female % Current PTSD 15.2 8.5 Race/Ethnicity White 81.3 93.8 Black 11.2 2.1 Latino 5.4 3.2 Other 2.1 0.9 Premilitary education Without high school diploma 23.5 0.7 High school graduate/no college 50.1 43.3 Some college 20.0 41.9 College graduate (4-year) 6.4 14.0 Age of entry to Vietnam 19 28.5 1.0 20 20.6 1.9 21 13.4 11.1 22 24 18.6 43.8 25+ 19.0 42.1 Differs from percentage of women, p <.05. probability of PTSD, largely because of too many missing responses to items on the Mississippi Scale. These 35 cases were removed from analysis. Because of the complex sampling design in which the probability of selection into the NVVRS varied substantially across groups of veterans (in particular, across race and ethnic groups), the data were weighted back to the population of Vietnam theater veterans from which they were sampled. Further, to account for this complex design, we used the software package SUDAAN (Shah, Barnwell, & Bieler, 1997) to obtain Taylor-series estimates of the standard errors. RESULTS Table 1 compares male and female veterans in the Vietnam theater sample. The top row displays current PTSD rates based on the RTI diagnostic algorithm and previously reported in Kulka et al. (1990). As noted previously, the rate of current PTSD in men who served in the Vietnam theater was nearly twice that of the women. As the remainder of the table shows, a number of characteristics likely to be related to risk for PTSD were not equally distributed by gender in Vietnam veterans. In each case, characteristics related to greater risk of PTSD were more prevalent among men. Racial and ethnic minority groups, particularly Blacks, were represented to a far greater degree among the male theater veterans. Furthermore, as expected, female veterans had substantially greater educational attainment than did their male counterparts, and were substantially older at the start of their Vietnam service. Given that the female veterans were advantaged relative to men on the prewar PTSD risk factors, it may be that the elevated PTSD rates in the male veterans were attributable to these advantages. Table 2 displays the results of regression analyses in which the predicted probability of PTSD was regressed on each of the variables displayed in the previous table, and then finally on all of the prewar risk factors together. The coefficient for the variable, male in each column represents the difference between male and female veterans in the mean of the predicted probability of PTSD. For example, in Model 1, the coefficient of 0.067 indicates that the unadjusted elevation in current PTSD prevalence among males was 6.7%. The remainder of the analyses in the table indicates that this difference was partially explained by differences in racial and ethnic distribution (Model 2), was reduced by more than half when adjusted for premilitary education (Model 3), and was almost completely explained by differences in the age at entry to Vietnam (Model 4). With all of the premilitary risk factors in the model, the PTSD rate was actually 2.0% lower among men than among women. Table 3 displays the results of regression analyses that again compared male and female PTSD probabilities, but this time the male veterans were divided into those with low, moderate, high, and very high probabilities of war-zone stress exposure as measured by our composite military-historical measure. The coefficient for each exposure group in the table represents the mean differences in the predicted probability of PTSD between this group and the women veterans. The control variables for Models 2 through 5 were the same as those in Table 2, though the coefficients are not presented.
Gender Differences in War-Related PTSD 647 Table 2. The Role of Premilitary Risk Factors in Gender Differences in Posttraumatic Stress Disorder Among Vietnam Veterans Model 1 Model 2 Model 3 Model 4 Model 5 B SE B SE B SE B SE B SE Gender a Female Male 0.067 0.015 0.055 0.160 0.026 0.140 0.012 0.016 0.020 0.160 Race/ethnicity b White Black 0.075 0.022 0.062 0.024 Latino 0.147 0.028 0.126 0.029 Other 0.222 0.097 0.255 0.089 Premilitary education c Without high school diploma High school graduate/no college 0.068 0.031 0.071 0.030 Some college 0.141 0.030 0.119 0.031 College graduate (4-year) 0.211 0.027 0.150 0.029 Age of entry to Vietnam d 19 20 0.041 0.035 0.031 0.034 21 0.113 0.034 0.091 0.034 22 24 0.120 0.031 0.076 0.032 25+ 0.154 0.029 0.139 0.029 a Represented in the model with a dummy variable scored 1 for men and 0 for women. b Represented in the model using three dummy variables with White being the contrast category. c Represented in the model using three dummy variables, Without high school diploma being the contrast category. d Represented in the model using four dummy variables with 19 being the contrast category. Significantly different from contrast group, p <.05. Table 3. Elevations in Male Veteran Posttraumatic Stress Disorder Rates at Different Levels of War-Zone Exposure Relative to Female Veterans a Model 1 Model 2 Model 3 Model 4 Model 5 B SE B SE B SE B SE B SE Women b Low-exposure men 0.036 0.027 0.028 0.027 0.005 0.026 0.023 0.026 0.042 0.027 Moderate-exposure men 0.056 0.017 0.047 0.017 0.018 0.016 0.009 0.017 0.020 0.018 High-exposure men 0.116 0.037 0.102 0.038 0.070 0.037 0.054 0.037 0.012 0.037 Very high-exposure men 0.229 0.058 0.201 0.057 0.182 0.055 0.153 0.056 0.105 0.054 a Model 1 gives the gross comparison of men at each exposure level to women, Model 2 controls race/ethnicity, Model 3 controls premilitary education, Model 4 controls age at entry to Vietnam, Model 5 controls all of the above. b Contrast category. Significantly different from female veterans, p <.05. From an examination of Model 1 in Table 3, it appears that male elevations in rates of PTSD existed even among those veterans with a relatively low probability of war-zone stress exposure. As the remaining results show, however, greater minority representation, lower levels of education, and the comparative youth of the male veterans effectively explained these elevations. Indeed, when all three of these factors were controlled, men at the lowest level of
648 Turner,Turse,andDohrenwend exposure exhibited a prevalence of current PTSD more than 4% lower than the female veterans. Although not quite statistically significant, this difference is comparable in magnitude to that obtained in the National Comorbidity Study from a representative national household sample (Kessler et al., 1995). In contrast, even though race/ethnicity and education were strongly related to probable exposure (for example, 8.0% of Black male veterans were in the highest exposure category, and 20.4% were in the top two categories; for White veterans the comparable figures were 2.4% and 10.4%, respectively), the highest exposure category retained an elevated prevalence of PTSD relative to women even when all other risk factors were controlled (10.5% higher). Thus, the elevated rates of PTSD in male veterans appear to be the result both of the greater prevalence of prewar risk factors among the men and of their greater exposure to war-zone adversity. Because of the qualitative differences between nurses and male combat soldiers in the nature of their exposures, we decided to further restrict the comparison to male medical personnel and female nurses. Most male personnel with a medical MOS, however, were field medics and hence substantially exposed to combat. When we repeated the analyses in Table 3 for nurses and male medical personnel only, the men in the high- and very high-exposure categories had a prevalence of PTSD 32.0% higher than the female nurses and this difference reduced only to 27.5% in the fully adjusted model. In contrast, medical personnel in low and moderate exposure circumstances did not differ in prevalence from the nurses. DISCUSSION The original NVVRS investigators found a twofold elevation in rates of PTSD among male relative to female Vietnam theater veterans. The analyses presented here show that, net of prewar risk factors, men with low probable severity of exposure had lower rates of PTSD than women a finding consistent with those from nonveteran samples. Elevations among male veterans were limited to those who served under circumstances of high exposure. The primary exposures to war-zone stressors for women serving in Vietnam did not involve engagement with the enemy. If they were nurses, as were the vast majority, they were exposed to varying degrees to the consequences of combat casualties in the form of death, life-threatening injury, and disfigurement. The general severity of the wounds and injuries witnessed by nurses was greater during the Vietnam War than during any previous war in U.S. history. Due, in large part, to new rapid medical evacuation methods, many horribly wounded soldiers who would have died on the battlefield in previous wars survived, but were left maimed and permanently disabled (Appy, 2003; Norman, 1990). In addition to these adverse witnessing experiences, nurses had the responsibility of providing medical care for severely wounded soldiers and, in the war setting, often had to intervene in life or death situations in a manner generally reserved for physicians in civilian life (Freedman & Rhoads, 1987). Further, this responsibility fell on nurses who were, on average, quite inexperienced upon arriving in Vietnam. According to the Assistant Chief of the Army Nurse Corps, 60% of Army nurses had less than 2 years nursing experience and of that 60%, most had less than 6 months experience (Marshall, 1987). In a further refinement to the analyses, the elevation in PTSD prevalence among highly exposed males was found to be even greater when the sample was restricted to medical personnel. This may reflect the additional exposure of medics to the bloody and grotesque consequences of combat because of their medical duties. Alternatively, having to treat wounded soldiers while under fire may have actually exposed them to a greater degree of life threat than generally experienced by their comrades in the same unit. Two important limitations to our study need to be discussed. First, risk factors for PTSD are included in the algorithm that created it. Specifically, Hispanic ethnic status and the presence of a self-reported traumatic event are among the variables included to adjust for discrepancies between the self-report measure and clinical diagnoses of PTSD. The adjustment for Hispanic status is less of a problem because race/ethnicity is part of our analysis only as a potential confounder of the relationship between gender
Gender Differences in War-Related PTSD 649 and PTSD. Further, the results did not change when the two minority groups were combined and the adjustment made in this way. More serious is the inclusion of traumatic event (self-reported and not necessarily war-related) in the algorithm because this builds in an association between events and the predicted probability of PTSD. Fortunately, our military historical measure of exposure is independent of self-report and is essentially an ecological assessment of probable severity of exposure given the time and circumstances of the veteran s military assignment. Nonetheless, to the extent that this is correlated with endorsement of traumatic events, caution is warranted. Another limitation of the analyses presented here is that, to date, we have no measure of exposure for the female veterans in Vietnam comparable to our MHMs for the men. We have been working to develop these measures for nurses primarily around the concept of casualty load. It will be important to continue this archival work to understand the etiology of PTSD in female veterans of the Vietnam War and, more generally, to uncover potentially traumatic war-zone stressors that are not explicitly attached to combat experiences. REFERENCES American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author. Appy, C. (2003). Patriots: The Vietnam War remembered from all sides. New York: Viking. Coffelt, R. D., Arnold, R. J., & Argabright, D. (2002). An electronic database of combat area causalities. Washington, DC: National Archives and Records Administration. Dirkzwager, A. J. E., Bramsen, I., & van der Ploeg, H. M. (2005). Factors associated with posttraumatic stress among peacekeeping soldiers. Anxiety, Stress & Coping, 18, 37 51. Dohrenwend, B. P., Neria, Y., Turner, J. B., Turse, N., Marshall, R., Lewis-Fernandez, R., et al. (2004). Positive tertiary appraisals and post-traumatic stress disorder in U.S. male veterans of the war in Vietnam: The roles of positive affirmation, positive reformulation, and defensive denial. Journal of Consulting and Clinical Psychology, 72, 417 433. Dohrenwend,B.P.,Turner,J.B.,Turse,N.A.,Adams,B.G., Koenen, K. C., & Marshall, R. (2006). The Psychological risks of Vietnam for U.S. veterans: A revisit with new data and methods. Science, 313, 979 982. Freedman, D., & Rhoads, J. (1987). Nurses in Vietnam: The forgotten veterans (p. 164). Austin, TX: Texas Monthly Press. Keane, T. M., Caddell, J. M., & Taylor, K. L. (1988). Mississippi Scale for Combat-Related Posttraumatic Stress Disorder: Three studies in reliability and validity. Journal of Consulting and Clinical Psychology, 56, 85 90. Kessler, R. C., Sonnega, A., Bromet, E., Hughes, M., & Nelson, C. B. (1995). Posttraumatic stress disorder in the National Comorbidity Survey. Archives of General Psychiatry, 52, 1048 1060. Kulka, R. A., Schlenger, W. E., Fairbank, J. A., Hough, R. L., Jordan, B. K., Marmar, C. R., et al. (1990). Trauma and the Vietnam generation: Report of findings from the National Vietnam Veterans Readjustment study. New York: Brunner-Mazel. Marshall, K. (1987). In the Combat Zone: Vivid personal recollections of the Vietnam War from the women who served there. New York: Viking. Norman, E. M. (1990). Women at war: The story of fifty military nurses who served in Vietnam. Philadelphia: University of Pennsylvania Press. Parslow, R. A., Jorm, A. F., & Christensen, H. (2006). Associations of pre-trauma attributes and trauma exposure with screening positive for PTSD: Analysis of a community-based study of 2,085 young adults. Psychological Medicine, 36, 387 395. Roemer, L., Litz, B., Orsillo, S., Ehlich, P. J., & Friedman, M. J. (1998). Increases in retrospective accounts of war-zone exposure over time: The role of PTSD symptom severity. Journal of Traumatic Stress, 11, 597 605. Schlenger, W. E., Caddell, J. M., Ebert, L., Jordan, B. K., Rourke, K. M., Wilson, D., et al. (2002). Psychological reactions to terrorist attacks: Findings from the National Study of Americans Reactions to September 11. Journal of the American Medical Association, 228, 581 588. Shah, B. V., Barnwell, B. G., & Bieler, G. S. (1997). SUDAAN users manual. Research Triangle Park, NC: Research Triangle Institute. Southwick, S. M., Morgan, C. A., & Nicolau, A. L. (1997). Consistency of memory for combat- related traumatic events in veterans of Operation Desert Storm. American Journal of Psychiatry, 154, 173 177. Trautman, R., Tucker, P., Pfefferbaum, B., Lensgraf, S. J., Doughty, D. E., Buksh, A., et al. (2002). Effects of prior trauma and age on posttraumatic stress symptoms in Asian and Middle Eastern immigrants after terrorism in the community. Community Mental Health Journal, 38, 459 474.