The Utility of the Personality Assessment Inventory in the Assessment of Posttraumatic Stress Disorder in OEF/OIF/OND Veterans

Abstract

We examined the use of the Personality Assessment Inventory (PAI) in a small sample of 47 U.S. military veterans of the conflicts in Iraq and Afghanistan. Approximately half of the sample met criteria for posttraumatic stress disorder (PTSD) based on the Clinician-Administered PTSD Scale. PAI profiles were compared between the PTSD and non-PTSD groups. The PTSD group had clinically significant scores (≥ 70T) on the PAI for 5 clinical scales (anxiety, anxiety-related disorders, depression, paranoia, and schizophrenia) and 10 clinical subscales consistent with the typical symptom picture for PTSD. Effect size correlations (r) between scales and diagnosis group membership were large (r ≥ .5) for several scales that reflect PTSD symptoms and for the PTSD LOGIT function. In a receiver operating characteristics curve analysis, the PTSD LOGIT function and the Traumatic Stress Subscale both demonstrated good diagnostic utility (areas under the curve > .80).

Keywords

personality assessment inventory posttraumatic stress disorder veterans

Posttraumatic stress disorder (PTSD) has a particularly high prevalence among military personnel and veterans. Lifetime prevalence of PTSD in the United States is estimated at 3.5% to 5.7% (Kessler, Chiu, Demler, & Walters, 2005; Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). Prevalence among veterans, however, appears to be twice as high, with one major study estimating the rate at 13.8% (Tanielian & Jaycox, 2008). In particular, PTSD is among the most commonly diagnosed conditions among veterans of recent conflicts in the Middle East, including Operations Enduring Freedom (OEF), Operation Iraqi Freedom (OIF), and Operation New Dawn (OND) (Epidemiology Program, 2015) with one study reporting that 21.8% of OEF/OIF/OND veterans treated by VA Medical Center (VAMC) facilities met criteria for PTSD (Seal et al., 2009).

PTSD is a complex disorder, with symptoms across different dimensions. It is also characterized by a high degree of comorbidity, which complicates the issue of accurate assessment. Studies suggest that 75% of individuals with PTSD meet criteria for at least one additional psychological disorder (Kessler et al., 2005). Depressive disorders, substance use disorders (SUDs), and anxiety disorders are the most commonly comorbid (Brady, Killeen, Brewerton, & Lucerini, 2000). One study of veterans with PTSD reported rates of 48.1% comorbidity with depression, 21.9% comorbidity with SUDs, 13.9% comorbidity with anxiety disorders, and 8.8% comorbidity with schizophrenia and bipolar disorders (Trivedi et al., 2015).

Because making an accurate diagnosis of PTSD can be difficult, there is a need for psychometrically sound tools to assess this complex disorder. Comprehensive personality assessments have been used by clinicians and researchers to better characterize, assess, and diagnose PTSD. The Personality Assessment Inventory (PAI; Morey, 1991) has emerged as a useful instrument for assessing PTSD. Given that the PAI has a range of scales across four categories (validity, clinical, interpersonal, and treatment), it is well-suited for this task.

The PAI (Morey, 1991) was developed within a construct validation framework, and scales were constructed using a rational/empirical process. The PAI profile includes 22 scales falling into four categories: validity scales (Inconsistency [ICN], Infrequency [INF], Negative Impression [NIM], and Positive Impression [PIM]), clinical scales (Somatic Complaints [SOM], Anxiety [ANX], Anxiety-Related Disorders [ARD], Depression [DEP], Mania [MAN], Paranoia [PAR], Schizophrenia [SCZ], Borderline Features [BOR], Antisocial Features [ANT], Alcohol Problems [ALC], and Drug Problems [DRG]), treatment considerations scales (Aggression [AGG], Suicidal Ideation [SUI], Stress [STR], Nonsupport [NON], and Treatment Rejection [RXR]), and two interpersonal scales (Dominance [DOM] and Warmth [WRM]). Many of these scales are comprised of component subscales.

The subscales of the PAI, which address more specific dimensions of each scale’s overarching construct, can be particularly useful in the assessment of PTSD, as they cover a broad range of dimensions (depressive symptoms, anxiety, hypervigilance, and others) with which PTSD symptoms and other responses to trauma coincide, and lend a degree of specificity to a diagnosis that is often made in the presence of comorbid disorders. Morey (1996) and Calhoun, Collie, Clancy, Braxton, and Beckham (2010) have discussed the ways that various subscales address specific aspects of the complex symptom structure of PTSD. Trauma exposure and re-experiencing symptoms are assessed by the traumatic stress subscale (ARD-T). Sleep disturbances are addressed in the physiological symptoms of depression subscale (DEP-P). Hyperarousal is reflected by the physiological anxiety (ANX-P), hypervigilance (PAR-H), irritability (MAN-I), and affective instability (BOR-A) subscales, as well as the aggression subscales (AGG-A, AGG-P, and AGG-V). Numbing symptoms, manifested as feeling distant from others, are reflected in the WRM scale and social detachment subscale (SCZ-S). Avoidance symptoms are addressed in the affective depression subscale (DEP-A).

The array of PAI clinical scales and subscales covers a broad range of clinical syndromes, permitting the detection of co-occurring disorders as well. The most frequently comorbid disorders (depression, SUDs, anxiety disorders) are all addressed by PAI scales, although the overlap between symptoms of PTSD and symptoms of anxiety and depression underscore the complexity of this assessment task.

Research on the utility of the PAI for assessing PTSD has typically involved testing differences between PTSD and non-PTSD groups on each scale of the profile. These studies have suggested that PTSD and non-PTSD groups show differences in mean scores on several PAI scales, including those assessing depression, anxiety, anxiety-related disorders, and somatic complaints (Cherepon & Prinzhorn, 1994; Drury et al., 2009; McDevitt-Murphy, Weathers, Adkins, & Daniels, 2005; McDevitt-Murphy, Weathers, Flood, Eakin, & Benson, 2007; Morey, 1991). Prior studies of veterans specifically have suggested that those with PTSD may be differentiated from those without PTSD on the Traumatic Stress subscale of the Anxiety-Related Disorders scale (ARD-T), as well as on several other scales and subscales reflecting the symptom domains of PTSD (Calhoun et al., 2010; Mozley, Miller, Weathers, Beckham, & Feldman, 2005). Overall, the extant literature has shown support for the use of the PAI as a tool in the characterization, assessment, and diagnosis of PTSD in veteran and civilian samples.

Morey (1991) included participants who had a diagnosis of PTSD in his original work on the PAI, and Cherepon and Prinzhorn (1994) examined the PAI profiles of adult female abuse survivors. Both studies found that traumatized participants had clinically significant elevations (defined in the manual as greater than 70T) on the ARD, ANX, and DEP full scales as well as the ARD-T, Affective Anxiety (ANX-A), Physiological Anxiety (ANX-P), and Affective Depression (DEP-A) subscales. Additionally, Cherepon and Prinzhorn found that the largest group difference between the traumatized and nontraumatized groups was on the ARD-T subscale. Holmes, Williams, and Haines (2001) compared 3 groups of motor vehicle accident survivors: those with PTSD, those with acute stress disorder, and those with subclinical symptoms. The PTSD group’s scores were significantly different from the other groups on the SOM, ANX, ARD, and DEP scales and the Negative Relationships (BOR-N), Social Detachment (SCZ-S), and Thought Disorder (SCZ-T) subscales.

Another study examined the PAI profiles of adult women who were classified as either PTSD or non-PTSD based on the Clinician-Administered PTSD Scale (CAPS; Blake et al., 1995). Results showed significant group differences on several clinical scales (ANX, DEP, ARD, SOM, PAR, BOR, and SCZ), one validity scale (NIM), as well as the NON and RXR treatment scales. The largest group differences, however, were found to be on the DEP-P and the ARD-T subscales (McDevitt-Murphy et al., 2005). This study also reported that the ARD-T and DEP-P subscales demonstrated excellent diagnostic utility and that correlations between PAI scales and four CAPS symptom clusters provided additional validity evidence. The ARD-T subscale has shown further discriminant validity in distinguishing PTSD from depression and social phobia in a sample of college students with mixed civilian trauma exposure (McDevitt-Murphy et al., 2007). Previous studies have established good diagnostic utility of the ARD-T subscale in nonveteran samples, with values of sensitivity and specificity greater than or equal to values of .79 and .77, respectively (Calhoun, Boggs, Crawford, & Beckham, 2009; McDevitt-Murphy et al., 2005). However, no study to date has examined the diagnostic utility of the ARD-T subscale in a sample composed purely of veterans.

The PAI has also shown good utility in the assessment of veterans with PTSD. One study examined the mean PAI profile of male combat veterans of the Vietnam War, Gulf War, Korean War, and World War II. This study showed clinically significant elevations on the DEP, SOM, ANX, ARD, SCZ, and NIM scales (Mozley et al., 2005). The ARD-T subscale was the highest point in the mean score profile for veterans with PTSD and correlated moderately with several established measures of PTSD. Mozley et al. also reported that subgroups of PTSD-diagnosed veterans with and without comorbid major depressive disorder differed on the DEP, ANX, and WRM scales.

Calhoun et al. (2010) examined a sample of veterans of Vietnam, Korea, World War II, Iraq, and Afghanistan, and Van Voorhees et al. (2014) examined a sample of veterans of Vietnam, Iraq, and Afghanistan. Both these studies found clinically significant elevations on the SOM, ANX, ARD, DEP, and SCZ scales. Calhoun et al. noted that veterans of earlier eras such as World War II and the Korean War have profiles showing relatively lower scores than other veterans on PAR, SCZ, ANT, AGG, and higher scores on WRM, suggesting that veterans of these past conflicts show a less hostile, less aggressive, and less isolated presentation than veterans of more recent wars. This difference could be due to the nature of these earlier conflicts or the difference in time since combat for these cohorts at time of assessment. Further research on veterans of the recent era is required to substantiate this as a consistent trend in PAI profiles of veterans of OEF/OIF/OND.

Samples of veterans with PTSD have typically differed from nonveteran samples in having higher overall scores on the PAI clinical scales and moderate elevations on treatment and interpersonal scales. These results provide a reference point for future research examining the utility of the PAI for assessing PTSD, although more work is needed. No study to date has reported on between-groups effect sizes on all PAI scales, comparing veterans of the OEF/OIF conflicts with and without PTSD. Such research is needed to determine whether the characteristic PTSD profile varies by population.

The PAI may also be used to derive scores on a set of LOGIT functions, in which subscale values are combined and weighted in an actuarial decision rule formula. The PTSD formula was derived from a linear LOGIT analysis of Morey’s original PAI clinical standardization sample, which contained a considerable proportion of veterans (40%). This decision rule reflects the probability that an individual meets criteria for a specific diagnosis when compared with other members of the standardization sample (Morey, 1991, 1996). The PTSD LOGIT function uses the ARD-T, ANX-P, DEP-P, SCZ-T, and BOR-A subscales, which correspond to different aspects of the broad constellation of PTSD symptoms, and are typically elevated in other samples of trauma survivors and PTSD-diagnosed individuals. In a community-based sample of women, a LOGIT function cut score of .05 showed a sensitivity of .82 and a specificity of .72 when compared to the CAPS (Calhoun et al., 2009). The LOGIT function also showed sensitivity in a sample comprised of individuals randomly selected from the PAI standardization sample combined with a group of veterans diagnosed with PTSD, correctly identifying 83% of those diagnosed with PTSD (Calhoun, Earnst, Tucker, Kirby, & Beckham, 2000). Because the LOGIT function has shown some diagnostic utility in nonveteran populations, further validation of its utility among veterans could serve to add a degree of confidence in diagnosis with a measure that addresses the multidimensional nature of PTSD.

The aim of the present study was to investigate the use of the PAI in a sample of veterans with OEF/OIF/OND combat exposure, comparing those with and without a diagnosis of PTSD. Based on the theoretical symptom structure of PTSD, the PTSD LOGIT function and consistent features of PAI profiles seen in previous veteran samples, we expected that the PTSD mean PAI profile would show clinically significant elevations on all DEP subscales, as well as the ANX-A, ANX-P, ARD-T, SCZ-T, SCZ-S and BOR-A subscales. Based the PTSD LOGIT function, and informed by previous studies’ diagnostic group differences, we hypothesized that the PTSD group would be most substantially differentiated from the non-PTSD group on the ANX-A, ANX-P, ARD-T, DEP-A, DEP-P, and SCZ-T subscales. We also aimed to investigate the diagnostic utility in this sample of the PTSD LOGIT function as well as the ARD-T subscale.

Method

Participants

We began with a sample of 58 veterans who completed the PAI as part of an extensive assessment battery. Participants with invalid PAI profiles were eliminated from this group by checking PAI profiles for completion and INF and ICN validity scale elevations. As per PAI interpretation guidelines (Morey, 1991), any scales or subscales with less than 80% item completion were not included. Additionally, any profiles with ICN T scores greater than or equal to 73 or INF T scores greater than or equal to 75 were not included. Following the application of these exclusion criteria, we retained a sample of 47 OEF/OIF veterans which included 37 men (78.7% of the sample) and 10 women (21.3%). Participants in this study ranged in age from 22 to 66 years (M = 37.2; SD = 11.2) and a majority identified as either White (n = 24; 51.1%) or Black (n = 22; 46.8%). About half of the veterans sampled (n = 24; 51.1%) completed a single overseas deployment, while a slightly smaller percentage of veterans (n = 21; 44.7%) completed multiple deployments. Twenty-three veterans (48.9% of the sample) met diagnostic criteria for PTSD, based on the CAPS. A chi-square test was conducted to determine whether those with multiple deployments were more likely to be diagnosed with PTSD, and this result was not significant.

Procedure

After review and approval of this study by the Institutional Review Boards of both the Memphis VAMC and The University of Memphis, veterans were recruited through medical clinics at the VAMC. Veterans were approached and invited to participate in a study investigating adjustment and health following combat. Eligibility criteria included having served at least one combat deployment as part of Operation Iraqi Freedom, Operation Enduring Freedom or Operation New Dawn (OEF/OIF/OND). Interested participants were scheduled for two assessment appointments where they completed a battery of measures. Structured interviews were administered in the first session, and these were conducted by doctoral students in clinical psychology who had been trained on the use of the CAPS and who had used the CAPS in at least one prior study. The PAI was completed in the second session. Both sessions typically occurred within a 1-week span. Participants were compensated $20 for each of the two assessment sessions.

Measures

Clinician Administered PTSD Scale

The CAPS (Blake et al., 1995) is a 30-item semi-structured clinical interview designed to assess the 17 symptoms of PTSD in DSM-IV (American Psychiatric Association, 2013). For each item, interviewers rate the frequency and intensity on separate 5-point scales ranging from 0 to 4. Items with a frequency score of at least 1 and an intensity score of at least 2 are rated as positive for the presence of a symptom, per the Frequency ≥1/Intensity ≥2 (or “F1/I2” scoring rule). The criteria for PTSD are met if there is a positive rating for symptoms consistent with the DSM-IV algorithm of 1 re-experiencing symptom, 3 numbing and avoidance symptoms, and 2 hyperarousal symptoms. The CAPS has shown consistency across raters and testing occasions, and has shown excellent convergent and discriminant validity, diagnostic utility, and sensitivity to clinical change (Weathers, Keane, & Davidson, 2001). The CAPS demonstrated excellent internal consistency in the present sample (α = .96).

Personality Assessment Inventory

The PAI (Morey, 1991) is a 344-item self-administered, objective inventory of adult personality that assesses multiple domains of personality and psychopathology, and provides information on critical clinical variables. The PAI is comprised of 22 nonoverlapping full scales (4 validity scales, 11 clinical scales, 5 treatment consideration scales, 2 interpersonal scales). Responses are recorded on 4-point graduated scales as false, somewhat true, mainly true, or very true. Scores are calculated by converting raw scores from each scale and subscale to T scores using a conversion table provided in the PAI manual. The PAI has shown good test–retest reliability, with coefficients ranging from .73 to .86 (Morey, 1996). Convergent and discriminant validity has been established with various psychometric instruments relating to constructs assessed by all the scales (Morey, 1996). Subsequent research has confirmed good internal consistency in samples of veterans with PTSD (Van Voorhees et al., 2014), extratest validity in inpatient substance abuse settings (Hopwood, Baker, & Morey, 2008), convergent validity in a federal inmate sample (Patry & Magaletta, 2014), and reliability and consistent factor structure in a neuropsychological sample (Busse, Whiteside, Waters, Hellings, & Ji, 2014). The internal consistency in the present sample ranged from moderate to strong for all scales (α = .69 to α = .95).

PAI PTSD LOGIT Function

The PAI PTSD LOGIT function is an equation that uses differential weighting of five different PAI subscales (ARD-T, ANX-P, DEP-P, SCZ-T, and BOR-A) to determine the probability of a PTSD diagnosis based on profile comparisons between PTSD and non-PTSD individuals; scores range from 0 to 1.0. The LOGIT value for any given individual is calculated using the following formula: Probability of diagnosis = exp[2(x_dx − 5)]/{1 + exp[2(x_dx − 5)]}, where “exp” is the exponential of the bracketed expression, and x_dx = .051(ARD-T) + .001(ANX-P) + .001(DEP-P) + .006(SCZ-T) − .010(BOR-A) + 2.45 (Morey, 1996).

Data Analysis

Data screening was conducted in accordance with recommendations from Tabachnick and Fidell (2006). Missing items on the PAI were mean-substituted using the average of the scale score, provided that at least 80% of the items in a scale were available. CAPS and PAI T scores were checked for skewness and kurtosis; no transformations were required. Next, descriptive statistics were computed for our sample. Correlation effect sizes were computed to compare veterans with and without PTSD on all clinical scales and subscales in the PAI as well as LOGIT function values. A receiver operating characteristic curve analysis was conducted for the ARD-T and LOGIT function values to determine optimal cut scores for PTSD diagnosis.

Results

The mean profile for the PTSD group was characterized by elevations in the clinically significant range (≥ 70T) on ANX, ARD, DEP, PAR, SCZ, and on several of the component subscales. The highest full-scale elevation was observed for DEP (M = 79.3T). The non-PTSD group showed a mean profile characterized by scores in the normal range for all scales and subscales. Table 1 shows T score means, standard deviations, and effect sizes (r_pb) for all PAI scales and subscales.

Table 1.

Means, Standard Deviations, and Between-Group Correlation Effect Sizes (r_pb) for Personality Assessment Inventory T Scores and LOGIT Function Scores.

Scales	Groups
	PTSD (n = 23)		Non-PTSD (n = 24)		Effect size (r)
	M	SD	M	SD	Effect size (r)
Validity Scales
Inconsistency (ICN)^a,b	49.8	8.2	53.7	8.0	−.24
Infrequency (INF)^a	54.2	7.8	50.3	9.8	.22
Negative Impression (NIM)	68.4	17.6	54.5	11.1	.44**
Positive Impression (PIM)^a,b	39.1	10.4	48.9	11.4	−.42**
Clinical Scales
Somatic Complaints (SOM)	69.9	16.0	57.4	10.7	.43**
Conversion (SOM-C)	68.6	16.6	55.5	10.1	.44**
Somatization (SOM-S)	68.2	15.8	56.0	12.1	.41**
Health Concerns (SOM-H)	65.5	13.6	57.5	11.0	.31*
Anxiety (ANX)	71.6	15.0	54.1	12.2	.55**
Cognitive (ANX-C)^a	66.6	14.5	51.3	10.5	.53**
Affective (ANX-A)	70.1	15.5	53.7	13.4	.50**
Physiological (ANX-P)	72.1	15.8	56.8	12.1	.49**
Anxiety Related Disorders (ARD)	74.9	13.7	54.3	12.7	.62**
Obsessive-Compulsive (ARD-O)	61.6	10.0	53.0	12.5	.36*
Phobias (ARD-P)^a	63.1	14.9	48.7	11.5	.49**
Traumatic Stress (ARD-T)	78.7	13.7	56.9	13.0	.64**
Depression (DEP)	79.3	15.7	58.5	12.6	.60**
Cognitive (DEP-C)^a	72.5	18.2	54.5	14.6	.49**
Affective (DEP-A)	77.0	15.9	55.5	13.0	.60**
Physiological (DEP-P)	74.3	10.2	60.6	11.5	.54**
Mania (MAN)	57.0	8.9	53.8	9.6	.18
Activity Level (MAN-A)	50.2	9.2	49.8	8.8	.02
Grandiosity (MAN-G)^a	46.6	9.6	53.3	10.5	−.32*
Irritability (MAN-I)	68.5	10.2	55.1	11.5	.53**
Paranoia (PAR)	71.9	13.7	59.9	12.9	.42**
Hypervigilance (PAR-H)	73.7	11.3	61.1	11.3	.50**
Persecution (PAR-P)^a	63.0	11.4	56.6	11.1	.28
Resentment (PAR-R)	67.6	15.4	56.9	16.1	.33*
Schizophrenia (SCZ)	71.2	16.8	57.3	12.4	.44**
Psychotic Experiences (SCZ-P)	53.4	14.3	47.4	8.1	.26
Social Detachment (SCZ-S)^a	71.8	16.5	59.5	15.7	.36*
Thought Disorder (SCZ-T)	70.9	16.5	58.2	13.5	.40**
Borderline Features (BOR)	69.2	13.6	55.1	10.6	.51**
Affective Instability (BOR-A)	72.9	12.5	56.4	11.0	.59**
Identity Problems (BOR-I)^a	65.9	14.6	53.8	11.2	.43**
Negative Relationships (BOR-N)^a	67.4	14.1	56.0	13.1	.40**
Self-Harm (BOR-S)	53.5	14.5	48.9	9.7	.19
Antisocial Features (ANT)	56.0	12.9	54.5	10.9	.07
Antisocial Behaviors (ANT-A)	53.8	11.9	54.4	10.8	−.03
Egocentricity (ANT-E)^a	52.8	12.0	50.7	10.5	.10
Stimulus Seeking (ANT-S)	58.3	15.8	55.7	13.1	.09
Alcohol Problems (ALC)^c	56.7	14.0	50.9	11.2	.23
Drug Problems (DRG)^d	54.6	19.4	50.5	8.4	.14
Treatment Scales
Aggression (AGG)^a	68.4	11.8	55.0	12.7	.48**
Aggressive Attitude (AGG-A)	65.3	13.7	52.1	12.8	.45**
Verbal Aggression (AGG-V)^a	62.0	10.2	55.2	10.2	.32*
Physical Aggression (AGG-P)^a,b	69.9	13.3	55.4	13.7	.48**
Suicidal Ideation (SUI)	57.3	19.3	50.8	9.4	.21
Stress (STR)	65.5	13.1	54.1	11.7	.43**
Nonsupport (NON)	69.5	17.1	55.5	12.2	.43**
Treatment Rejection (RXR)	36.8	10.0	47.2	9.1	−.49**
Interpersonal Scales
Dominance (DOM)	52.3	10.0	52.8	13.0	−.02
Warmth (WRM)	34.7	10.9	42.3	11.6	−.33*
LOGIT	.21	.18	.04	.05	.55**

Note. Posttraumatic stress disorder (PTSD) diagnostic status based on F1/I2 CAPS scoring rule.

PTSD (n = 22). ^bNon-PTSD (n = 23). ^cPTSD (n = 21). ^dPTSD (n = 20).

p < .05. **p < .01.

Large differences, defined as effect size correlations (r) of .5 or greater (Rosenthal, Rosnow, & Rubin, 2000) between the PTSD and non-PTSD groups, were found for the ANX, ANX-C, ANX-A, ARD, ARD-T, DEP, DEP-A, DEP-P, Irritability (MAN-I), BOR, and BOR-A scales, with ARD-T being the largest (r_pb = .64). Moderate effect sizes (r) of .3 to .5 (Rosenthal et al., 2000) were found for the SOM scale, all SOM subscales, ANX-P, Obsessive Compulsive (ARD-O), Phobias (ARD-P), Cognitive Depression (DEP-C), Grandiosity (MAN-G), PAR, all PAR subscales, SCZ, SCZ-S, SCZ-T, Identity Problems (BOR-I), BOR-N, the AGG scale, all AGG subscales, STR, NON, RXR, and WRM. Small effect sizes (r) of .1 to .3 (Rosenthal et al., 2000) were found for MAN, Psychotic Experiences (SCZ-P), Self-Harm (BOR-S), Egocentricity (ANT-E), ALC, DRG, and SUI.

We also calculated PTSD LOGIT function scores for each participant and found a large and significant effect size between groups (r_pb = .55). Receiver operating characteristic analysis of the PTSD LOGIT function scores yielded an area under curve of .85 (standard error [SE] = .05; 95% confidence interval [CI] = .75-.96). The optimal cut score for the LOGIT function was determined to be .044. This cut score showed a sensitivity of .83 and a specificity of .71. Receiver operating characteristic analysis of the ARD-T T scores yielded an area under curve of .87 (SE = .05; 95% CI = .76-.97). The optimal cut score for ARD-T was determined to be 63.5T. This cut score showed a sensitivity of .91 and a specificity of .75.

Discussion

We administered the PAI to a small sample of veterans from a VAMC who had completed combat deployments as part of OEF/OIF/OND as a preliminary effort to characterize the mean PAI profile among OEF/OIF/OND veterans with PTSD. Approximately half of the sample met criteria for PTSD using a structured interview. The mean PAI profile for the PTSD group showed clinically significant elevations (≥ 70T) on subscales that address the diverse symptom constellation of PTSD. These subscales are ANX-A, ANX-P, ARD-T, all DEP subscales, PAR-H, SCZ-S, SCZ-T, and Affective Instability (BOR-A). The present sample’s clinically significant ARD-T score is consistent with all extant research on use of the PAI in PTSD assessment. ANX-A, ANX-P, the DEP subscales, SCZ-T, and BOR-A have shown clinically significant or at least moderate elevations (≥ 60T) in veteran samples (Calhoun et al., 2010; Mozley et al., 2005) as well as in community-based samples of PTSD-diagnosed women (McDevitt-Murphy et al., 2005), undergraduate students (McDevitt-Murphy et al., 2007), the PAI clinical standardization sample (Morey, 1996), and child abuse survivors (Cherepon & Prinzhorn, 1994). The only exception was Calhoun et al.’s (2009) community-based sample of women, whose SCZ-T score was not moderately elevated. These subscale elevations have been relatively consistent across studies, suggesting that they represent stable features of the PTSD PAI profile, regardless of population. PAR-H has shown some degree of elevation throughout the extant literature, but has not consistently cleared 60T in any population aside from veterans of conflicts after World War II and Korea, lending support to Calhoun et al.’s (2010) notion that veterans of more recent conflicts present differently. SCZ-S elevations over 60T have only been consistently observed in veteran samples. While the reason for this discrepancy is unclear, it could indicate an exacerbation of numbing symptoms and resultant interpersonal difficulties in the veteran population. These veteran-specific interpersonal difficulties are further supported by the fact that WRM scores in our sample and other veteran samples have consistently shown means less than 40T and NON scores over 60T, trends which are not consistently observed in the nonveteran samples, further indicating a more emotionally distant and socially isolated presentation for this population.

Further evidence for the presence of heightened difficulty within the interpersonal realm for the veteran population was found in the AGG scale. In the present sample, all AGG subscales were at least moderately elevated. As in other veteran samples, the full AGG scale T score for the PTSD group was in the moderate-high range (≥ 65T). Calhoun et al., 2010; Crawford, Calhoun, Braxton, & Beckham, 2007; Mozley et al., 2005; Van Voorhees et al., 2014), whereas civilian samples have consistently shown mean scores below 60T for PTSD groups. This differential elevation in the AGG scale between veteran and nonveteran samples suggests that this PAI scale may have clinical utility in detecting concomitant difficulties with aggression among individuals with PTSD, which is important in light of research indicating that anger and aggression are frequent problems experienced by veterans with PTSD (Calhoun et al., 2002). The non-PTSD group in our sample was within the same range as nonveteran, non-PTSD samples (Drury et al., 2009; McDevitt-Murphy et al., 2005; McDevitt-Murphy et al., 2007), suggesting that the elevation in the AGG scale in veterans with PTSD is not related to having been in combat or the military in general, but rather speaks to a difference in aggressive tendencies that may be associated with PTSD among veterans. This elevation of aggression levels specific to the subpopulation of veterans with PTSD is also supported by prior research showing that veterans with PTSD have higher aggression levels than combat veterans without PTSD (Beckham, Feldman, Kirby, Hertzberg, & Moore, 1997; Begic & Jokic-Begic, 2001; Calhoun et al., 2002; Crawford et al., 2007; Freeman & Roca, 2001; McFall, Wright, Donavan, & Raskind, 1999) as well as other psychiatric patients without PTSD (McFall, Fontana, Raskind, & Rosenbeck, 1999).

The present sample’s largest between-groups effect size was on the ARD-T subscale, replicating a consistent finding across other PTSD samples (Cherepon & Prinzhorn, 1994; McDevitt-Murphy et al., 2005). This is noteworthy given that the ARD-T scale does not include questions reflecting the full range of PTSD symptoms, but emphasizes the unique aspects of PTSD such as re-experiencing symptoms as well as sense of having been changed by a past traumatic event (Morey, 1991), and that in most prior studies, the comparison groups have been comprised of trauma survivors who may be demonstrating some symptoms of PTSD short of the full criteria. Prior research has demonstrated good convergent validity with other measures of PTSD and discriminant validity from measures of anxiety and depression (McDevitt-Murphy et al., 2005). Thus, the present findings provide additional evidence that the ARD-T scale is a valuable aspect of the PAI in terms of helping to inform a diagnosis (but it should be emphasized that this subscale alone would not be sufficient to warrant a PTSD diagnosis).

Other large effect sizes for ANX and DEP subscales were consistent with previous research comparing PTSD and non-PTSD groups (Drury et al., 2009; Holmes et al., 2001; McDevitt-Murphy et al., 2005; Mozley et al., 2005). The large effect sizes seen for ANX-A and ANX-C are consistent with the hyperarousal and intrusive memories associated with PTSD. Hyperarousal symptoms could also be further reflected in the large effect sizes for MAN-I and PAR-H. The large effect size seen for the DEP-P subscale could reflect physiological symptoms caused by sleep disturbances and nightmares that are associated with intrusion symptoms (Calhoun et al., 2010). The large effect size observed for the DEP-A scale likely reflects the considerable anhedonia that occurs with PTSD (Calhoun et al., 2010). Scales consistent with the PTSD symptom picture differentiated PTSD participants from non-PTSD in the present sample, indicating that higher T scores in the veteran samples are due not just to the effects of combat exposure or military experience, but to PTSD.

The present findings add to the small but growing body of PAI-PTSD studies, and together these studies suggest that samples of veterans generate higher overall mean PAI profiles than nonveterans with PTSD. It should be noted that veterans tend to score higher than nonveterans on other measures of PTSD as well. For example, the PTSD Checklist (Weathers, Litz, Herman, Huska, & Keane, 1993) has consistently demonstrated that higher cut scores are more appropriate in veteran samples (Forbes, Creamer, & Biddle, 2001; Weathers et al., 1993). It is possible that the potential for compensation-seeking in the VA system is one factor contributing to these higher scores. It is also possible that combat-related PTSD may differ in severity and symptom presentation, relative to other forms of PTSD.

The notion that veterans of more recent conflicts may be more hostile and isolated than their counterparts from prior conflicts was suggested by Calhoun et al. (2010) and is echoed in the present findings by clinically significant differences (5 points or more) on some scales between the PTSD group in the present study and Calhoun et al.’s (2010) World War II/Korea PTSD group. The present sample showed mean T scores on PAR and AGG that were both at least 8 points higher than Calhoun et al.’s World War II/Korea group, and our WRM T scores were 5 points lower than their counterparts from those previous conflicts. Additionally, the mean PAI profile of the present sample’s PTSD group and that of Calhoun et al.’s OEF/OIF group were very similar overall.

The present analyses suggested diagnostic utility for the LOGIT function and ARD-T scale consistent with other research on discriminating between PTSD and non-PTSD individuals. Calhoun et al.’s (2009) prior study revealed similar rates of sensitivity and specificity. Additionally, the present sample’s rate of correctly identified PTSD cases was nearly identical to Calhoun et al.’s (2000) veteran sample. ARD-T subscale T scores used in previous studies (Calhoun et al., 2009; McDevitt-Murphy et al., 2005) showed comparable rates of sensitivity and specificity as well. These findings suggest that both the ARD-T and the LOGIT function offer some utility in identifying PTSD cases within the veteran population.

One surprising finding in this literature has been the consistent lack of clinically elevated scores on the scales relating to substance abuse (ALC and DRG), despite the fact that SUDs are highly comorbid with PTSD (Brady et al., 2000). This is true in civilian and veteran samples. Interestingly, an earlier study compared PAI profiles from a group of individuals with PTSD to a group with depression, a group with social phobia, and a well-adjusted comparison group (McDevitt-Murphy et al., 2007). That study showed little difference between these groups on the ALC and DRG scales, although a separate publication examining differences in substance use patterns in this sample found that the groups were differentiated on other measures of substance misuse (McDevitt-Murphy, Murphy, Monahan, Flood, & Weathers, 2010). The ALC and DRG scales have demonstrated validity for assessing alcohol and drug problems, respectively, in an inpatient sample (Parker, Daleiden, & Simpson, 1999) and in a correctional sample (Patry, Magaletta, Diamond, & Weinman, 2011). This suggests that perhaps the ALC and DRG scales are detecting severe patterns of substance misuse associated with dependence, but perhaps are less sensitive to the hazardous substance use patterns exhibited by individuals with PTSD.

Limitations of the present study include the use of a small sample, constraining our ability to generalize from this group. The sample was comprised nearly exclusively of men, and it will be important for future work in this area to provide insight into how PAI profiles may differ between male and female veterans. Regrettably, we did not collect data on whether veterans in our sample were involved in conflicts prior to OEF/OIF/OND, and so we are unable to parse out the possible effect of prior combat. Additionally, the effect of comorbid diagnoses, trauma type, and trauma severity were not explored in this study but would likely be a fruitful area of future research. Finally, we used the DSM-IV version of the CAPS, and it will be important to examine PAI profiles of individuals diagnosed using the DSM-5 criteria for PTSD. Further study comparing PAI profiles generated by groups of combat veterans with PTSD and by other specific trauma groups might shed light on the nature of differences in the presentation of PTSD across these populations.

In sum, the present findings suggest that the PAI is a useful instrument for the assessment of PTSD in veterans. Several scales that are reflective of PTSD symptoms show large differences between PTSD and non-PTSD veterans. Future research should continue to investigate the unique characteristics of PTSD in OEF/OIF/OND era combat veterans.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a grant from the National Institute of Alcohol Abuse and Alcoholism (K23 AA 016120-01) and by the Memphis VAMC Office of Research and Development.

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Washington, DC: Author.

Beckham

J. C.

Feldman

M. E.

Kirby

A. C.

Hertzberg

M. A.

Moore

S. D.

(1997). Interpersonal violence and its correlates in Vietnam veterans with chronic posttraumatic stress disorder. Journal of Clinical Psychology, 53, 859-869. doi:10.1002/(SICI)1097-4679(199712)53:8<859::AID-JCLP11>3.0.CO;2-J

Begic

Jokic-Begic

(2001). Aggressive behavior in combat veterans with post-traumatic stress disorder. Military Medicine, 166, 671-676.

Blake

D. D.

Weathers

F. W.

Nagy

L. M.

Kaloupek

D. G.

Gusman

F. D.

Charney

D. S.

Keane

T. M.

(1995). The development of a clinician-administered PTSD scale. Journal of Traumatic Stress, 8, 75-90. doi:10.1002/jts.2490080106

Brady

K. T.

Killeen

T. K.

Brewerton

Lucerini

(2000). Comorbidity of psychiatric disorders and posttraumatic stress disorder. Journal of Clinical Psychiatry, 61, 22-32. doi:10.5772/26689

Busse

Whiteside

Waters

Hellings

(2014). Exploring the reliability and component structure of the Personality Assessment Inventory in a neuropsychological sample. The Clinical Neuropsychologist, 28, 237-251. doi:10.1080/13854046.2013.876100

Calhoun

P. S.

Beckham

J. C.

Feldman

M. E.

Barefoot

J. C.

Haney

Bosworth

H. B.

(2002). Partners’ ratings of combat veterans’ anger. Journal of Traumatic Stress, 15, 133-136. doi:10.1023/A:1014808008142

Calhoun

P. S.

Boggs

C. D.

Crawford

E. F.

Beckham

J. C.

(2009). Diagnostic efficiency of the Personality Assessment Inventory LOGIT function for posttraumatic stress disorder in women. Journal of Personality Assessment, 91, 409-415. doi:10.1080/00223890903087497

Calhoun

P. S.

Collie

C. F.

Clancy

C. P.

Braxton

L. E.

Beckham

J. C.

(2010). Use of the PAI in assessment of posttraumatic stress disorder among help-seeking veterans. In Blais

M. A.

Baity

M. R.

Hopwood

C. J.

(Eds.), Clinical applications of the personality assessment inventory (pp. 93-111). New York, NY: Routledge.

10.

Calhoun

P. S.

Earnst

K. S.

Tucker

D. D.

Kirby

A. C.

Beckham

J. C.

(2000). Feigning combat-related posttraumatic stress disorder on the Personality Assessment Inventory. Journal of Personality Assessment, 75, 338-350. doi:10.1207/S15327752JPA7502_11

11.

Cherepon

J. A.

Prinzhorn

(1994). Personality Assessment Inventory (PAI) profiles of adult female abuse survivors. Assessment, 1, 393-399. doi:10.1177/107319119400100408

12.

Crawford

E. F

Calhoun

P. S.

Braxton

L. E.

Beckham

J. C.

(2007). Validity of the Personality Assessment Inventory aggression scales and violence potential index in veterans with PTSD. Journal of Personality Assessment, 88, 90-98. doi:10.1080/00223890709336839

13.

Drury

Calhoun

P. S.

Boggs

Araujo

Dennis

M. F.

Beckham

J. C.

(2009). Influences of comorbid disorders on Personality Assessment Inventory profiles in women with posttraumatic stress disorder. Journal of Psychopathological Behavior Assessment, 31, 119-128. doi:10.1007/s10862-008-9101-5

14.

Epidemiology Program, Post-Deployment Health Group, Office of Public Health, Veterans Health Administration, Department of Veterans Affairs. (2015). Analysis of VA health care utilization among Operation Enduring Freedom (OEF), Operation Iraqi Freedom (OIF), and Operation New Dawn (OND) Veterans. Washington, DC: Author. Retrieved from http://www.publichealth.va.gov/docs/epidemiology/healthcare-utilization-report-fy2015-qtr2.pdf

15.

Forbes

Creamer

Biddle

(2001). The validity of the PTSD checklist as a measure of symptomatic change in combat-related PTSD. Behavior Research and Therapy, 39, 977-986. doi:10.1016/S0005-7967(00)00084-X

16.

Freeman

T. W.

Roca

(2001). Gun use, attitudes toward violence, and aggression among combat veterans with chronic posttraumatic stress disorder. Journal of Nervous and Mental Disease, 189, 317-320. doi:10.1097/00005053-200105000-00008

17.

Holmes

G. E.

Williams

C. L.

Haines

(2001). Motor vehicle accident trauma exposure: Personality profiles associated with posttraumatic diagnoses. Anxiety, Stress & Coping, 14, 301-313. doi:10.1080/10615800108248359

18.

Hopwood

C. J.

Baker

K. L.

Morey

L. C.

(2008). Extratest validity of selected Personality Assessment Inventory scales and indicators in an inpatient substance abuse setting. Journal of Personality Assessment, 90, 574-577. doi:10.1080/00223890802388533

19.

Kessler

R. C.

Chiu

W. T.

Demler

Walters

E. E.

(2005). Prevalence, severity, and comorbidity of twelve-month DSM-IV disorders in the National Comorbidity Survey Replication (NCS-R). Archives of General Psychiatry, 62, 617-627. doi:10.1001/archpsyc.62.6.617

20.

Kessler

R. C.

Sonnega

Bromet

Hughes

Nelson

C. B.

(1995). Posttraumatic stress disorder in the National Comorbidity Survey. Archives of General Psychiatry, 52, 1048-1060. doi:10.1001/archpsyc.1995.03950240066012

21.

McDevitt-Murphy

M. E.

Murphy

J. G.

Monahan

C. J.

Flood

A. M.

. . . Weathers

F. W.

(2010). Unique patterns of substance misuse associated with PTSD, depression, and social phobia. Journal of Dual Diagnosis, 6, 94-110. doi:10.1080/15504261003701445

22.

McDevitt-Murphy

M. E.

Weathers

F. W.

Adkins

J. W.

Daniels

J. B.

(2005). Use of the Personality Assessment Inventory in assessment of posttraumatic stress disorder in women. Journal of Psychopathology and Behavioral Assessment, 27, 57-65. doi:10.1007/s10862-005-5380-2

23.

McDevitt-Murphy

M. E.

Weathers

F. W.

Flood

A. M.

Eakin

D. E.

Benson

T. A.

(2007). The utility of the PAI and the MMPI-2 for discriminating PTSD, depression, and social phobia in trauma-exposed college students. Assessment, 14, 181-195. doi:10.1177/1073191106295914

24.

McFall

Fontana

Raskind

Rosenbeck

(1999). Analysis of violent behavior in Vietnam combat veteran psychiatric inpatients with posttraumatic stress disorder. Journal of Traumatic Stress, 12, 501-517. doi:10.1023/A:1024771121189

25.

McFall

Wright

P. W.

Donavan

D. M.

Raskind

(1999). Multidimensional assessment of anger in Vietnam veterans with posttraumatic stress disorder. Comprehensive Psychiatry, 40, 216-220. doi:10.1016/S0010-440X(99)90006-8

26.

Morey

L. C.

(1991). Personality Assessment Inventory: Professional manual. Lutz, FL: Psychological Assessment Resources.

27.

Morey

L. C.

(1996). An interpretive guide to the Personality Assessment Inventory. Odessa, FL: Psychological Assessment Resources.

28.

Mozley

S. L.

Miller

M. W.

Weathers

F. W.

Beckham

J. C.

Feldman

M. E.

(2005). Personality Assessment Inventory (PAI) profiles of male veterans with combat-related posttraumatic stress disorder. Journal of Psychopathology and Behavioral Assessment, 27, 179-189. doi:10.1080%2F00223890903087497

29.

Parker

J. D.

Daleiden

E. L.

Simpson

C. A.

(1999). Personality Assessment Inventory substance-use scales: Convergent and discriminant relations with the addiction severity index in a residential chemical dependence treatment setting. Psychological Assessment, 114, 507-513. doi:10.1037/1040-3590.11.4.507

30.

Patry

M. W.

Magaletta

P. R.

(2014). Measuring suicidality using the Personality Assessment Inventory: A convergent validity study with federal inmates. Assessment, 22, 36-45. doi:10.1177/1073191114539381

31.

Patry

M. W.

Magaletta

P. R.

Diamond

P. M.

Weinman

B. A.

(2011). Establishing the validity of the Personality Assessment Inventory drug and alcohol scales in a corrections sample. Assessment, 18, 50-59. doi:10.1177/1073191110368484

32.

Rosenthal

Rosnow

R. L.

Rubin

D. B.

(2000). Contrasts and effect sizes in behavioral research: A correlational approach. New York, NY: Cambridge University Press.

33.

Seal

Metzler

T. J.

Gima

K. S.

Bertenthal

Maguen

Marmar

C. R.

(2009). Trends and risk factors for mental health diagnoses among Iraq and Afghanistan veterans using Department of Veterans Affairs Health Care, 2002-2008. American Journal of Public Health, 99, 1651-1658. doi:10.2105/AJPH.2008.150284

34.

Tabachnick

B. G.

Fidell

L. S.

(2006). Using multivariate statistics (5th ed.). Needham Heights, MA: Allyn & Bacon.

35.

Tanielian

Jaycox

L. H.

(2008). Invisible wounds of war: Psychological and cognitive injuries, their consequences, and services to assist recovery. Santa Monica, CA: RAND.

36.

Trivedi

R. B.

Post

E. P.

Sun

Pomerantz

Saxon

Piette

J. D.

. . . Nelson

(2015). Prevalence, comorbidity and prognosis of mental health among US veterans. American Journal of Public Health, 105, 2564-2569. doi:10.2105/AJPH.2015.302836

37.

Van Voorhees

E. E.

Dennis

P. A.

Elbogen

E. B.

Clancy

C. P.

Hertzberg

M. A.

Beckham

J. C.

Calhoun

P. S

. (2014). Personality Assessment Inventory internalizing and externalizing structure in veterans with posttraumatic stress disorder: Associations with aggression. Aggressive Behavior, 40, 582-592. doi:10.1002/ab.21554

38.

Weathers

F. W.

Keane

T. M.

Davidson

J. R.

(2001). Clinician-Administered PTSD Scale: A review of the first ten years of research. Depression and Anxiety, 13, 132-156. doi:10.1037/t00072-000

39.

Weathers

F. W.

Litz

B. T.

Herman

D. S.

Huska

J. A.

Keane

T. M.

(1993, October). The PTSD Checklist (PCL): Reliability, validity, and diagnostic utility. Paper presented at the annual meeting of the International Society for Traumatic Stress Studies, San Antonio, TX.