Unique Patterns of Substance Misuse associated with PTSD, Depression, and Social Phobia

Abstract

Objectives

This study investigated the relations between post-trauma psychopathology and substance abuse in a sample of trauma-exposed college students (n = 136) assigned to four groups based on primary diagnosis: posttraumatic stress disorder, depression, social phobia, or well-adjusted (participants who had low levels of distress). Groups were compared on a series of dimensions of substance use/abuse.

Results

Participants in the PTSD group evidenced greater substance use and abuse than those in the social phobia and well-adjusted groups on several dimensions and greater alcohol consumption than the depressed group. Correlation analyses suggested that most dimensions of substance abuse were related more strongly to avoidance and numbing (cluster C) symptoms than to reexperiencing and hyperarousal.

Conclusions

The present findings suggest that trauma-related psychopathology may be associated with a more hazardous pattern of substance use than depression and social phobia.

Several recent studies have documented high rates of comorbidity between posttraumatic stress disorder (PTSD) and substance use disorders (SUD)(see Jacobsen, Southwick & Kosten, 2001 and Chilcoat & Menard, 2003 for reviews). This finding has been reported across a wide range of samples, including military veterans (Kulka et al., 1990), female survivors of sexual assault (Ullman, Filipas, Townsend, & Starzynski, 2005), and disaster survivors (Adams, Boscarino, & Galea, 2005). The combination of PTSD and substance abuse appears to be particularly severe and chronic compared to either disorder alone (Brady, Killeen, Saladin, Dansky, & Becker, 1994; Ouimette, Ahrens, Moos, & Finney, 1998). PTSD-SUD appears to be associated with a more chronic and severe course than SUD with other co-occurring psychiatric disorders (Ouimette et al., 1998), but no previously published studies have investigated the extent to which individuals with PTSD are at increased risk for substance abuse compared to trauma-exposed individuals meeting criteria for other specific disorders.

A related issue that has received only limited attention in the literature is the relationship between specific PTSD symptom clusters and substance abuse. PTSD, as defined in DSM-IV-TR (APA, 2000) is comprised of three symptom clusters: reexperiencing (Criterion B in DSM), avoidance/emotional numbing (Criterion C), and hyperarousal (Criterion D). Results from factor analyses suggest that the cluster of symptoms included on criterion C are best represented by two different dimensions: avoidance and emotional numbing (Asmundson et al., 2000; King, Leskin, King, & Weathers, 1998). Only a handful of studies have examined the associations between specific PTSD symptom clusters and substance abuse variables and findings have been mixed. An early investigation reported that re-experiencing and hyperarousal symptoms were associated with alcohol, but not drug, abuse and that avoidance/numbing and intrusive symptoms were associated with drug abuse in a sample of veterans (McFall, MacKay & Donovan, 1992). A study of female substance abusers suggested that alcohol dependence was correlated only with hyperarousal symptoms, while anxiolytic dependence was correlated with hyperarousal and numbing symptoms and opiate dependence was associated with hyperarousal, intrusive, and numbing symptoms (Stewart, Conrod, Pihl, & Dongier, 1999). A recent study of survivors of intimate partner violence found significant associations between each of the PTSD symptom dimensions and substance use (Sullivan & Holt, 2008). On the whole, studies more frequently report associations between the reexperiencing and hyperarousal symptoms and substance abuse, and less frequently find avoidance and numbing symptoms to be significantly related to substance use/abuse (Simons, Gaher, Jacobs, Meyer, & Johnson-Jiminez, 2005; Stewart, Conrod, Samoluk, Pihl, & Dongier, 2000; Read, Brown, & Kahler, 2004). No previously published studies have investigated the relations between PTSD symptom clusters and substance abuse among college students, a particularly important group due to their high rate of substance abuse and the high rate of potentially traumatic events that may occur within the context of substance use, such as sexual victimization, and alcohol-related injury (Hingson, Heeren, Winter, & Wechsler, 2005).

Many investigations of relationships between PTSD and substance abuse have focused on alcohol and illicit drugs but excluded nicotine, although the relations between PTSD and nicotine/tobacco use have been explored in an independent literature. A recent study exhaustively reviewed literature on smoking and PTSD among persons experiencing traumatic events and concluded that although smoking rates appear to be particularly high among persons with PTSD, and that respondents seem to report increases in smoking following traumatic events, the literature has not reached a consensus on whether this association is specific to PTSD-related pathology or to negative affect that may be shared with other diagnoses (Feldner, Babson, & Zvolensky, 2007).

The present study had two primary aims. The first aim was to examine the extent to which individuals with PTSD differed from trauma-exposed persons with other primary diagnoses (specifically depression and social phobia) on measures of substance use and abuse. Second, we planned to investigate the role of specific PTSD symptom dimensions in predicting substance abuse. We included measures of consumption and substance-related problems for alcohol, nicotine and illicit drugs. Depression and social phobia were chosen as comparison diagnoses for three reasons. First, they are particularly relevant to disentangling the specific relationship of PTSD to substance abuse outcomes, apart from variance due to general distress or anxiety and depressive symptoms more broadly. Both depression and social phobia have been associated with substance abuse. Depression has been associated with drinking to cope motives in college students, which seems to result in a more hazardous pattern of use (Cooper, 1994). The relationship between social anxiety and alcohol abuse is complex, with adult samples showing a positive association (Buckner & Schmidt, 2009), but college samples showing a negative correlation (Ham & Hope, 2005).

Second, PTSD shares symptoms with depression and social phobia, although the degree of symptom overlap with PTSD is substantial for depression, but somewhat less so for social phobia. This disparity was expected to be reflected in the pattern of substance abuse associated with the three diagnoses, with greater discrimination predicted between PTSD and social phobia than between PTSD and depression. Third, depression and social phobia are more prevalent in the general population than other relevant disorders that might have been selected for examination and both are sufficiently prevalent in a sample of undergraduate students which made the contrasted groups design feasible. These data were collected as part of a larger study that examined the assessment of trauma-related pathology and health (McDevitt-Murphy, Weathers, Flood, Eakin & Benson, 2007; Flood, McDevitt-Murphy, Weathers, Benson & Eakin, 2009).

Method

Participants

Four hundred fifty undergraduates from a large public university in the southeastern United States completed a screening questionnaire packet as an optional activity for extra credit in psychology courses. We applied a set of cut scores to the screening measures in order to identify participants who appeared likely to meet criteria for one of the four groups (PTSD, depression, social phobia, well-adjusted). Upon screening into the study, participants were invited to participate further if they appeared likely to be appropriate for one of the groups. This resulted in a total sample of 136. Participants were predominantly female (n = 112; 82.4%). In terms of race, a majority of participants described themselves as Caucasian (n = 114; 83.8%), fifteen participants described themselves as African-American (11.0%), two described themselves as Asian (1.5%) and one participant endorsed multiple descriptors. As noted earlier, all participants reported at least one event that satisfied the two-part definition of a trauma in Criterion A of the DSM-IV criteria for PTSD. The most common index events included transportation accidents (n = 28, 20.6%); sexual assault/unwanted sexual experience (n = 17, 12.5%); sudden death of someone close to respondent (n = 18, 13.2%); and physical assault or assault with a weapon (n = 15, 11.1%).

The PTSD group included 30 participants who met criteria for PTSD using the original Frequency-1/Intensity-2 (F1/I2) scoring rule on the CAPS (Weathers, Ruscio, & Keane, 1999), described in the measures section. The depression group included 23 participants who met criteria for current major depression (n = 19), minor depression (n = 1), or dysthymia (n = 3) as their primary diagnosis based on the SCID. The social phobia group included 21 participants who met criteria for current social phobia as their primary diagnosis on the SCID. The well-adjusted group consisted of 18 participants who obtained low scores on screening measures and were presumed not to meet criteria for any of the DSM-IV diagnoses considered. Forty-four of the participants who screened positive on at least one of the screening measures and were interviewed were subsequently excluded because they did not meet diagnostic criteria for any of the disorders of interest, and were therefore inappropriate for any of the groups. Group analyses included 92 participants, including 30 in the PTSD group, 23 in the depression group, 21 in the social phobia group and 18 in the well-adjusted group. Correlational analyses also include the 44 people excluded from the contrast groups.

There was a high degree of comorbidity in this sample. Of the 30 participants in the PTSD group, 25 (83.3%) met criteria for at least one additional current Axis I diagnosis, most commonly social phobia (n = 16, 53.3%), specific phobia (n = 9, 30.0%), generalized anxiety disorder (n = 7; 23.3%) and dysthymia (n = 7, 23.3%). Of the 23 participants in the depression group, 11 (47.8%) met criteria for at least one additional current Axis I diagnosis, most commonly specific phobia (n = 7; 30.4%), panic disorder (n = 3; 13.0%), and generalized anxiety disorder (n = 3, 13.0%). Of the 21 participants in the social phobia group, 4 (19.0%) met criteria for at least one additional current Axis I diagnosis. All of these participants met criteria for a current specific phobia, and two (9.5%) of them also met criteria for generalized anxiety disorder.

Measures

Screening measures

Trauma history was assessed using the Life Events Checklist (LEC) from the CAPS (Blake et al., 1995) which inquires about exposure to 17 categories of traumatic life events. The LEC was included in the initial screening packet, and a space for participants to write a brief narrative description of their “worst event” along with a Criterion A questionnaire to ascertain whether the worst event met both parts of the DSM-IV definition of a traumatic event (APA, 2001). The screening packet also included the PTSD Checklist (PCL; Weathers et al., 1993), the Social Phobia Scale (SPS; Mattick & Clarke, 1989) and the Beck Depression Inventory, Revised (BDI-II; Beck et al., 1996) which are all psychometrically sound brief symptom measures.

Diagnostic measures

Psychiatric diagnoses were assigned based on structured interviews. The CAPS (Blake et al., 1995) assesses the 17 PTSD symptoms included in the DSM-IV diagnostic criteria. On the CAPS, the frequency and intensity and intensity of PTSD symptoms are rated on separate 5-point (0-4) rating scales. We used the F1/I2 scoring rule on the CAPS (Weathers et al., 1999). By this convention, a symptom is rated as present if the frequency rating is at least “1” and the intensity rating is at least “2” (both rated on 5-point rating scales ranging from 0 to 4) and the PTSD diagnosis is made if a participant’s pattern of symptoms fits the DSM-IV algorithm for B, C, and D symptoms. We assessed current (past month) symptoms on the CAPS.

The SCID I (First, Gibbon, Spitzer, & Williams, 1996) was used to assess current mood and anxiety disorders. The SCID I has a module for each Axis I diagnosis, which includes the DSM-IV criteria in question format. Each symptom is coded as present, subthreshold or absent and each module produces a dichotomous rating for the diagnosis. Interrater reliability coefficients for the SCID have been reported to be in the range of 0.61 to 1.00 for specific Axis I diagnoses (Ventura, Liberman, Green, Shaner, & Mintz,1998; Walby, Odegaard, & Mehlum, 2006).

Substance Use/Abuse

Quantity and frequency of alcohol use over the previous six months were assessed using a single item each. Response options for the frequency item included, “I do not drink at all”, “I drink less than once per month”, “I drink less than once per week”, “I drink once or twice per week”, “I drink three to four times per week”, and “I drink daily or almost daily”. Respondents were asked to indicate the number of standard drinks typically consumed per occasion. Participants also completed a single item inquiring about their “usual level of intoxication” on drinking occasions, which included four response options: “No effect from alcohol”, “ Mildly intoxicated (feel a slight “buzz” from alcohol); “Moderately intoxicated (some impairment in judgment, feeling lightheaded or flushed)”; “Significantly intoxicated (legally drunk, significant impairment in judgment, staggering)”. Participants who reported any alcohol use in the past 6 months completed the Rutgers Alcohol Problem Index (RAPI; White & Labouvie, 1989) as a measure of alcohol-related problems (a total RAPI score of 0 was entered for participants who reported abstaining from alcohol). The RAPI assessed the frequency of occurrence of 23 alcohol-related problems over the past six months and has demonstrated good reliability and validity with college students (Martens, Neighbors, Dams-O’Connor, Lee, & Larimer, 2007). We used a modification of the original RAPI, which included two additional items that query drinking and driving. The 25-item measure has demonstrated good psychometric characteristics in other samples of college drinkers (Neighbors, Larimer, Geisner, & Knee, 2004; Neighbors, Lee, Lewis, Fossos, & Larimer, 2007). The 25-item RAPI demonstrated excellent internal consistency in this sample (Cronbach’s α = .93).

Participants completed questions of quantity and frequency of cigarette use over the past six months. Frequency of smoking was rated with a six point scale including response options: “I do not smoke at all”, “I smoke less than once per month”, “I smoke less than once per week”, “I smoke once or twice per week”, “I smoke three to four times per week”, and “I smoke daily or almost daily”. Participants who reported any smoking in the past six months were asked to complete the Revised Tolerance Questionnaire (RTQ; Tate & Schmitz, 1993), which is a revised version of the Fagerstrom Tolerance Questionnaire (FTQ; Fagerstrom & Schneider, 1989). The RTQ has been used extensively among diverse samples including young adult smokers (Quinlan & McCaul, 2000; Tate & Schmitz, 1993) and has demonstrated strong internal consistency, with α coefficients ranging from .72 to .88 (Ratner et al., 2000, Armitage, 2007). In the current sample, the RTQ evidenced good internal consistency (Cronbach’s α = .89).

Quantity and frequency of drug use over the past six months were assessed with a single question each. Frequency of drug use was assessed with a six point scale including response options: “I do not use drugs at all”, “I use drugs less than once per month”, “I use drugs less than once per week”, “I use drugs once or twice per week”, “I use drugs three to four times per week”, and “I use drugs daily or almost daily.” Participants who reported drug use in the past six months also completed the Drug Abuse Screening Test (DAST; Skinner, 1982). The DAST is a 28-item questionnaire assessing the presence or absence of a range of negative consequences of drug abuse over the past six months. Variations of the DAST have been used to screen diverse samples ranging from college undergraduates (McCabe et al., 2006) to clinical populations (Yudko, 2007). The DAST-28 has shown adequate psychometric properties with high internal consistencies (Cronbach’s α = .92 in the present sample) and good test–retest reliability, r = .85 (Yudko, 2007).

Procedure

Participants were invited for participation if they (a) reported experiencing at least one stressor on the LEC that met the two-part definition of a trauma in Criterion A of the DSM-IV PTSD criteria (APA, 1994); (b) exceeded the screening criterion on at least one of the three symptom measures (PCL, BDI, or SPS), or scored below the screening criteria on all three measures (the criterion for well-adjusted group membership); and (c) indicated interest in further participation for additional extra credit following completion of the screening measures (see measures).In order to create three diagnostic groups (PTSD, depression, social phobia) all participants who screened positive on at least one of the three symptom measures and agreed to further participation were administered structured diagnostic interviews, including the Clinician-Administered PTSD Scale (CAPS; Blake et al., 1995) and the mood and anxiety disorder modules of the Structured Clinical Interview for DSM-IV (SCID I; First et al., 1996). A “well-adjusted” trauma-exposed control group was created by recruiting a randomly selected subset (30%) of participants who scored below all three screening criteria. Those participants completed the self-report battery, but not the clinical interviews. Participants in all groups other than well-adjusted completed three assessment sessions, the first of which included the interview measures. Participants in the well-adjusted group were not interviewed and completed the self-report measures in two assessment sessions. To control for order effects, the sequence of assessment instruments was randomized.

Interviews were conducted by doctoral students in clinical psychology and a licensed clinical psychologist. All interviews were audiotaped, and reliability analyses were conducted on 24 (20%) randomly selected interviews. Two raters each listened to twelve tapes and rated both the CAPS and SCID interviews. Raters agreed 100% on principal diagnosis, and thus on group assignment. Excellent interrater reliability was also found for CAPS severity scores. Intraclass correlation coefficients (one-way random effects model, single measure; Shrout & Fleiss, 1979) were derived for CAPS total severity (the sum of frequency and intensity for each item), and for severity scores for B, C, and D symptoms. For total CAPS severity, the intraclass correlation was .99, and for B, C, and D severity, the intraclass correlations were .96, .95, and .90, respectively.

Results

Diagnostic Group Differences on Substance use/abuse measures

Table 1 displays descriptive statistics by group for the nine measures of substance use and abuse. ANOVA results for four variables (typical level of intoxication, frequency of smoking, RAPI, and RTQ) were statistically significant. Results from pairwise contrasts are also indicated in Table 1. The PTSD group reported significantly more drinks per occasion than the Depression group. The PTSD group also reported drinking to significantly higher levels of intoxication than all other groups. On the RAPI, the PTSD group mean was significantly higher than both the Social Phobia and the Well-adjusted group means. With regard to smoking, the PTSD group reported significantly more frequent smoking and higher levels of nicotine dependence than the Social Phobia group. On a measure of frequency of drug use, the PTSD group reported significantly more frequent drug use than the well-adjusted group. On the DAST, the PTSD group mean was significantly higher than the well-adjusted group mean and on the RTQ, the PTSD group mean was significantly higher than the social phobia group.

TABLE 1.

Group Differences of Measures of Substance Use and Abuse

	PTSD (P) n = 30 M (SD)	Depression (D) n = 23 M (SD)	Social phobia (SP) n = 21 M (SD)	Well-adjusted (WA) n = 18 M (SD)	F value	Significant contrasts
Frequency of alcohol consumption	2.27 (1.41)	1.78 (1.24)	1.65 (1.22)	1.72 (1.497)	1.13
No. of drinks per occasion	5.04 (3.29)	3.20 (2.24)	3.58 (3.64)	4.18 (4.23)	1.38	PTSD > D
Typical level of alcohol  intoxication	1.67 (0.78)	0.91 (0.79)	1.10 (0.99)	1.00 (1.06)	3.59*	PTSD > D, SP, WA
Frequency of drug use	0.63 (1.28)	0.48 (1.08)	0.19 (0.40)	0.00 (0.00)	0.506
Frequency of smoking	2.40 (2.25)	1.82 (2.30)	0.62 (1.32)	1.89 (2.27)	3.04*	PTSD, D, WA > SP
No. of cigarettes per month	118.08 (181.48)	70.00 (159.60)	33.37 (137.30)	61.00 (105.10)	1.15
RAPI total	14.93 (18.22)	9.13 (12.82)	5.05 (6.95)	3.67 (4.28)	3.83*	PTSD > SP, WA
DAST total	31.00 (4.76)	30.00 (4.06)	29.76 (4.30)	28.61 (1.79)	1.34	PTSD > WA
RTQ total	19.30 (8.36)	15.39 (7.68)	13.14 (6.63)	16.00 (7.70)	2.81*	PTSD > SP

^*Note. p < .05

^**p < .01.

PTSD = post-traumatic stress disorder; RAPI = Rutgers Alcohol Problems Index; DAST = Drug Abuse Screening Test; RTQ = Revised Tolerance Questionnaire.

Relations between PTSD symptoms and substance abuse

We investigated relations between PTSD symptom clusters (CAPS severity scores) and substance use, nicotine dependence, and alcohol and drug problems. Correlations are presented in Table 2. Reexperiencing symptoms were significantly correlated with only one variable, RAPI. C symptoms were significantly correlated with frequency of smoking, quantity of cigarettes smoked, nicotine dependence and drug and alcohol-related problems. C symptoms were further separated into the avoidance and numbing dimensions, based on factor analyses suggesting that these are best considered separate constructs (Asmundson et al., 2000). The pattern of correlations for the disaggregated C symptoms suggested stronger relationships for the numbing symptoms than for the avoidance symptoms. D symptoms (Hyperarousal) were not significantly correlated with any of the substance use/abuse measures.

TABLE 2.

Correlations Between PTSD Severity Scores and Measures of Substance Use and Abuse (N = 118^a)

	PTSD total severity	Re-experiencing (B symptoms)	Avoidance/numb ing (C symptoms)	Disaggregated C symptoms		Hyperarousal (D symptoms)
				Avoidance	Numbing
Frequency of alcohol  consumption	.05	.04	.11	.04	.13	−.03
No. of drinks	.08	.12	.11	.00	.14	−.01
Typical level of alcohol  intoxication	.04	.08	.11	.09	.09	−.09
Frequency of drug use	.06	.05	.16	.05	.18	−.07
Frequency of smoking	.19*	.13	.31**	.20*	.30**	.01
No. of cigarettes per month	.18	.11	.28**	.20*	.26**	.02
RAPI total	.29**	.20*	.35**	.21*	.34**	.16
DAST total	.14	.07	.24**	.16	.23*	.01
RTQ total	.21*	.17	.31**	.20*	.29**	.03

^aNote. N varied slightly for some variables, due to missing data; N = 117 for frequency of smoking, number of cigarettes per month, frequency of drug use, and frequency of alcohol consumption; N = 113 for number of drinks per occasion; and N = 112 for typical level of intoxication.

^*p < .05

^**p < .01.

PTSD = post-traumatic stress disorder; RAPI = Rutgers Alcohol Problems Index; DAST = Drug Abuse Screening Test; RTQ = Revised Tolerance Questionnaire.

Discussion

The present study investigated relations between psychopathology and aspects of substance use and abuse in a sample of trauma-exposed college students. Analyses of group differences across three diagnostic groups and one group of well-adjusted participants suggest that participants with PTSD tend to exhibit more extreme profiles of substance abuse than trauma-exposed participants with diagnoses of depression or social phobia or well-adjusted participants. With respect to alcohol, participants with PTSD reported having more drinks per occasion than participants with depression, and drinking to higher levels of intoxication than participants in any of the other groups. On the RAPI, a measure of alcohol problems, the PTSD group scored significantly higher than the social phobia and well-adjusted groups. All of these findings suggest a more hazardous pattern of alcohol use among participants with PTSD, even compared to trauma-exposed participants with other psychological disorders. This pattern is consistent with a large body of literature suggesting a specific link between PTSD and alcohol abuse (Ouimette & Brown, 2003), and in particular, a recent study of adults suggesting that PTSD placed individuals at higher risk of heavy drinking, compared to major depressive disorder, or other anxiety disorders (Arch, Craske, Stein, Sherbourne, & Roy-Bourne, 2006).

Although there were similar levels of drug use across the groups, the PTSD group had a significantly higher levels of drug abuse symptoms that the well-adjusted group. With respect to cigarette smoking, a complex pattern emerged. In the current study the PTSD group demonstrated a greater degree of nicotine dependence than the social phobia group. The PTSD and depression groups were not differentiated on any aspect of smoking or nicotine dependence, in contrast to some published findings suggesting that in addition to shared variance between PTSD and depression, PTSD symptoms may pose incremental risk for heavy smoking among, beyond the effect of co-occurring depressive symptoms (Thorndike, Wernicke, Pearlman & Haaga, 2006). The social phobia group reported extremely low levels of smoking and nicotine dependence, such that they reported significantly less frequent smoking than all other groups, including the well-adjusted group. This was a surprising finding, given that a large epidemiological study found higher rates of current smoking among persons with current social phobia, compared to the contrast group of well-adjusted adults (Lasser et al., 2000).

In addition to group comparisons, we investigated the associations between substance abuse and specific PTSD symptom clusters. Overall, the association between PTSD and substance use/abuse in this sample seemed to be largely related to C symptoms, and more specifically to emotional numbing. Five of the nine substance use/abuse variables were correlated with C symptoms, while only one was correlated with B (reexperiencing) and none were correlated with D (hyperarousal) symptoms. These results are inconsistent with other reports in the literature, which suggest that reexperiencing and hyperarousal symptoms may drive the association between PTSD and substance abuse (McFall et al., 1992; Read et al., 2004; Simons et al., 2005; Stewart et al., 1999; Stewart et al., 2000). Our results for smoking are partially consistent with a prior report on the relations between PTSD symptoms and smoking in a sample of college students, which found smoking to be related to both avoidance and hyperarousal symptoms (Feldner et al., 2007).

Overall, the present results suggest that emotional numbing symptoms contribute more than other PTSD symptoms to substance abuse. One construct that may explain this relationship is experiential avoidance. Although we did not measure experiential avoidance in this sample, some published studies have found this construct to be predictive of drinking to cope motives (Stewart, Zvolensky, & Eifert, 2002) and to be correlated with emotional numbing (Marx & Sloan, 2005; Tull & Roemer, 2003). Individuals with relatively high levels of emotional numbing may engage in substance misuse with the aim of avoiding aversive emotional experiences, similar to the process described as self-medication. The discrepancy between the present findings and prior literature that indicated a greater role for reexperiencing and hyperarousal symptoms in predicting drinking may be related to the age of this sample. Most prior investigations used adult samples recruited from clinical settings whereas the current sample was comprised exclusively of college students screened for trauma exposure. Given that college drinking is typically most strongly associated with drinking for social or positive reinforcement motives (Grant, Stewart, O’Connor, Blackwell, & Conrod, 2007), it is possible that in this sample, the relative importance of coping with these specific PTSD symptoms (reexperiencing & hyperarousal) as a motive for drinking is diminished. The finding that reexperiencing symptoms were significantly associated with alcohol problems and not with drug/alcohol consumption or drug problems was unexpected. Individuals with these symptoms may not drink or use drugs frequently but may be at high risk for problems when they do drink.

These results parallel, to some extent, a recent finding in the literature suggesting that mood lability was predictive of later substance-related problems, but not substance use, for both alcohol and marijuana, in a sample of college students (Simons & Carey, 2007). Prior research has also shown that drinkers who endorse motives of coping with depression or coping with anxiety demonstrate a more hazardous pattern (Cooper, 1994). It is possible that significant psychopathology may render substance users more vulnerable to the negative consequences of substance use, as a function of behaving impulsively while drinking or using drugs, or perceiving events that took place while drinking or using drugs in a more critical or negative light. The nature of this relationship should be pursued in future research.

The contrasted-groups design that included three diagnostic groups and a well-adjusted control group was a strength of the present study. Although participants were not recruited from clinical settings, they were carefully diagnosed using gold-standard interview measures. Further, in this study we compared diagnostic groups and also examined continuous scores on symptom measures, allowing us to examine these relationships from both a categorical and a dimensional approach. Similar to clinical samples, there was a high degree of comorbidity in this sample, which likely influenced our results. The high rate of social anxiety within the PTSD group may have had an additive effect, contributing to the increased risk for hazardous substance use.

The strength of inferences that may be drawn from this study is limited by several important factors. First, the use of a sample of college students may limit generalizations to the wider adult population. The sample was largely Caucasian, which limits generalizability to more diverse samples. There were very few men in this study, which precluded the investigation of gender differences. Another limitation is the fact that the timeframes assessed by the measures we used differed. Thus, diagnostic ratings were made based on the past month, while the questionnaires that focused on substance use/abuse asked about the past six months.

The aforementioned limitations lead naturally to suggestions for future research. Exploration of gender differences is warranted. Evidence from both college drinking studies and studies of PTSD-SUD relations suggest that there may be different processes operating for men and women (Najavits et al., 1997; Stewart, Ouimette, & Brown, 2002). An adequately powered study investigating the influence of gender on these relations would shed light on these issues. Further work on this topic with ethnically diverse, adult (non-college) samples is also needed. Additionally, there is a need for prospective studies investigating the temporal sequencing of substance abuse and trauma exposure, and the longitudinal relations between PTSD and substance abuse severity.

The present results suggest that PTSD shows strong and specific associations with several dimensions of substance use and abuse among college students. One practical implication of these results is that campus mental health and substance abuse prevention efforts should be integrated, as there may be functional relations between these domains and considerable overlap between the students with the greatest levels of psychiatric and substance abuse risk. These results also provide support for integrated treatment models targeting both substance use and psychiatric symptoms (Conrod, Stewart, Comeau, & Maclean, 2006). Indeed there is some evidence that depressed college students are unlikely to respond to standard alcohol interventions (Geisner, Neighbors, Lee, & Larimer, 2007), and that interventions that target anxiety among college drinkers may be associated with drinking reductions (Watt, Stewart, Birch, & Berniew, 2006).