Trauma Cue-Elicited Alcohol Craving as a Function of Adult Versus Childhood-Onset Interpersonal Traumatic Events in Young Adult Drinkers

Abstract

Objective:

Childhood trauma may influence risk for alcohol use disorder and posttraumatic stress disorder through negative and positive reinforcement drinking. Laboratory studies evaluating childhood trauma in relation to these phenotypes are limited.

Method:

This study examined the influence of childhood index traumas on responses to trauma and alcohol cues among 184 college students (50.0% female) endorsing lifetime interpersonal trauma and current weekly alcohol use. Participants’ subjective alcohol craving and distress were measured in response to four narrative (trauma vs. neutral) and beverage (alcohol vs. water) cue combinations.

Results:

Forward-fitted linear mixed-effects models indicated main effects of childhood index traumas on distress (β= 6.151, p = .001) and craving (β = 0.656, p = .023), wherein individuals with childhood index traumas showed evidence of elevated levels of distress and craving. Childhood index trauma interacted with the narrative cue to predict distress (β = -10.764, p = .002), wherein individuals with childhood index traumas showed greater levels of distress to the neutral cue, and individuals with adult index traumas showed greater levels of distress to the trauma cue. Childhood index trauma interacted with the beverage cue to predict craving (β = -0.599, p = .011), wherein childhood index traumas were associated with greater levels of craving to neutral cues. Childhood index trauma did not significantly interact with the beverage cue to predict distress or the narrative cue to predict craving (ps > .05).

Conclusions:

Childhood trauma may be more relevant to positive rather than negative reinforcement aspects of alcohol use disorder during young adulthood.

College marks a period of heightened risk for the onset of unsafe alcohol use patterns and alcohol use disorder (AUD; O’Malley & Johnston, 2002; Slutske, 2005). College-aged drinkers with a history of trauma exposure, in particular, exhibit elevated risk for the onset or maintenance of alcohol problems (Ullman et al., 2005), with both pre-college and college-onset traumatic events demonstrating associations with subsequent increases in alcohol use (Berenz et al., 2016; Klanecky et al., 2008; Walsh et al., 2014). However, literature supports the notion that childhood trauma may impart unique risk for AUD and comorbid conditions (Arnow, 2004; Bulik et al., 2001; Copeland et al., 2018; Gilpin et al., 2017). The connection between childhood trauma and subsequent psychopathology may be particularly salient among individuals exposed to interpersonal trauma (e.g., sexual assault), which conveys greater conditional probability for psychopathology compared with accidental traumas (e.g., natural disaster; Kessler et al., 1995). Interpersonal trauma is generally more severe and more likely to be chronic; it is relatively uncommon for an individual to have experienced only one sexual or physical assault in their lifetime (Classen et al., 2005; Gidycz et al., 1993; Kessler et al., 2017). Among adult trauma survivors, the median age of their first interpersonal trauma exposure is 17–18 years (Kessler et al., 2017).

Childhood trauma relates to AUD presentations in clinical populations as well. Individuals with comorbid posttraumatic stress disorder (PTSD) and AUD are more likely to have suffered childhood trauma than individuals with PTSD or AUD alone (Blanco et al., 2013; Evren et al., 2011). Childhood trauma is associated with dysregulation of neurobiological stress and reward response systems, increased emotional reactivity, and impaired quality of life (Dillon et al., 2009; Dunn et al., 2018; Evren et al., 2011; Kuhlman et al., 2015; McLaughlin et al., 2019), which may contribute to the development of a more severe course of PTSD and/or AUD. In treatment settings, childhood trauma history is associated with greater PTSD symptom severity (Cloitre et al., 1997; Hembree et al., 2004). Further, adults who identify their currently most distressing traumatic event as occurring during childhood show higher PTSD symptom severity relative to adults who experienced their most distressing trauma in adulthood (Ogle et al., 2013).

The impact of childhood trauma on adult AUD may arise from the initial effect on problematic drinking in adolescence. Among childhood and adult trauma survivors with uni-morbid PTSD or AUD, comorbid PTSD-AUD, or healthy controls, the childhood trauma group evidenced earlier age at first drink and onset of heaviest alcohol consumption compared with the adult trauma group (Waldrop et al., 2007).

Childhood trauma is more likely to occur before adolescent alcohol use (Clark et al., 1997), and childhood maltreatment has been shown to increase risk of adolescent binge drinking, controlling for relevant demographic variables and parental AUD (Shin et al., 2009). Therefore, not only is childhood adversity a precursor to alcohol use, but childhood trauma is also linked to earlier and more risky drinking behavior. Because earlier onset alcohol use and adolescent binge drinking are predictive of AUD (Englund et al., 2008; Grant et al., 2001; Viner & Taylor, 2007), the increased AUD risk in adolescence and young adulthood is particularly high for children who have experienced trauma.

Given relationships among childhood trauma, PTSD, and AUD, individuals with unresolved childhood trauma may be prone to engage in alcohol use under conditions of stress. In one study, individuals exposed to childhood abuse were more likely to report drinking to cope during the first year of alcohol use (Rothman et al., 2008), which predicts later alcohol problems (Grayson & Nolen-Hoeksema, 2005). Using a clinical laboratory model of self-medication alcohol use in a treatment-seeking sample of individuals with co-morbid PTSD-AUD, greater trauma and alcohol cue-elicited alcohol craving was observed among individuals exposed to childhood trauma, relative to those who did not experience a trauma until adolescence (Schumacher et al., 2006).

Childhood trauma may also contribute to neuroadaptations in reward systems that underlie alcohol-seeking behavior in AUD. Animal literature consistently demonstrates that early life stressors promote changes in reward and reinforcement pathways in the brain (Matthews & Robbins, 2003). In particular, animal models support a causal relationship between early and repeated incidents of social trauma (e.g., single prolonged stress model; social defeat stress model) and both dampened valuation of non-alcohol rewards as well as enhanced alcohol reward, even at first exposure to alcohol (Boutros et al., 2018; Casement et al., 2015; Enman et al., 2015; Koob & Le Moal, 1997, 2005; Montagud-Romero et al., 2018). Extant human literature indicates that childhood trauma may interact with blunted neural reward reactivity to influence problematic alcohol use by way of stress-related anhedonia symptoms and coping-motivated substance use (Corral-Frías et al., 2015). Thus, childhood trauma may prime the brain to devalue non-alcohol rewards and experience greater reward from alcohol, leading to increased alcohol craving. Further, overlap in the stress and alcohol neural circuitries may reinforce connections between exposure to childhood trauma and alcohol-seeking behaviors (Herman, 2012).

Despite support for the suggestion that childhood trauma exposure may influence risk for AUD and PTSD-AUD comorbidity by way of negative reinforcement drinking and heightened reward response to alcohol, laboratory studies evaluating childhood trauma in relation to these phenotypes are limited. As a laboratory model of conditioned alcohol craving responses to trauma and alcohol cues, the trauma and alcohol cue reactivity paradigm is aptly positioned to inform the role of childhood trauma in negative and positive reinforcement aspects of AUD during young adulthood. Young adult college students are an ideal population in which to examine the relationships between child and adult interpersonal trauma exposure and AUD risk, as college students exhibit high rates of interpersonal trauma and AUD (Knight et al., 2002; O’Malley & Johnston, 2002) yet are less likely to have developed additional complex medical and psychiatric comorbidities.

The current study aimed to examine the impact of a history of childhood trauma on responses to trauma and alcohol cues. It was hypothesized that individuals would display greater subjective alcohol craving and distress in response to a childhood index trauma relative to an adult interpersonal trauma narrative (i.e., Childhood Index Trauma × Narrative Cue interaction). In addition, it was hypothesized that individuals with a childhood index trauma would display greater subjective alcohol craving and distress to alcohol beverage cues (i.e., Childhood Index Trauma × Beverage Cue interaction) than individuals with an adult index trauma. Covariates were selected on the basis of theoretical relevance and statistical significance, based on the established best-fitting models of trauma and alcohol cue reactivity in this sample (Berenz et al., 2021). For subjective craving, covariates included narrative cue, beverage cue, cumulative trauma, alcohol use frequency, study site, PTSD symptom severity, and a PTSD × Narrative Cue interaction term. For distress, covariates included narrative cue, beverage cue, cumulative trauma, PTSD symptom severity, and a PTSD × Narrative Cue interaction term; the covariates of gender, age, alcohol use frequency, and study site were also examined but were excluded from the covariate model because they did not significantly predict distress.

Method

Study overview

The present investigation used secondary data collected from 2015 to 2019 at two universities from a laboratory study of trauma and alcohol cue reactivity (Berenz et al., 2021). A “site” variable was created (0 = a university in the mid-Atlantic, 1 = an urban university in the Midwest) for use as a covariate in analyses. At both universities, the Institutional Review Boards approved all study procedures.

Participants

Of the 223 individuals enrolled in the main study, 24 were not scheduled for the laboratory session of the study because of ineligibility at Session 1, 8 had missing data on one or more primary assessments, and 7 completed the laboratory session but had data excluded from analyses because of issues during the laboratory procedure (e.g., participant not following study instructions). A total of 184 participants were included in current study analyses.

An intentionally gender-balanced sample (50% female; ages 18–25) was recruited via paper flyers and online advertisements. Inclusion criteria involved being a college student, endorsing a history of one or more interpersonal traumatic events meeting Criterion A for PTSD according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5; American Psychiatric Association, 2013), and current weekly alcohol use. Exclusion criteria were current use of medications that could interfere with cue-elicited craving (e.g., naltrexone), current or past participation in exposure-based therapy for PTSD, and past-6-month participation in AUD or substance use disorder treatment.

Measures

Session 1 (Baseline)

(A) TIMELINE FOLLOWBACK: The Timeline Followback (TLFB; Sobell & Sobell, 1992) is a calendar method to assess presence and quantity of daily alcohol and illicit drug use of the past 90 days. The TLFB was used to confirm participants’ regular frequency of alcohol consumption (i.e., daily, weekly, or bi-weekly). The TLFB demonstrates good concurrent validity with other measures of alcohol consumption (Grant et al., 1995) and good reliability (Sobell & Sobell, 1996).

(B) TRAUMATIC LIFE EVENTS QUESTIONNAIRE: The Traumatic Life Events Questionnaire (TLEQ; Kubany et al., 2000) is a self-report measure of exposure to 22 types of potentially traumatic events (PTEs) that includes questions related to event frequency, timeline, and severity. The TLEQ was used to determine interpersonal traumatic events to be followed up via clinical interview (below) and cumulative trauma (i.e., participants’ total number of PTE types experienced). The TLEQ has shown evidence of excellent psychometric properties and is recommended as a measure of exposure to PTEs (Gray et al., 2004).

(C) AGE AT TRAUMA NARRATIVE: Participants’ age at their self-identified worst interpersonal traumatic event meeting DSM-5 Criterion A for PTSD was queried. If the participant's trauma narrative event occurred before age 16 years, it was classified as a childhood index trauma. If the participant's trauma narrative event occurred after age 16 years, it was classified as an adult index trauma. Age at trauma narrative was used to create a dichotomous childhood index trauma variable (0 = adult index trauma, 1 = childhood index trauma).

(D) CLINICIAN ADMINISTERED PTSD SCALE: The Clinician Administered PTSD Scale (CAPS-5; Blake et al., 1990) was administered to ascertain current (past-30-day) PTSD diagnostic status (present vs. absent), as well as a PTSD symptom severity score. The CAPS has evidenced high interrater, test–retest, and convergent reliability and high internal consistency (Blake et al., 1995; Weathers et al., 2001). Criterion A for PTSD was assessed for all interpersonTIMELINE FOLLOWBACKal PTEs endorsed on the TLEQ. The remaining DSM-5 PTSD symptoms were queried with regard to all interpersonal traumatic events that met Criterion A for PTSD.

(E) MINI-INTERNATIONAL NEUROPSYCHIATRIC INTERVIEW: The Mini-International Neuropsychiatric Interview (MINI; Sheehan et al., 1998) is a semi-structured clinical interview used to assess current DSM-5 anxiety, mood, eating, and substance use disorders, as well as potential symptoms of psychosis and suicidality. Psychometric analyses conducted with the MINI indicate high validation and reliability scores (Sheehan et al., 1998). The MINI was used to assess current AUD diagnostic status and symptom severity.

Session 2 (Laboratory measures)

(A) SUBJECTIVE UNITS OF DISTRESS SCALE: The Subjective Units of Distress Scale (SUDS; Wolpe, 1958) is a commonly used 0–100 point scale (0 = totally relaxed, 100 = highest distress/fear/anxiety/discomfort that you have ever felt) for assessing emotional distress. Participants responded to the questions, “Please rate your SUDS level while you imagined the last scene” following each of the four cue combinations to examine cue-related changes in distress.

(B) THREE-ITEM ALCOHOL CRAVING SCALE: Alcohol craving was measured by three questions assessed on a 0–10 Likert-type scale: “I crave a drink right now,” “I want a drink right now,” and “I have a desire for a drink right now” (Kozlowski et al., 1996). This scale was administered following each cue trial during Session 2 of the study to assess cue-elicited craving. The numerical responses to each of these questions were averaged to calculate a total craving score.

Procedure

Participants who met eligibility criteria on a brief telephone screen were scheduled for Session 1. During Session 1, informed consent was obtained and participants completed a battery of questionnaires, including the TLEQ. The experimenter administered the TLFB and an alcohol preference form to determine the alcoholic drink for use as an in vivo alcohol cue in Session 2. The MINI and CAPS-5 were administered by a trained doctoral student. A personalized trauma narrative was obtained that included the participant's subjective, behavioral, and physiological responses at the time of the trauma, as well as a description of the physical setting of the participant's self-identified worst interpersonal traumatic event. Participants’ age (in years) at which their self-identified worst traumatic event occurred was also queried. This trauma narrative was adapted to create a personalized 60-second trauma script recorded by the experimenter for use as a trauma narrative cue in Session 2. Participants were scheduled for Session 2 within 2 weeks of Session 1 and instructed to abstain from alcohol for 24 hours before the session. Participants who no longer met criteria after completing Session 1 received compensation but were not scheduled for Session 2.

In Session 2, participants’ acute abstinence from alcohol was verified via breath alcohol analysis and self-report. Cue reactivity procedures were identical to those published in Coffey et al. (2010), the gold standard of assessment. Participants were seated behind an adjustable-height table for beverage cue presentation and monitored with a nonrecording video camera behind a room partition. Each cue reactivity assessment consisted of the following procedures: (a) The participant closed their eyes and listened to a 60-second audio narrative (i.e., personalized trauma or neutral [i.e., a recording of the participant changing a lightbulb] narrative cue) over headphones. (b) The participant actively imagined the narrative scene for 2 minutes while viewing an in vivo beverage cue (i.e., personalized alcohol or neutral spring water cue) placed on the table in front of them. (c) The participant completed the self-report measures (i.e., craving scale, SUDS). These procedures were conducted for a practice trial and four experimental trials, consisting of all four possible cue combinations: trauma-alcohol (TA), trauma-neutral (TN), neutral-alcohol (NA), and neutral-neutral (NN). Presentation of the four cue combinations was counterbalanced between and within subjects by sex.

Data analysis

Analyses were conducted using R 3.6.0 (R Core Team, 2019), wrangled with tidyr 1.1.1 (Wickham & Henry, 2019), summarized with dplyr 1.0.2 (Wickham et al., 2020), and plotted with ggplot2 3.3.2 (Wickham, 2016). Manipulation checks were successful, such that alcohol cues elicited craving and salivation, and trauma cues elicited distress (Berenz et al., 2021).

To address the primary study aims, forward-fitted linear mixed-effects (LME) models were conducted using lme4 1.1-23 (Bates et al., 2015). Random structure was checked with the RePsychLing 0.0.4 (Baayen et al., 2015), followed by tests used with emmeans 1.5.0 (Lenth, 2020). Random effects for all models included the intercept and the slope of the repeated factors narrative cue (0 = neutral, 1 = trauma) and beverage cue (0 = water, 1 = alcohol), and random correlations between slopes were blocked. For subjective craving, covariates included narrative cue, beverage cue, cumulative trauma, alcohol use frequency, study site, PTSD symptom severity, and a PTSD × Narrative Cue interaction term (Model 2a). For distress, covariates included narrative cue, beverage cue, cumulative trauma, PTSD symptom severity, and a PTSD × Narrative Cue interaction term (Model 1a). Models 1b and 2b added the main effect of childhood index trauma (0 = adult index trauma, 1 = childhood index trauma). Models 1c and 2c added the interaction of Narrative Cue × Childhood Index Trauma. Models 1d and 2d added the interaction of Beverage Cue × Childhood Index Trauma. Deviance tests were used to determine whether model fit improved at each iteration of the model, with the best-fitting model being carried forward at each step (Singer & Willett, 2003). Separate iterations of LME models were conducted for the dependent variables of craving and distress.

Results

Sample characteristics and descriptive statistics

See Table 1 for descriptive statistics. The sample comprised equal proportions of men and women, and more than one third of participants identified as non-White. Participants endorsed exposure to approximately six different types of traumatic events on average. The mean CAPS total score was moderately high, with half of the sample screening positive for past-month PTSD. The average age of participants’ self-identified worst interpersonal traumatic event was 16 years, with more than a third classified as a childhood trauma. Childhood and adult index trauma groups did not significantly differ on gender, race, cumulative lifetime trauma, PTSD symptom severity or past-month diagnosis, or AUD diagnosis (ps > .05).

TABLE 1.

Descriptive statistics

Variable	Total sample (N = 184) M (SD) or %	Childhood index trauma (n = 73) M (SD) or %	Adult index trauma (n = 111) M (SD) or %	Group difference test t or χ2
Gender	50% female	42.5% female	55.0% female	χ2= 4.85, p = .183
Race				χ2= 1.99, p = .159 (White vs. non-White)
African American	12.5%	15.1%	10.8%	-
Asian	16.8%	16.4%	17.1%	-
Caucasian/White	53.8%	46.6%	58.6%	-
Multiracial	9.2%	13.7%	6.3%	-
Other	6.5%	5.5%	7.2%	-
Number of lifetime trauma types	6.61 (3.33)	7.00 (3.46)	6.35 (3.23)	t =-1.30, p = .197
Total PTSD symptom severity	15.10(11.30)	14.55 (11.79)	15.46 (11.02)	t =0.53, p = .594
Past-month PTSD diagnosis	49.5%	42.5%	54.1%	χ2 = 2.37, p = .124
Current AUD diagnosis	83.2%	83.6%	82.9%	χ2 = 0.04, p = .845
Mild	30.8%	34.2%	28.8%	-
Moderate	23.2%	20.5%	25.2%	-
Severe	28.1%	28.8%	27.9%	-
Trauma narrative age	16.48 (4.59)	11.89 (3.69)	19.50 (1.72)	t = 18.84, p <.001
Index trauma types	-			-
Unwanted sexual contact	46.5%	37.0%	53.2%	-
Physical assault	25.9%	30.1%	23.4%	-
Threat with a weapon	8.1%	4.1%	10.8%	-
Witnessing violence	17.8%	26.0%	11.7%	-
Other	1.6%	2.7%	0.9%	-

Notes: Gender coded as male (cis-gender) = 1; female (cis-gender) = 2; transgender male = 3; transgender female = 4; other = 5. Number of lifetime trauma types assessed via TLEQ (Traumatic Life Events Questionnaire). Total posttraumatic stress disorder (PTSD) symptom severity assessed via CAPS-5 (Clinician Administered PTSD Scale for DSM-5) total score. Total score past-month PTSD diagnosis assessed via CAPS-5. AUD = alcohol use disorder; AUDIT= Alcohol Use Disorders Identification Test; positive screen for moderate alcohol problems = AUDIT ≥ 8. Current AUD diagnosis and current major depressive disorder diagnosis assessed via MINI (Mini-International Neuropsychiatric Interview). Trauma narrative age = age at participant's self-identified worst interpersonal event meeting DSM-5 Criterion A for PTSD. Childhood index trauma = trauma narrative occurring before age 16 years. Adult index trauma = trauma narrative occurring after age 16.

Distress

Consistent with our hypothesis, there was a significant main effect of childhood index trauma on distress (Model 1b; β = 6.151, p = .001), which improved model fit above the covariate model (Model 1a), such that individuals with a childhood index trauma evidenced greater levels of distress. In addition, childhood index trauma significantly interacted with the narrative cue to predict distress (Model 1c; β = -10.764, p = .002; Figure 1). Model 1c significantly improved model fit above Model 1b. Individuals with a childhood index trauma showed greater levels of distress to the neutral narrative cue, and individuals with an adult index trauma showed greater levels of distress to the trauma narrative cue. Childhood index trauma did not significantly interact with the beverage cue to predict distress (Model 1d; β = -1.566, p = .333), and did not improve model fit above Model 1b. See Table 2 for model fit statistics predicting distress and Table 3 for comparisons of model fit for distress.

FIGURE 1.

Interaction between childhood index trauma and narrative cues (i.e., trauma vs. neutral) and distress. Childhood Index Trauma = Trauma narrative occurring before age 16 years; Adult Index Trauma = trauma narrative occurring after the age 16; SUDS = Subjective Units of Distress Scale (on scale of 0–100). Error bars denote standard error.

TABLE 2.

Estimates, standard error, and model fits for distress

Variable	Model 1a	Model 1b	Model 1c	Model 1d
Fixed effects estimate (SE) (Intercept)	14.579(1.090)***	12.161(1.286)***	11.580(1.297)***	11.803(1.336)***
Narrative cue: Trauma	36.674(1.750)***	36.674(1.750)***	40.927(2.195)***	36.674(1.750)***
Beverage cue: Alcohol	2.110(0.791)**	2.110(0.791)**	2.110(0.791)**	2.729(1.015)**
Cumulative trauma	0.774(1.081)	0.423(1.062)	0.423(1.062)	0.423(1.062)
PTSD	6.264(1.148)***	6.548(1.115)***	6.571(1.113)***	6.549(1.115)***
PTSD × Narrative Cue	4.642(1.776)**	4.642(1.776)**	4.471(1.733)*	4.632(1.776)**
Childhood index trauma		6.151(1.886)**	7.622(1.946)***	7.055(2.104)**
Narrative Cue × Childhood Index Trauma			−10.764(3.490)**
Beverage Cue × Childhood Index Trauma				−1.566(1.613)
Random effects (Variance)
Intercept | Subject	154.69	139.26	138.11	138.44
Narrative Cue | Subject	482.74	482.74	455.13	482.74
Beverage Cue | Subject	27.78	27.78	27.78	27.21
Residual	89.12	89.12	89.12	89.12

Notes: PTSD = posttraumatic stress disorder symptoms as measured by the CAPS (Clinician Administered PTSD Scale for DSM-5); cumulative trauma = number of lifetime trauma types assessed via TLEQ (Traumatic Life Events Questionnaire); childhood index trauma = trauma narrative occurring before age 16 years.

^*p < .05;

^**p < .01;

^***p < .001.

TABLE 3.

Summary of comparisons between mixed-effect models to determine whether each predictor improved model fit over the previous model with respect to distress

Model comparison	Change in df	Deviance	χ2	p
Model 1b versus Model 1a(covariate model)	+1	6,243.5	9.891	.002
Model 1c versus Model 1b	+1	6,234.2	9.232	.002
Model 1d versus Model 1b	+1	6,242.6	0.895	.344

Notes: df = degrees of freedom; Model 1b = Narrative Cue + Beverage Cue + Cumulative Trauma + Study Site + PTSD × Narrative Cue + Childhood Index Trauma; Model 1c = Narrative Cue + Beverage Cue + Cumulative Trauma + PTSD × Narrative Cue + Childhood Index Trauma + Childhood Index Trauma × Narrative Cue; Model 1d = Narrative Cue + Beverage Cue + Cumulative Trauma + PTSD × Narrative Cue + Childhood Index Trauma + Childhood Index Trauma × Beverage Cue.

Subjective craving

As predicted, there was a significant main effect of childhood index trauma on craving (Model 2b; β = 0.656, p = .023), which significantly improved model fit above the covariate model (Model 2a), such that individuals with a childhood index trauma demonstrated greater levels of subjective alcohol craving. However, childhood index trauma did not significantly interact with the narrative cue to predict craving (Model 2c; β = 0.357, p = .148) and did not improve model fit above Model 2b. Further, childhood index trauma significantly interacted with the beverage cue to predict craving (Model 2d; β = -0.599, p = .011; Figure 2). Model 2d significantly improved model fit above Model 2b. Individuals with a childhood index trauma showed greater levels of craving to the neutral beverage cue than individuals with an adult index trauma. See Table 4 for model fit statistics predicting subjective craving and Table 5 for comparisons of model fit for subjective craving.

FIGURE 2.

Interaction between childhood index trauma and beverage cues (i.e., alcohol vs. water) and self-reported craving. Childhood Index Trauma = Trauma narrative occurring before age 16 years; Adult Index Trauma = trauma narrative occurring after age 16; Craving = self-reported craving for alcohol (on scale of 0–10). Error bars denote standard error.

TABLE 4.

Estimates, standard error, and model fits for subjective craving

Variable	Model 2a	Model 2b	Model 2c	Model 2d
Fixed-effects estimate (SE) (Intercept)	1.468***(0.199)	1.244***(0.217)	1.290***(0.220)	1.206***(0.218)
Narrative cue: Trauma	1 077*** (0.122)	1 077*** (0.122)	0.094***;(0.156)	1 077*** (0.122)
Beverage cue: Alcohol	1.404***(0.117)	1.405***(0.117)	1.404***(0.117)	1.642***(0.147)
Cumulative trauma	0.117(0.162)	0.067(0.161)	0.067(0.161)	0.067(0.161)
Study site	0.693*(0.285)	0.609*(0.285)	0.609*(0.285)	0.609*(0.285)
Alcohol use frequency	0.437**(0.146)	0.513***(0.148)	0.513***(0.148)	0.513***(0.148)
PTSD	0.754***(0.166)	0.784***(0.165)	0.783***(0.165)	0.784***(0.164)
PTSD × Narrative Cue	0.442***(0.123)	0.442***(0.123)	0.448***(0.122)	0 444*** (0.122)
Childhood index trauma		0.656*(0.287)	0.540(0.298)	0.753*(0.290)
Narrative Cue × Childhood Index Trauma			0.357(0.246)
Beverage Cue × Childhood Index Trauma				-0.599*(0.232)
Random effects (Variance)
Intercept | Subject	2.704	2.619	2.623	2.621
Narrative Cue | Subject	1.264	1.264	1.249	1.264
Beverage Cue | Subject	1.025	1.025	1.025	0.950
Residual	1.515	1.515	1.515	1.515

Notes: PTSD = Posttraumatic stress disorder symptoms as measured by the CAPS (Clinician Administered PTSD Scale for DSM-5); cumulative trauma = number of lifetime trauma types assessed via TLEQ (Traumatic Life Events Questionnaire); alcohol use frequency = total number of drinking days during the past 30 days on the TLFB (Timeline Followback); study site = data collection site (0 = a university in the mid-Atlantic, 1 = an urban university in the Midwest); childhood index trauma = trauma narrative occurring before age 16 years.

^*p < .05;

^**p < .01;

^***p < .001.

TABLE 5.

Summary of comparisons between mixed-effect models to determine whether each predictor improved model fit over the previous model with respect to self-reported craving

Model comparison	Change in df	Deviance	χ2	p
Model 2b versus Model 2a (covariate model)	+1	3,048.4	5.013	.025
Model 2c versus Model 2b	+1	3,046.3	2.032	.154
Model 2d versus Model 2b	+1	3,041.9	6.502	.011

Notes: df = degrees of freedom; Model 2b = Narrative Cue + Beverage Cue + Cumulative Trauma + Study Site + Alcohol Use Frequency + PTSD × Narrative Cue + Childhood Index Trauma; Model 2c = Narrative Cue + Beverage Cue + Cumulative Trauma + Study Site + Alcohol Use Frequency + PTSD × Narrative Cue + Childhood Index Trauma + Childhood Index Trauma × Narrative Cue; Model 2d = Narrative Cue + Beverage Cue + Cumulative Trauma + Study Site + Alcohol Use Frequency + PTSD × Narrative Cue + Childhood Index Trauma + Childhood Index Trauma × Beverage Cue.

Discussion

This study examined the impact of childhood index traumas on patterns of laboratory-assessed responses to trauma and alcohol cues in a sample of young adult drinkers with a history of interpersonal trauma exposure. First, the influence of childhood index traumas on trauma and alcohol cue-elicited distress was assessed, above and beyond relevant demographic and theoretical covariates. As hypothesized, individuals with a childhood index trauma displayed greater distress, regardless of cue combination. This finding is consistent with literature linking childhood trauma with stress response system dysregulation and increased risk for psychopathology (Dunn et al., 2013; Enoch, 2011; Kaplow & Widom, 2007; Kuhlman et al., 2013, 2015; Lansford et al., 2007). Further, childhood index trauma significantly interacted with the narrative cue to predict distress, such that individuals with childhood index traumas evidenced greater distress to neutral narrative cues, and individuals with adult index traumas evidenced greater distress to trauma narrative cues. Although this pattern was unexpected, individuals with a childhood index trauma may be primed to experience greater baseline anxiety and distress. This explanation accords with fear conditioning literature, in which individuals with PTSD demonstrate stronger conditioned fear responses to both conditioned fear and safety stimuli (Davis et al., 2000; Grillon & Morgan, 1999; Rothbaum & Davis, 2003), suggesting that they experience difficulty discriminating threatening and safe situations. Future work comparing conditioned fear responses among individuals with childhood and adult trauma histories would help explain the elevated distress observed in the childhood index trauma group.

On the other hand, adult index traumas may generate more distress because of greater vividness and recency of the trauma memory. Because prolonged-exposure therapy for PTSD relies on eliciting distress during recounting of the trauma memory (Eftekhari et al., 2006; Foa, 2011), prolonged-exposure modifications to enhance the vividness of childhood index traumas in treatment may be beneficial. Indeed, literature suggests that childhood trauma predicts greater PTSD symptom severity post–prolonged-exposure treatment (Hembree et al., 2004). However, the effect of augmented prolonged-exposure protocols on treatment outcomes for childhood trauma survivors with PTSD is relatively underinvestigated. To inform these clinical implications, future research is warranted to compare vividness ratings during imaginal or in vivo exposure to child and adult trauma memories. In contrast, childhood index trauma did not significantly interact with the beverage cue to predict distress. Differences in distress between childhood and adult index trauma groups may be better accounted for by exposure to an imaginal narrative cue than an in vivo alcohol cue.

In addition, the impact of childhood index traumas on trauma and alcohol cue-elicited subjective alcohol craving was examined, above and beyond an established base-fitting covariate model in this sample (Berenz et al., 2021). As predicted, individuals with a childhood index trauma endorsed greater alcohol craving than individuals with an adult index trauma. Comparable findings were reported in a study using a similar trauma and alcohol cue reactivity paradigm, wherein childhood trauma was associated with greater alcohol craving (Schumacher et al., 2006). Given the established role of craving in AUD (Wemm et al., 2019), efforts to understand links between childhood trauma and craving are warranted. One hypothesis is that childhood trauma impacts alcohol-related reward processing. This explanation is supported by literature on reward-related neuroadaptations as a consequence of childhood trauma, wherein greater alcohol reward responsivity is associated with problematic drinking behaviors (e.g., alcohol craving; Casement et al., 2015; Corral-Frías et al., 2015; Koob & Le Moal, 2005; Montagud Romero et al., 2018). Future studies to examine alterations in reward functioning following childhood trauma exposure are needed.

Contrary to expectation, childhood index trauma did not significantly interact with the narrative cue to predict craving beyond the simple effect of childhood index trauma on craving. Among young adult drinkers with interpersonal trauma histories, it may be that reactivity to the narrative, regardless of age at the time of the event, is more predictive of an acute craving response (Berenz et al., 2021). In contrast, childhood index trauma significantly interacted with the beverage cue to predict craving, such that individuals with a childhood index trauma showed greater craving to the neutral beverage cue than did individuals with an adult index trauma. This effect cannot be accounted for by the simple effect of childhood index trauma on alcohol craving more generally, or by how much individuals drink outside of the lab, as alcohol use frequency was covaried in the models. It is possible that the childhood index trauma group experienced generalization of alcohol craving to non-alcohol beverage cues, although replication and extension of this finding is needed to draw any firm conclusions.

The study findings should be interpreted in the context of limitations. First, the sample was selected on the basis of interpersonal trauma exposure; results may not extend to survivors of other trauma types (e.g., combat) or to individuals without trauma exposure. Second, study participants were young adults, because this age cohort has shown increased risk for the onset of AUD and PTSD (Kessler et al., 2005; O’Malley & Johnston, 2002; Overstreet et al., 2017; Read et al., 2012). These findings may not generalize to other age groups or to individuals with long-standing PTSD-AUD symptoms. Replication of the current study aims in an adult sample would inform the unique contribution of childhood trauma to processes implicated in the etiology of chronic AUD and/or PTSD. Third, the cumulative trauma index used in this study did not account for severity of individual trauma events, total number of events across trauma types, or lifetime childhood trauma history. The current models would be enhanced by evaluating the influence of childhood trauma load on the outcomes of interest. Relatedly, participants’ index trauma was used to categorize participants into childhood and adult groups; each group may have experienced both childhood and adult trauma. This limitation is particularly salient considering evidence that revictimization (e.g., exposure to child and adult traumas) is associated with greater distress and psychiatric risk, including PTSD and alcohol abuse (Classen et al., 2005; Kilpatrick et al., 1997). Replication of study aims among individuals exposed to only childhood or adult trauma types is warranted. Last, the cross-sectional design of the study prohibits causality or directionality from being established. Despite limitations, this study is the first to examine the unique influence of childhood index traumas on subjective distress and alcohol craving responses to a trauma and alcohol cue reactivity paradigm. Study findings highlight the utility of examining the contribution of childhood trauma to PTSD and AUD processes in laboratory and clinical settings.