The Role of Perceived Stress in the Relation between Childhood Maltreatment and Severity of Alcohol Use Disorder: A Mediation Analysis

Abstract

Introduction

Experiences of Childhood Maltreatment (CM) relate to relapse and lower treatment success in Alcohol Use Disorder (AUD), one of the most prevalent substance use disorders. However, the exact mechanisms of this relationship still remain unclear. This study examines perceived stress and “drinking to cope with negative affect” (coping) as possible mediators in this relationship. Moreover, it aims at uncovering the differential effects of the subtypes of CM.

Methods

N = 96 individuals (42% women; mean age 41 ± 13 years) including healthy controls and individuals with varying severity of AUD and CM completed the Alcohol-Dependence Scale, Childhood Trauma Questionnaire, Perceived Stress Scale and German Inventory of Drinking Situations. Mediation analyses including perceived stress as a mediator between CM (and subtypes) and severity of AUD, as well as a serial mediation of the relationship between CM and AUD severity by perceived stress and coping were conducted.

Results

Perceived stress significantly mediated the relation between CM and AUD severity and the serial mediation by perceived stress and coping turned out significant. Subtype-specific analyses did not yield significant results.

Conclusion

This study reinforces perceived stress as a potential mechanism in the relation between CM and AUD severity. Moreover, coping further mediated the relationship between CM and AUD severity. Our results suggest including screening for CM (subtypes) in clinical routine in order to individually emphasize interventions focusing on stress regulation, as well as on developing healthy coping mechanisms, in patients with AUD. This might prevent heightened stress sensitivity, relapse and further maintenance of AUD.

Introduction

Childhood maltreatment (CM) is a common risk factor regarding the development and maintenance of Substance Use Disorders (SUDs) [1]. It includes physical, emotional, or sexual abuse and physical or emotional neglect. Underlining the extent to which detrimental consequences of CM might affect society, a meta-analysis described prevalence rates of the general population of 12.7% for sexual abuse (SA), 22.6% for physical abuse (PA), 36.3% for emotional abuse (EA), 16.3% for physical neglect (PN) and 18.4% for emotional neglect (EN) [2]. Overall, CM has a negative impact on mental and somatic health [3] as it can interfere with children’s development [4], notably with regard to neuroregulatory systems [5]. Not only was it found that people with SUD are likely to have been faced with CM [6], but it was also observed to be a risk for relapse and related to lower success in treatment for SUD [7]. Moreover, in patients with Alcohol Use Disorder (AUD), a history of CM related to an earlier onset of drinking [8]. As AUD, along with tobacco use disorder, is the most prevalent substance use disorder across the globe [9] affecting approximately 237 million men and 46 million women in 2017 [10], a deeper understanding of this disorder is warranted to inform optimal treatment.

Additionally, different types of CM, such as abuse and neglect, result in specific outcomes [11–13] and impact alcohol use differently [14]. For example, emotional abuse was shown to be a significant predictor of alcohol consumption [14, 15] and, in particular, related to an earlier onset of AUD [16]. In patients seeking treatment for AUD, emotional abuse most strongly predicted the severity of the disorder [15]. Also, Shin and colleagues [17] found a significant mediation of psychological distress on the relation between emotional abuse and alcohol-related problems in a young adult sample. Conversely, other studies found that all CM types equally increase the risk of alcohol consumption [18, 19]. According to a systematic review from Grummit and colleagues [20], most studies, however, failed to include neglect in their childhood adversity assessment and, thus, missed an important aspect of CM. Patock-Peckham and colleagues [21] found the effect of physical neglect to be specifically associated with increased impaired control over drinking. It is also important to note that CM subtypes do not necessarily come alone, but that a cumulative effect of experiencing several types of CM in one’s childhood was also observed [5, 17].

While the literature on mediators and moderators of the relation between CM and AUD is still inconsistent and mostly does not consider specific subtypes of CM, several studies suggest that symptoms of anxiety, trauma, and psychosocial or emotional stress exert a mediating effect. Individuals using substances, as compared to the general population, had higher prevalence rates of CM, as well as higher rates of psychological stress [22]. Moreover, a study by Sinha and colleagues [23] found that in patients with AUD, as compared to social drinkers, stress and alcohol cues exposure produced an enhanced and persistent state of alcohol craving. This state was marked by increased anxiety and negative emotions, which in turn were predictive of craving and relapse [23]. However, psychosocial stress alone might not be able to explain the whole relation between CM and AUD severity, since a single mediator could miss underlying mechanisms contributing to the relation [20]. Previous studies observed mechanisms to cope with psychosocial stress (e.g., states of negative affect) to be of relevance, as individuals with CM were more likely to start drinking to cope with negative feelings [24]. Temmen and Crockett [25] found that chronic life stress and drinking motives were both associated with alcohol consumption, however, men and women differed regarding the type of chronic stress (occupational stress for men, psychosocial stress for women). This followed the self-medication hypothesis [26], stating that individuals who experience distressing situations or heightened perceived stress tend to use substances, notably alcohol, to cope with and reduce negative emotions that may arise from experiencing psychosocial stress. This was shown in a longitudinal study by Park and colleagues [27], who examined adolescents using self-report questionnaires including trauma symptoms and drinking motives. The results suggest that trauma symptoms influence drinking motives - both mediating the relation between CM and AUD.

Thus, current literature hints toward a mediating effect of stress in the relation between the overall experience of CM and AUD, while drinking to cope with negative affect (coping) might be of additional relevance. Some studies suggested a specific influence to this end regarding subtypes of CM, specifically emotional abuse as discussed above. Therefore, the current project aimed to examine the mediating effects of perceived stress and coping in the relation between CM and AUD severity in an adult sample. In an attempt to examine the specific effects of each type of CM, separate analyses were conducted. Additionally, sex was included in the analyses, as previous studies observed CM to have a greater impact on men [11, 28], or women [6]. However, differential effects [29] and a lack of significant differences between sexes [30, 31] were also reported.

We hypothesized that (1) perceived stress mediates the relation between CM and AUD severity; (2) the type of CM exerts a specific influence on AUD severity, namely emotional abuse being of highest relevance; (3) perceived stress further mediates the relation between CM and AUD severity by its influence on drinking motives.

Materials and Methods

Procedure and Participants

N = 102 individuals participated in this study with N = 96 (42% women) presenting complete data. For data collection, they were invited to the Central Institute of Mental health, Mannheim, Germany, between January 2019 and September 2023. The studies were registered with ClinicalTrials.gov (identifiers: NCT03758053, NCT05048758). The local Ethics Committee of the Medical Faculty Mannheim, Heidelberg University, Germany, approved the study procedure (Approval No. 2018-560N-MA) and all participants gave their written informed consent prior to study participation. Participants were also informed about the specific study goals in a debriefing session and were invited to declare their interest in receiving study publications.

Flyers, social media, or newspaper reports for public announcements were used to recruit prospective participants with varying severity of alcohol use and CM. To this end, interested individuals were screened for their alcohol consumption and included in case they reported heavy drinking behavior (alcohol/day ≥40 g [female], 60 g [male] on min. 5 days/week). In addition, patients currently in treatment for AUD were recruited from the Central Institute of Mental Health, Mannheim, Germany to ensure the inclusion of individuals with a broad range of AUD severity. Further, individuals with none to minimal alcohol consumption (i.e., alcohol/day ≤12 g [female], 24 g [male] on up to 5 days/week) were included as healthy controls, i.e., the lower end of the severity distribution of AUD. In the original study design, individuals were invited twice, for a baseline appointment and a second examination day (see online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000539711). Only data from the first appointment will be used here for the cross-sectional analysis. At the baseline appointment, written informed consent was obtained and a breath alcohol test as well as drug- and pregnancy urine screenings were conducted. Subsequently, sociodemographic data was collected and the Structured Clinical Interview for the fourth version of the Diagnostic and Statistical Manual of Mental Disorders (SCID-IV) [32] interview was conducted to examine possible mental disorders. Additionally, AUD-criteria were defined according to DSM-5 [33]. Drinking behaviour during 90 days prior to the examination was assessed using the FORM-90 interview [34].

Participants were eligible for the current analyses, if they (1) were between 18 and 65 years old, and did not report (2) diagnoses of any mental disorders in the last 12 months other than nicotine use disorder or a lifetime severe mental disorder such as schizophrenia or bipolar disorder, (2) current severe neurological or somatic condition or history of severe head trauma (3) treatment with psychotropic medication. They were excluded in case of (4) positive drug and pregnancy screenings. After the baseline appointment, participants received an individual link via e-mail and completed several questionnaires online using the web-based software EvaSys (Electric Paper Evaluationssysteme GmbH, Lueneburg, Germany).

Measures

Sociodemographic information included age, sex (“man/woman”), education, employment status, and family status. Besides the Beck Depression Inventory (BDI, [35, 36]) and Fagerström Test for Nicotine Dependence (FTND, [37]), additional psychometric measures were conducted. The Alcohol Dependence Scale (ADS, Cronbach’s alpha (α) = 0.92 [38]) is a 25-item self-report instrument. It is used to assess the severity of alcohol dependence during the previous 12 months, as well as problems related to alcohol drinking; with higher scores reflecting higher alcohol dependence severity (or AUD). The Childhood Trauma Questionnaire (CTQ, α = 0.80 [39, 40]) is a Likert-type self-report instrument used to assess the severity of childhood maltreatment before the age of 18. It consists of 28 items containing emotional, physical, sexual abuse and emotional and physical neglect subscales. Total scores for each subtype ranges from 5 to 25, with higher scores representing more severe maltreatment. The Perceived Stress Scale (PSS, α = 0.84 [41, 42]) is a self-report measure assessing perceived, individual stress in the last month using 10 items on a Likert-scale. The German Inventory of Drinking Situations (DITS40, α = 0.84–0.98 [43, 44]) is a Likert-type self-report questionnaire with 40 items. It assesses the situations in which people with Alcohol use disorders drank excessively in the past year. Two factors represent drinking either to cope with negative affect and situations, or in (positive) social situations. Only the negative affect scale was used in the subsequent analyses.

Statistical Analysis

SPSS (Statistics for Windows, Version 27.0. IBM Corp., Armonk, NY, USA) was used to perform all analyses. Sum scores of corresponding questionnaires were calculated according to their manuals. Bivariate correlation analyses were conducted to assess the relationship between different variables, such as perceived stress, CM, and AUD severity. The PROCESS macro for SPSS [45] was used for subsequent mediation and serial mediation analyses. Bootstrapping (5,000 subsample) was applied to this small sample to address the distribution of the data and a confidence interval of 95% was defined for the indirect effects [46]. The following models were used: Model 4 captured the simple mediation of the relationship between CM and AUD severity by perceived stress. Additional mediation analyses of the subscales of CTQ as predictors were performed including the remaining subscales as covariates (except for SA due to a lack of significant correlation with other subscales) [47]. This allows to examine a subscale-specific influence of CM. Model 6 was used to analyse the serial mediation including PSS and DITS40 (negative affect drinking). For all analyses, sex and age were included as covariates. Sex-separated, exploratory mediation analyses including CM as a mediator were further conducted.

Results

Sample Description

The sample included 96 participants (58% males and 42% females; N = 6 individuals were excluded due to missing questionnaire data regarding PSS or CTQ). Overall, individuals exhibited only mild to moderate severity regarding the psychometric assessments, e.g., CTQ or ADS (see online suppl. materials 2 and 3 including Table 1; Fig. 2). Of note, prevalence and severity of SA was very low and this subscale did not correlate significantly with the other four subscales (see online suppl. material 4). For bivariate correlations of all variables used in the mediation analyses, please see the online supplementary Table 2 of the supplementary material 4. As displayed in Table 1, no significant differences between men and women were observed except for the amount of alcohol consumption during a drinking day.

Table 1.

Sample description including sociodemographic and clinical variables

	Males, mean (SD)	Females, mean (SD)	Statistics
N	56	40
Age, years	41.1 (13.6)	40.4 (12.8)	t (94) = 0.263, p = 0.721
Marital status (married:divorced:single)	13:9:34	8:7:25	χ2 (2) = 151, p = 0.927
Living status (alone:together with others)	26:30	15:25	χ2 (1) = 0.760, p = 0.383
Years of education	15.1 (2.7)	15.6 (2.5)	t (94) = −1.011, p = 0.315
Smoker (yes:abstinent:no)	17:2:37	11:1:28	χ2 (2) = 0.204, p = 0.903a
DITS40 social drinking	12.1 (7.7)	14.2 (10.3)	t (68.5) = −1.111, p = 0.270
DITS40 negative affect	6.2 (7.2)	8.9 (10.1)	t (66.7) = −1.431, p = 0.157
PSS	14.5 (7.5)	15.2 (9.4)	t (94) = −0.356, p = 0.723
CTQ overall	39.5 (14.6)	42.2 (15.8)	t (94) = −0.949, p = 0.345
CTQ Emotional abuse	8.7 (4.8)	10.2 (5.2)	t (94) = −1.404, p = 0.164
CTQ Physical abuse	6.7 (3.0)	6.5 (3.0)	t (94) = 0.331, p = 0.742
CTQ Sexual abuse	5.9 (3.5)	5.8 (2.4)	t (94) = 0.051, p = 0.960
CTQ Emotional neglect	11.0 (5.5)	12.3 (5.8)	t (94) = −1.090, p = 0.278
CTQ Physical neglect	7.2 (2.7)	7.6 (3.2)	t (94) = −0.699, p = 0.486
DSM-5 criteria	4.7 (3.4)	3.9 (3.5)	t (94) = 1.140, p = 0.257
Individuals with:without AUD	44:12	24:16	χ2 (1) = 3.90, p = 0.048
ADS	9.8 (7.6)	9.7 (8.1)	t (94) = 0.060, p = 0.953
BDI	8.1 (9.2)	9.2 (9.6)	t (93) = −0.550, p = 0.583
FORM-90 alcohol per day (g)	135.9 (159.7)	73.7 (66.1)	t (77.8) = 2.572, p = 0.012

Independent samples t tests or Welch-tests, as well as χ² tests were performed. The categorization into “with” and “without AUD” was performed according to the DSM-5 criteria that were fulfilled (more than two out of 11 criteria indicate a mild AUD).

SD, standard deviation; g, grams; n, sample size; ADS, Alcohol Dependence Scale; CTQ, Childhood Trauma Questionnaire; FORM-90, amount of alcohol consumption per drinking day over the last 90 days; DITS40, German Inventory of drinking situations, PSS, Perceived Stress Scale; BDI, Beck’s Depression Inventory.

^aAt least one cell with expected frequency <5.

Mediation of the Relationship between CM and AUD Severity by Perceived Stress

A mediation analysis was conducted to test the mediating effect of perceived stress on the relation between CM and the severity of AUD while controlling for age and sex (Fig. 1). The direct effect (path c) was found significant b = 0.142, p < 0.001 (SE = 0.053, 95% CI: [0.038; 0.246]). Perceived stress was positively related to CM (b = 0.168, SE = 0.055, p < 0.01, 95% CI: [0.059; 0.276]). A positive relation was shown between perceived stress and AUD severity (b = 0.440, SE = 0.089, p < 0.001, CI: [0.263; 0.617]). After introducing the indirect path, the direct effect (c′) between CM and AUD severity decreased to non-significance (b = 0.068, SE = 0.049, p = 0.169, CI: [−0.030; 0.166]). We found that the relationship between childhood maltreatment and AUD severity was fully mediated by PSS, indirect effect ab = 0.074, 95% CI: [0.023; 0.139].

Fig. 1.

Mediating effect of perceived stress on childhood maltreatment using CTQ sum score and relevant subscales, and AUD. N = 96. Controlled for participant’s sex and age. c, direct effect before mediation; c′, direct effect after mediation; sum, sum score; EA, emotional abuse; PA, physical abuse; EN, emotional neglect. PROCESS Bootstrap-analysis 5000 Model 4 (Hayes 2013). * p < 0.05, ** p < 0.01.

Sex-separate mediation analyses (online suppl. material 5, Fig. 3 and 4) did yield significant results for women only, i.e., a significant indirect effect (ab = 0.089, 95%-CI: [0.006; 0.211]).

Mediation of the relationship between CM and AUD severity by perceived stress with regards to subtypes of CM

Models were tested separately for each type of maltreatment. Again, perceived stress was defined as a mediator, and AUD severity as the outcome variable. All models were controlled for age and sex. When including remaining subscales of CM as additional covariates in the mediation analyses, no significant results remained. Uncorrected analyses yielded significant results for emotional and physical neglect, and emotional abuse. All results are displayed in Table 2.

Table 2.

Subtype-specific mediation analyses

X (predictor)	a path, X → M (CM → PSS)	b path, M → Y (PSS → AUD severity)	c path, total effect X → Y (CM → AUD severity)	c′ path, controlling for the indirect mediational path (CM → AUD severity)	ab path, indirect mediational path
EA controlleda	b = 0.462, p = 0.062	b = 0.429, p < 0.001	b = 0.486, p < 0.05	b = 0.288, p = 0.175	ab = 0.198
	SE = 0.244, 95% CI: (−0.023; 0.948)	SE = 0.090, 95% CI: (0.251; 0.607)	SE = 0.230, 95% CI: (0.029; 0.944)	SE = 0.210, 95% CI: (−0.130; 0.706)	95% CI: (−0.003; 0.434)
EA	b = 0.517, p < 0.01	b = 0.416, p < 0.001	b = 0.507, p < 0.001	b = 0.292, p < 0.05	ab = 0.215
	SE = 0.155, 95% CI: (0.209; 0.824)	SE = 0.089, 95% CI: (0.240; 0.593)	SE = 0.146, 95% CI: (0.217; 0.797)	SE = 0.140, CI: (0.015; 0.570)	95% CI: (0.070; 0.413)
PA controlleda	b = −0.201, p = 0.579	b = 0.429, p < 0.001	b = 0.233, p 0.493	b = 0.319, p = 0.297	ab = −0.086
	SE = 0.360, 95% CI: (−0.9,152; 0.514)	SE = 0.090, 95% CI: (0.251; 0.607)	SE = 0.339, 95% CI: (−0.441; 0.908)	SE = 0.304, 95% CI: (−0.286; 0.924)	95% CI: (−0.374; 0.277)
PA	b = 0.479, p = 0.092	b = 0.446, p < 0.001	b = 0.708, p < 0.001	b = 0.494, p < 0.05	ab = 0.214
	SE = 0.282, 95% CI (−0.080; 1.039)	SE = 0.085, CI (0.277; 0.615)	SE = 0.261, 95% CI (0.189; 1.227)	SE = 0.234, CI (0.030; 0.959)	95% CI (−0.018; 0.198)
EN controlleda	b = 0.155, p = 0.548	b = 0.429, p < 0.001	b = −0.199, p = 0.415	b = −0.266, p = 0.227	ab = 0.067
	SE = 0.258, 95% CI: (−0.257; 0.667)	SE = 0.090, 95% CI: (0.251; 0.607)	SE = 0.243, 95% CI: (−0.682; 0.284)	SE = 0.218, 95% CI: (−0.699; 0.168)	95% CI: (−0.160; 0.332)
EN	b = 0.423, p < 0.01	b = 0.462, p < 0.001	b = 0.276, p = 0.060	b = 0.080, p= 0.549	ab = 0.196
	SE = 0.150, 95% CI: (0.126; 0.720)	SE = 0.089, 95% CI: (0.285; 0.640)	SE = 0.145, 95% CI: (−0.012; 0.564)	SE = 0.134, CI: (−0.185; 0.346)	95%-CI: (0.052; 0.380)
PN controlleda	b = 0.053, p = 903	b = 0.429, p < 0.001	b = 0.243, p = 0.555	b = 0.220, p = 0.551	ab = 0.023
	SE = 435, 95% CI: (−0.810; 0.917)	SE = 0.090, 95% CI: (0.251; 0.607)	SE = 0.410, 95% CI: (−0.572; 1.058)	SE = 367, 95% CI: (−0.510; 0.950)	95% CI: (−0.396; 0.344)
PN	b = 0.653, p < 0.05	b = 0.456, p < 0.001	b = 0.564, p < 0.05	b = 0.267, p= 0.282	ab = 0.297
	SE = 0.285, 95% CI: (0.086; 1.219)	SE = 0.088, 95% CI: (0.281; 0.630)	SE = 0.272, 95% CI: (0.024; 1.104)	SE = 0.247, CI: (−0.223; 0.758)	95%-CI: (0.055; 0.587)

No subtype-specific results were observed when including remaining subtypes of CM as covariates.

EA, emotional abuse; PA, physical abuse; EN, emotional neglect; PN, physical neglect.

^aControlled for EA, PA, EN, PN, respectively. Sex and age were included as covariates in all analyses.

Mediation of the Relationship between CM and AUD Severity by Perceived Stress and Coping

To test our third hypothesis, a serial mediation analysis of perceived stress and negative affect drinking motives on CM and AUD severity, controlling for sex and age, was performed (Fig. 2). The direct effect between CM and AUD severity was found significant, b = 0.142, p > 0.01 (SE = 0.053, 95% CI: [0.038; 0.246]). Perceived stress was significantly positively related to childhood maltreatment (b = 0.168, SE = 0.055, p < 0.01, 95% CI: [0.059; 0.276]). There was a significant positive relation between coping and perceived stress (b = 0.485, SE = 0.084, p < 0.001, 95% CI: [0.318; 0.653] and a significant positive relation between CM and coping (b = 0.112, SE = 0.046, p < 0.05, 95%: CI [0.020; 0.204]). AUD severity was positively related to perceived stress (b = 0.211, SE = 0.094, p < 0.05, 95% CI: [0.024; 0.397]), and coping (b = 0.472, SE = 0.100, p < 0.001, 95% CI: [0.273; 0.671]). After introducing the indirect path, the direct effect (c′) between CM and AUD severity decreased to non-significance (b = 0.016, SE = 0.046, p = 0.736, CI: [−0.075; 0.106]). The relationship between CM and AUD severity was found to be serially mediated by perceived stress and coping, indirect effect ab = 0.038, 95%-CI: [0.011; 0.078].

Fig. 2.

Serial mediation of perceived stress and coping mechanisms on CM and AUD. N = 696. Controlled for participant’s sex and age. c, direct effect before mediation; c′, direct effect after mediation. PROCESS Bootstrap-analysis 5000 Model 6 (Hayes 2013). * p < 0.05, ** p < 0.01.

Discussion

While CM is known to increase the risk for subsequent AUD, mechanisms contributing to the development and maintenance of this disorder are still not fully understood. The present study examined the mediating effect of perceived stress on the relation between childhood maltreatment and the severity of AUD and accounted for subtype-specific effects, drinking motives, age, and sex. We confirmed a mediation of the relationship between CM and AUD severity by perceived stress. Additionally, a serial mediation of the relationship between CM and AUD severity by perceived stress and coping was observed. Contrary to hypothesis 2, we were not able to confirm subtype-specific effects in our sample.

The first hypothesis predicted a mediation of the relationship between CM and AUD severity by perceived stress in a sample of individuals with varying severity of AUD. In line with previous studies, our results suggest that perceived stress may play a mediating role in the relation between CM and AUD severity [23]. Indeed, Sinha and colleagues [23] demonstrated that AUD patients, as compared to social drinkers, showed increased sensitivity to stress and persistence of distress following exposure to stressful cues. Seo and colleagues [48] explain that exposure to trauma is associated with impaired cortisol levels at awakening, as well as increased stress sensitivity of the amygdala. This finding is consistent with other studies finding an enlarged amygdala and hyper responsiveness to threatening stimuli, and reduced hippocampal volume [49]. This is also associated with alterations in memory, stress, and emotional regulation in people who experienced early life stress [50, 51]. Yet, the age of onset as well as the duration of CM can also influence neural activity and structure (see [52]). This might lead to inconsistent results or null findings when only including the overall CM severity in the analyses. Following our second hypothesis, our data does not fully support a differential influence of the type of CM. While the exploratory analyses seem to support a mediation for specific subtypes such as emotional abuse, emotional neglect, and physical neglect; the analyses including the remaining subscales did not reveal any significant results. Thus, our exploratory results have to be considered with caution. Overall, however, a subtype specific influence of CM has been reported previously by Shin and colleagues [17], who found emotional abuse to be specifically of relevance for the mediation of psychological distress on alcohol use in young adults. Additionally, these results are partly in accord with those of Betz and colleagues [12] who found that CM domains related to threat (i.e., emotional and physical abuse) are associated with perceived stress and helplessness. Burns and colleagues [53] have found emotional abuse to be specifically related to emotional regulation difficulties, which, could be explained by the chronic exposure to negative emotions [53]. Similarly, Gama and colleagues [54] found CM to be related to cognitive processing impairments during emotional situations which relates to disruptions in emotional regulation abilities. This would then lead to the use of alcohol as a way to cope with negative affect. Regarding emotional neglect, Strine and colleagues [29] found emotional neglect to mediate the relation between CM and alcohol problems. Our lack of significant results when examining subtype-specific effects of CM might result from either an actual lack of specificity or certain statistical aspects. Regarding the latter, our sample exhibited overall mild to moderate severity of CM which might lead to reduced power regarding the (possibly small) subtype-specific effect. Also, the sample size (N = 96) is quite small as compared to other studies reporting subtype-specific effects of, i.e., emotional abuse, by Schwandt and colleagues [15] who included 280 individuals with AUD.

We confirmed our third hypothesis which predicted a serial mediation of perceived stress and negative affect coping mechanism. In our sample, we assessed drinking alcohol to cope with (negative) affect following, e.g., criticism, inter- and intrapersonal conflicts, unsuccessful problem solving. This learned coping mechanism can be explained by the anxiolytic effect of alcohol [55], which, in addition to the rewarding effect, reinforces conditioned learning. Additionally, alcohol consumption alters neural reward, affect, and control circuitries which further facilitate the consumption of alcohol, especially during stressful situations [56]. Previously learned, maladaptive strategies, such as drinking to cope, might thus be more easily available during stressful situations. Our results are in line with a previous study from Park and colleagues [27] who found a serial mediation by post-traumatic stress symptoms and coping in an adolescent sample. Similarly, Hogarth and colleagues [31] found that the relation between childhood abuse and substance misuse is mediated by using substances to cope with negative affect. The aspect of coping (with stress and negative affect) as an additional mediator to the relation between CM and AUD is further supported by previous literature [24–26]. Our results thus confirm previous literature and extend the findings to a heterogeneous sample of individuals with varying severity of AUD.

Of note, we conducted additional sex-separated analyses to investigate possible differential effects of sex (see supplementary material 5), as previous studies observed differential effects of CM between men and women [6, 11, 28, 29]. While the mediation in the women’s sample was found significant, the men’s sample did not show a significant mediation. Accordingly, sex was kept as a covariate as it is known to be a variable of interest [57, 58].

Some limitations need to be considered. Firstly, the cross-sectional design of our study does not allow for a causal interpretation of the results. A longitudinal design with the same variables is needed to confirm the results, and could additionally evaluate bi-directionality. In addition, the retrospective self-report assessment of CM is subject to recall bias. Further, not only is it shown that stress affects AUD [59], but also that AUD has an effect on stress pathways, notably on the extended amygdala [60]. Our study only includes the suggestion for a one-way path from stress to AUD severity but does not account for the effect of AUD on brain areas, specifically the stress system. Limitations also extend to questionnaire characteristics. The ADS captures AUD severity by also assessing the physical aspects of AUD. These symptoms do not necessarily occur in individuals with no or mild AUD. Consequently, the ADS might not be the most suitable means to assess the full dimensional range of AUD severity as some items are not suitable for healthy individuals. The results of this study should therefore be interpreted accordingly. Regardless, these data set can be used for future analyses. Yet, subsequent studies are required to test the subtype-specific findings in a larger sample while fully controlling for all subtypes of CM, respectively, as a subtype-specific effect might be rather small. Lastly, CM is a complex concept to investigate, notably as other types of adverse childhood experiences exist. Prior studies have found the age and the duration of childhood maltreatment to affect brain structure development [1, 52]. It should also be further investigated whether these factors might have direct differential effects on alcohol consumption, as compared to an overall measure of CM, namely the CTQ sum score. Further, previous research suggested additional mediators such as impulsivity [1, 6], depression [1], and ethnicity [24, 61] to be of relevance. As the present study sample excluded individuals with severe comorbid mental disorders, limitations to this end occur as well. Similarly, the distribution of severity of CM and AUD was skewed in our sample. The study also included individuals with minimal CM as well as individuals with none to minimal amount of alcohol consumption (see online suppl. Table 1) which represented a fairly large share of the data. Therefore, several lines of research remain to be investigated to fully understand the mechanisms between CM and AUD.

Despite these limitations, our results add to the body of knowledge on the mental health consequences of childhood maltreatment and have several implications. Firstly, early identification and interventions for CM could help prevent heightened stress sensitivity and its negative consequences on health. Therefore, it is essential not only screening for CM in the treatment of SUD, but also screening in youth with CM to identify the type of CM and assess their risk for later development of SUD [62]. Secondly, as demonstrated by serial mediation, alternative coping strategies during negative affect should be trained, e.g., during cognitive behavioral therapy. As mentioned by Hyman and colleagues [22], interventions focusing on stress tolerance and desensitization, and emotion regulation skills would be of particular interest for patients with a history of CM. An example would be using mindfulness-based approaches, as Garland and colleagues [63] found those methods to positively influence the treatment of AUD through the focus on emotional and stress regulation. Besides, cognitive pathways involved in stress as well as neuroendocrine alterations need to be taken into account.

Conclusion

The present study set out to examine the mediating role of perceived stress in the relation between CM and AUD. While perceived stress was found to significantly mediate the effect of CM on AUD, coping was found to add explanatory value in the serial mediation model. These results support the idea that interventions on stress regulation and coping strategies should be included in targeted prevention and treatment methods for AUD.

Statement of Ethics

The local Ethics Committee of the Medical Faculty Mannheim, Heidelberg University, approved the study procedure (Approval No. 2018-560N-MA) and all participants gave their written informed consent prior to study participation.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

The project was supported in part by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) Grant No. 324164820 GRK2350/1 and 402170461. The funders had no role in the study design, data collection, analysis, decision to publish, or preparation of the manuscript.

Author Contributions

Design the current study: S.G. Analyzed the data: E.B., S.G. Supported the data analysis: M.D. Wrote the paper: E.B., S.G. Supported the writing of the paper: M.D., S.V.‐K. Interpreted the data: E.B., M.D., S.V.‐K., S.G. Conceived and designed the original experiments: F.K., S.V.‐K., S.G. Procured the funding of the original studies: F.K., S.V.‐K. Commented on the manuscript and provided intellectual input: E.B., M.D., C.T., F.K., S.V.‐K., S.G.

Funding Statement

The project was supported in part by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) Grant No. 324164820 GRK2350/1 and 402170461. The funders had no role in the study design, data collection, analysis, decision to publish, or preparation of the manuscript.

Data Availability Statement

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further enquiries can be directed to the corresponding author S.G.