Experienced childhood trauma (CT) has significant negative consequences for cognitive, psychological, and physical health development (Jaffee, 2017) and is an environmental risk factor that elevates the risk of developing alcohol use disorders (AUDs) (Afifi et al., 2012; Fenton et al., 2013). CT includes harmful acts such as sexual, physical, and emotional abuse as well as physical and emotional neglect, i.e., caregiver failure to respond to a child’s emotional and physical needs. The global prevalence of sexual abuse is estimated at 11.8% (Stoltenborgh et al., 2011) and of physical abuse at 22.6% (Stoltenborgh et al., 2013a). Regarding emotional abuse, the prevalence in the European Union is 29.1% (Sethi et al., 2013). The global prevalence rates of physical neglect (16.3%) and emotional neglect (18.4%) are less reliable because few epidemiological studies have investigated these two aspects of CT in the general population (Stoltenborgh et al., 2013b).

Experienced CT is common in patient samples with AUDs, being reported by 22–74% (Huang et al., 2012; Langeland & Hartgers, 1998; Lotzin et al., 2016; Schückher et al., 2019). The prevalence rates may vary between studies because CT has been investigated in different samples of AUD patients (e.g., male/female, inpatient/outpatient, and AUD/AUD + drug abuse + psychiatric comorbidity) (e.g., Dom et al., 2007; Gilbert et al., 2009; Magnusson et al., 2012; Zlotnick et al., 2006).

Few studies have examined how experienced CT affects treatment outcome and what impacts psychological aspects have. The few studies investigating CT in relation to treatment outcome have found that alcohol/substance-dependent patients with experienced CT were less adherent to treatment (Jaycox et al., 2004), relapsed more often (Farley et al., 2004; Walitzer & Dearing, 2006), and were abstinent for a shorter period (Greenfield et al., 2002). In a 12-month follow-up, a smaller proportion of alcohol-dependent women with a history of childhood abuse reported abstinence than did a non-abused group of alcohol-dependent women (Schückher et al., 2019).

AUD patients with experienced CT are overrepresented in terms of impaired self-efficacy and self-directedness (self-control), as well as anxiety and depressive symptoms, compared with patients without experienced CT (Berent et al., 2018; de Carvalho et al., 2015; Jaffee, 2017). Major depression and anxiety disorder affect treatment outcome negatively (Kushner et al., 2005; Samet et al., 2013).

Self-efficacy and self-directedness can be seen as “tools” for dealing with demanding life situations. Self-efficacy and self-directedness develop during childhood when the child interacts with important persons inside the family and with important persons outside the family, such as teachers (Bandura, 1986; Cloninger et al., 1993). A socially supportive environment during childhood and adolescence can reduce the risk of psychopathology despite experienced CT (Jaffee, 2017).

Self-efficacy represents an individual’s imagined ability to engage in desirable behavior (Bandura, 1986). People high in self-efficacy believe that their own decisions and behavior shape their lives, i.e., that they have control over their own lives; people low in self-efficacy believe the opposite. High self-efficacy is associated with beneficial AUD treatment outcome, probably through the use of coping strategies to avoid relapses (Greenfield et al., 2000; Trucco et al., 2007). The Alcohol Abstinence Self-Efficacy Scale (AASE) has been developed specifically to examine self-efficacy related to high-risk situations for consuming alcohol (Di Clemente et al., 1994).

Self-directedness is a personality trait in Cloninger’s biopsychosocial theory of personality (Cloninger et al., 1993). It consists of an individual’s ability to steer behavior through self-control and self-image. Low self-directedness is strongly associated with experienced CT (de Carvalho et al., 2015). High self-directedness is associated with less alcohol consumption during alcohol-treatment follow-up (Foulds et al., 2016) and with alcohol-treatment retention (Arnau et al., 2008).

The aim of this study was to investigate whether experienced CT has an independent effect on alcohol-treatment outcome (measured by DSM-IV criteria and total abstinence) 2.5 and 5 years after treatment initiation.

We hypothesized that: 1) individuals with AUDs and experienced CT would have lower self-efficacy, lower self-directedness, and poorer mental health; and 2) in these individuals, CT would have an independent effect on treatment outcome, when controlling for other significant variables such as self-efficacy, self-directedness, and mental health.

Materials and Methods

The data used were collected as part of the Gothenburg Alcohol Research Project (GARP), an ongoing multidisciplinary, longitudinal project investigating risk factors and factors influencing treatment outcome (Berglund et al., 2008, 2013; Dahlgren et al., 2011; Fahlke et al., 2012). The participants in this study (n = 349; men n = 263 [75.4%], women n = 86 [24.6%]) were recruited between 2004 and 2012 from three alcohol-treatment settings: two traditional 12-step settings (12-step in-patient care, n = 132 [37.8%]; 12-step out-patient care, n = 113 [32.4%]) and one municipally sponsored setting (individual and group psychodynamic therapy and relapse prevention, n = 104 [29.8%]). The two 12-step treatment facilities are owned by a foundation whose members consist of companies or parts of the public sector in western Sweden. When employees have alcohol problems, they may be offered this 12-step treatment. The third treatment facility is municipally owned in western Sweden’s largest city, where people can contact reception if they have alcohol problems

Participants and Procedure

Before starting treatment, individuals were screened by a psychiatrist at the treatment facilities. If they had a severe psychiatric comorbidity (e.g., schizophrenia or bipolar disorder) or drug dependency, they were then referred to psychiatric addiction healthcare. Those who would begin treatment were invited to participate in the study; if they agreed, they underwent a structured interview and completed self-rating questionnaires within 1–3 weeks of start of treatment. Inclusion criteria in the study were adult individuals (>18 years old) with alcohol abuse or dependence according to DSM-IV, the diagnostic system current during the first data collection (2004–2012). The interviews were conducted by a doctoral student in psychology, a licensed psychologist, or a psychiatrist, all from the GARP research group.

There were 323 participants (92.6%) with alcohol dependence and 26 (7.4%) with alcohol abuse. The prevalence of experienced CT was the same in those with alcohol dependence and those with alcohol abuse, although the latter had better mental health, self-directedness, and self-efficacy (data not shown). The length of treatment was: 1) variable in the municipally sponsored setting; 2) 28 days high-intensive + one year aftercare in the in-patient care 12-step setting; and 3) one year low-intensive (once a week) in the out-patient care 12-step setting. The participants ranged from 21 to 71 years old (47.8 ± 10.3 years). The follow-up interviews occurred 2.5 and 5 years after treatment entry. The drop-out rate between the baseline and follow-up interviews was 29.2% (n = 102, including one deceased) 2.5 years after treatment initiation and 41.8% (n = 159, including 13 deceased) 5 years after treatment initiation. There were no significant differences between the 2.5- and 5-year drop-outs regarding experienced CT and background variables, with one exception (see Table 1). In those who remained at the 2.5-year follow-up, there were more than expected in the municipally sponsored setting (34.0%) and fewer than expected in the 12-step out-patient setting (29.6%) (χ2 = [2, n = 349] = 7.53, p < 0.05, φC = 0.02).

Table 1 Differences in background variables, experiences of CT and alcohol-related variables between drop-outs and those who have remained in the study (2 ½ year and 5 year). Data are presented as means (± SD) or frequencies (% in parentheses)

Informed consent was obtained from all participants. The study was approved by the regional ethical board at the University of Gothenburg (No. 487–03).

Measures

Addiction Severity Index

The Addiction Severity Index (ASI) is a standardized structured interview used to assess alcohol and drug abuse and dependence in adults (McLellan et al., 1992). It also covers mental health and functioning.

The ASI interviews were conducted when the participants started treatment, and 2.5 and 5 years thereafter. We gathered the following background data from the ASI interview: age, sex, education, relationship status, employment, and living situation.

For the purposes of GARP, the ASI interview was supplemented with additional questions regarding alcohol dependence and alcohol abuse (according to DSM-IV criteria). The participants were also asked to describe their alcohol consumption during an average week in the last 12 months. Questions regarding alcohol consumption in the follow-up interview included: “Have you drunk any alcohol at all during the last 12 months?” and “How much have you drunk in an average week during the last 12 months?”

Childhood Trauma Questionnaire

The Childhood Trauma Questionnaire – Short Form (CTQ-SF; Bernstein et al., 2003), a 28-item self-reporting scale with five subscales, was used to assess the participants for CT. The Emotional Abuse (EA) subscale captures non-physical actions leading to negative self-image and a feeling of lack of worth in the child; these actions can also include constant harsh criticism or expressions of hatred toward the child. The Physical Abuse (PA) subscale captures actions that cause harm leading to medically noticeable effects, such as illness and physical or psychological suffering. Sexual Abuse (SA) captures sexual actions towards the child, with or without physical contact (e.g., verbal comments only), and sexual exploitation of the child. Physical Neglect (PN) captures lack in a material sense, for example, of food, clothing, and hygiene, and lack of care during illness, whereas Emotional Neglect (EN) captures lack of emotional response and interaction with the child. The total score on CTQ-SF (CTQtot) ranges from 25 to 125. In each subscale (range 5–25), cut-off levels distinguish between no or minimal trauma, low trauma, moderate trauma, severe trauma, and extreme trauma. The cut-offs used in this study are between moderate and severe trauma, and the cut-off levels are >12 for EA, >9 for PA, >7 for SA, >14 for EN, and >9 for PN (Bernstein & Fink, 1998). CTQ-SF has been shown to have good reliability, internal consistency, and criterion validity (Bernstein et al., 2003; Thombs et al., 2007).

Alcohol Abstinence Self Efficacy Scale (AASE)

Self-efficacy represents an individual’s imagined ability to perform a desirable behaviour (Bandura, 1986). People with high self-efficacy believe that their own decisions and behaviour shape their lives, i.e., that they have control over their own lives. People with low self-efficacy believe the opposite. AASE is a 40-item self-report inventory measuring confidence to abstain from alcohol and resist alcohol temptation in five different high-risk situations on a 5-point scale. The high-risk situations are the following: negative affect, positive mood, physical pain, and alcohol craving and urges (Di Clemente et al., 1994).

Temperament and Character Inventory (TCI)

TCI is a personality inventory based on psychological, biological, and sociological theories (Cloninger et al., 1994). It has four temperament dimensions with an assumed biological basis and three character dimensions, which are assumed to develop through a complex social learning process during childhood. This study used the character dimension “self-directedness”, which captures the individual’s self-acceptance, self-control, and ability to maneuver and steer behavior. It consists of five facets: responsibility, purposefulness, resourcefulness, self-acceptance, and enlightened (Cloninger et al., 1994).

Symptom Check List – 90

The Symptom Checklist – 90 (SCL-90; Derogatis et al., 1974) is a 90-item self-report inventory measuring different aspects of distress. In this study, the sum of all items (i.e., General Severity Index [GSI]) was used to measure the severity of psychiatric symptoms. The SCL-90 scale has been translated into Swedish and validated and normed in a Swedish context. The calculations were made according to Swedish age- and gender-adjusted t-points (Fridell et al., 2002). A t-point of >60 indicates considerable mental health problems.

Confounding Variables

Social Stability

Given the importance of a supportive social environment for successful treatment, we included an index of social stability in our analyses to control for the potential influence of social context on treatment outcome. This index was constructed using items capturing living situation and occupation. If the participants 1) had their own accommodation and 2) worked, studied, or had a retirement pension, they were categorized as socially stable, otherwise not.

DSM-IV Criteria at the 2.5-Year Follow-up

At the 5-year follow-up, we controlled for total number of DSM-IV criteria at the 2.5-year follow-up.

Data Preparation and Statistics

The participants were divided into two groups: a group with experienced CT (at least one experience of moderate to extremely severe trauma, i.e., sexual abuse, physical abuse, emotional abuse, physical neglect, and emotional neglect) and a group without such experience.

The treatment outcome at follow-up was 1) number of DSM-IV criteria met at follow-up or 2) abstinence. Abstinence was defined as not drinking at all or drinking alcohol on only one or a few occasions (i.e., two or three times) since the end of treatment (at 2.5-year follow-up) or over the last 12 months (at 5-year follow-up).

Statistical analysis was performed using IBM SPSS Statistics version 29. The chosen significance level was p < 0.05. Independent t-tests or chi-square tests were used when comparing variables between the two groups. To explore which cells contributed significantly to the association, adjusted standardized residuals were used. To examine effect size, Cohen’s d or Cramer’s V was used. Linear multiple regression (with the outcome variable number of DSM-IV criteria met at the 2.5- and 5-year follow-ups) and logistic regression (with the outcome variable abstinence achieved at the 2.5- and 5-year follow-ups) was carried out to evaluate the effect of CT on treatment outcome, controlling for other relevant variables, as follows: Model 1 – mental health, self-efficacy, self-temptation, and self-directedness; Model 2 – mental health, self-efficacy, self-temptation, and self-directedness + covariate: social stability; and Model 3 (only at the 5-year follow-up) – mental health, self-efficacy, self-temptation, and self-directedness + covariates: social stability and DSM-IV criteria met at the 2.5-year follow-up. The assumptions behind our regression models were met.

Results

One hundred and twenty seven participants (39.6%) with CTQ-SF data (28 missing) had experienced moderate to severe CT in one or more categories. More specifically, 20.9% reported experiencing moderate to severe emotional neglect, 25.4% moderate to severe physical neglect, 10.5% moderate to severe emotional abuse, 10.8% moderate to severe physical abuse, and 4.6% moderate to severe sexual abuse.

Background and Confounding Variables at Baseline

The two groups (Group 1: experience of moderate to severe CT in one or more categories; Group 2: no experienced CT) were similar in background variables with the exception of the confounding variable, with the CT group being less socially stable than the other group (χ2 = [1, n = 321] = 6.43, p < 0.05, φC = 0.15; see Table 2).

Table 2 Background variables for those without (n = 194) and with (n = 127) experienced CT. Data are presented as means (± SD) or frequencies (% in parentheses)

Mental Health, Self-Directedness, and Self-Efficacy at Baseline

The score for overall mental health symptoms, measured by SCL-90 (GSI), indicated that the CT group had significantly more severe overall mental health symptoms than did the group without experienced CT (t[300] = 5.05, p < 0.001, d = 0.62). The score for self-directedness, measured by TCI, indicated that the CT group had less self-directedness than did the group without CT (t[308] = 5.53, p < 0.001, d = 0.55). Moreover, The CT group had more alcohol temptation regarding negative affect (t[307] = 3.99, p < 0.001, d = 0.47), social positive (t[307] = 2.33, p < 0.05, d = 0.27), physical and other concerns (t[307] = 4.00, p < 0.001, d = 0.72), and craving and urges (t[308] = 4.02, p < 0.001, d = 0.47) than did the group without experienced CT. The group without experienced CT had more alcohol-related self-efficacy regarding negative affect (t[308] = 3.22, p < 0.01, d = 0.37), social positive (t[308] = 2.00, p < 0.05, d = 0.24), physical and other concerns (t[308] = 2.90, p < 0.01, d = 0.33), and craving and urges (t[306] = 3.01, p < 0.01, d = 0.35) than did the group with experienced CT (see Table 3).

Table 3 Mental health (SCL-90), self-directedness (TCI), and self-efficacy (AASE) at baseline for those without (n = 194) and with (n = 127) experienced CT; data are presented as means (± SD)

Descriptive Data on Treatment Outcome

The CT group met significantly more DSM-IV criteria at both the 2.5-year follow-up (t[224] = 2.12, p < 0.05, d = 0.29) and 5-year follow-up (t[173] = 2.43, p < 0.05, d = 0.38) than did the group without experienced CT. There were similar rates of abstinence in both groups at the follow-ups (see Table 4).

Table 4 Descriptives of treatment outcome variables for those without experienced CT (baseline: n = 194, 2.5-year follow-up: n = 134, 5-year follow-up: n = 110) and those with experienced CT (baseline: n = 127, 2.5-year follow-up: n = 92, 5-year follow-up: n = 66); data presented as means (± SD) or frequencies (% in parentheses)

Predictors of Treatment Outcome at 2.5-Year and 5-Year Follow-up

Number of DSM-IV Criteria

Multiple regression analyses were conducted to examine the influence of experienced CT on the number of DSM-IV criteria met 2.5 and 5 years after treatment initiation, controlling for self-efficacy, self-directedness, and mental health, as well as for the confounding variable social stability. At the 2.5-year follow-up, using variables from the start of treatment (i.e., experienced CT, self-efficacy, self-directedness, mental health [added to Model 1], and social stability [added to Model 2]), CT was not found to be a significant predictor in either Model 1 or Model 2. The only significant predictor at the 2.5-year follow-up (Model 2) was social stability (unstandardized β = – 0.877, p < 0.05), as shown in Table 5. At the 5-year follow-up, using variables from the start of treatment (i.e., experienced CT, self-efficacy, self-directedness, mental health, and social stability) and from the 2.5-year follow-up (i.e., covariate: DSM-IV criteria met at the 2.5-year follow-up), experienced CT was not a significant predictor in Model 1, Model 2, or Model 3. Social stability was significant in Model 2 (unstandardized β = 0.875, p < 0.05), whereas DSM-IV criteria met at the 2.5-year follow-up (unstandardized β = 0.390, p < 0.001) was the only significant predictor in Model 3 (which included all variables), as shown in Table 6.

Table 5 Linear multiple regression: dependent variable, number of DSM-IV criteria met 2.5 years after treatment initiation; unstandardized beta-coefficients (standard error in parentheses), p-value
Table 6 Linear multiple regression: dependent variable, number of DSM-IV criteria met 5 years after treatment initiation; unstandardized beta-coefficients (standard error in parentheses), p-value

Abstinence

Logistic regression analyses were conducted to examine the influence of experienced CT on abstinence 2.5 and 5 years after treatment initiation. At the 2.5-year follow-up, neither Model 1 nor Model 2 reached statistical significance (Model 1: χ2[11, N = 180] = 9.88, p = 0.54; Model 2: χ2[12, N = 180] = 11.82, p = 0.55). Therefore, experienced CT was not found to be a significant predictor at the 2.5-year follow-up in either Model 1 or Model 2 (data not shown). At the 5-year follow-up, neither Model 1 nor Model 2 reached statistical significance (Model 1: χ2[11, N = 138] = 11.93, p = 0.37; Model 2: chi-square χ2[12, N = 138] = 16.36, p = 0.18). However, Model 3 was found to be statistically significant (χ2[13, N = 138] = 28.42, p = 0.008). The significant predictors in Model 3 were: DSM-IV criteria met at the 2.5-year follow-up (OR = 0.60, p = 0.01, 95% CI [0.40, 0.89]) and temptation social/positive (OR = 0.82, p < 0.05, 95% CI [0.67, 0.99]; see Table 7). Experienced CT was not found to be a significant predictor at the 5-year follow-up.

Table 7 Logistic regression at the 5-year follow-up, with abstinence 5 years after treatment initiation as the outcome variable

Discussion

The aim of this study was to investigate whether experienced CT has an independent effect on alcohol-treatment outcome 2.5 and 5 years after treatment initiation in a sample with AUDs.

Most participants in this study were employed and had their own accommodation. The treatment programmes did not address people with AUDs together with severe psychiatric illnesses or extensive psychiatric disabilities. Despite this, mental health problems occurred among the participants. This could, to some extent, be explained by the alcohol problems themselves, but was also explained by experienced CT. This group of essentially socially stable individuals with AUDs who participated in this study may well represent most of those with alcohol problems undergoing treatment, who do not have severe psychiatric illnesses and disabilities.

We hypothesized that: 1) individuals with AUDs and experienced CT would have lower self-efficacy, lower self-directedness, and poorer mental health; and 2) in these individuals, CT would have an independent effect on treatment outcome, when controlling for other significant variables such as self-efficacy, self-directedness, and mental health.

In this sample, 39.6% had experienced moderate to severe CT, which is about 9% more than in the normal population (Witt et al., 2017), but less than in an alcohol-dependent sample, in which the prevalence of experienced CT was found to be about 50% (Lotzin et al., 2016) using the same instrument and cut-offs as used here. One possible explanation for the different prevalence rates is that the German sample deviated more from the general population in terms of employment and relationship status, which may serve as indicators of more severe background problems, such as experienced CT.

Individuals with AUDs and with experienced CT had more mental health problems, lower levels of self-directedness, and lower levels of self-efficacy in various situations where they might be tempted to consume alcohol. This supports the first hypothesis in our study. The effect sizes for mental health problems and self-directedness were medium, while the effect sizes for various aspects of self-efficacy were small to medium. These findings suggest that patients with a history of CT had a more vulnerable starting point when beginning treatment for their AUDs, which aligns with previous research indicating that individuals with AUDs and a history of CT often have higher levels of mental health problems and lower levels of self-directedness and alcohol-related self-efficacy (Berent et al., 2018; de Carvalho et al., 2015; Jaffee, 2017).

Regarding our second hypothesis, we examined whether a history of CT has an independent effect on treatment outcome 2.5 and 5 years after treatment initiation. Our outcomes were the number of DSM-IV criteria met and abstinence. We controlled for mental health problems, self-directedness, and self-efficacy. Confounding variables were social stability (e.g., having a stable living situation and employment), as previous research has shown that social stability is an important factor to consider (Berglund, 2009), and, at the 5-year follow-up, the number of DSM-IV criteria met at the 2.5-year follow-up, as previous studies have suggested that stability in abstinence or reduced alcohol consumption can increase the chances of continued abstinence or reduced consumption in the long term (Witkiewitz et al., 2021).

Our study found that experienced CT had no independent effect on treatment outcome, as measured by the number of DSM-IV criteria met and abstinence rates 2.5 and 5 years after treatment initiation. Instead, social stability was found to be the only significant predictor at the 2.5-year follow-up. The inclusion of social stability in the multiple regression analysis increased the explained variance from 0.05 to 0.069. However, social stability did not significantly affect treatment outcome when the outcome measure was abstinence at the 2.5-year follow-up. At the 5-year follow-up, the number of DSM-IV criteria met at the 2.5-year follow-up was the most significant predictor of both the number of DSM-IV criteria met and abstinence rates. Fewer DSM-IV criteria met at the 2.5-year follow-up were associated with fewer DSM-IV criteria met and a higher probability of abstinence at the 5-year follow-up. Temptation social/positive was also found to have an independent effect on abstinence at the 5-year follow-up, with lower levels of temptation being associated with a higher probability of abstinence.

Our interpretation of the present results is that experienced CT in individuals with AUDs has a minor impact in terms of predicting treatment outcome. Instead, social stability appears to be the main factor influencing treatment outcome in the first 2.5 years. Social stability, defined here as having accomodation and being employed, studying, or retired with pension, has been shown in previous research to be important for treatment retention and positive treatment outcome (Berglund, 2009). At the 5-year follow-up, social stability no longer had any independent effect; instead, the number of DSM-IV criteria met at the 2.5-year follow-up had a relatively large impact on treatment outcome (explaining 15% of the variance). We interpret this as indicating that good treatment results after 2.5 years are a decisive factor for continued good treatment results, and vice versa. If individuals can achieve stability in their alcohol consumption within 2.5 years, there is a good chance that it will continue in the long term. Surprisingly, variables such as mental health, self-directedness, and self-efficacy were not significant in predicting treatment outcome, with the exception of temptation social/positive, which predicted abstinence at the 5-year follow-up.

This study has some limitations. Experienced CT was reported by adult individuals who had to recollect their distant past. Alcohol consumption data were based solely on self-reports and the drop-out rate was fairly high: about 29% at the 2.5-year and about 45% at the 5-year follow-up. However, there were no differences in drop-out rate between the individuals with and without experienced CT. The primary focus of the alcohol-treatment settings was to help individuals reduce their alcohol consumption, but we could not control for other therapeutic interventions they may have had, for example, interventions regarding mental health problems and childhood experiences. Unfortunately, we only have data recorded at the start of treatment (i.e., baseline) for the control variables, such as self-efficacy, self-directedness, and mental health. We do not have data on any changes in these variables during treatment or by the end of treatment. Consequently, the full effects of these control variables cannot be accurately determined, as they may change over the course of treatment.

This study found that, although experience of moderate to severe CT is relatively common among individuals with AUDs, experienced CT did not have an independent effect on treatment outcome. The group with experienced CT reported more mental health problems, lower self-efficacy, and lower self-directedness, but these additional challenges did not affect treatment outcome for their alcohol problems. This may be due to resilience factors that these individuals possess, which compensate for their negative experiences, or due to coping strategies that allow them to manage their challenges. The clinical implication of this study is that socially well-established individuals with AUDs and experienced CT obtain the same benefit from alcohol treatment as do those without such experience. In individuals with AUDs and co-occurring severe psychiatric disorders or disabilities who have experienced CT, the outcome may be different, so research on treatment outcome is also needed in these groups as well.