Substance Use Disorders, Relapse Rates, and Contributing Factors

Substance Use Disorders (SUDs) are defined as chronic relapsing disorders responsible for 11.8 million deaths annually (Cheron & Kerchove d’Exaerde, 2021). This figure surpasses the yearly mortality rate attributed to cancer and ranks fifth among global causes of death (Cheron & Kerchove d’Exaerde, 2021). Despite extensive clinical and preclinical research spanning decades, relapse rates persist at a staggering 60% for all substance classes except cocaine, opioids, and amphetamine within the initial twelve months of abstinence; for those, the relapse rate climbs to around 80% (McKetin et al., 2018; Sinha, 2011; Soyka et al., 2017). The complexity of relapse rates, particularly in cases of polysubstance use, may be influenced by factors beyond those typically reported. While polysubstance use in general is associated with higher relapse rates (Iqbal et al., 2023), research by Pan and colleagues in 2022 and Heidebrecht et al. (2018) highlighted the significance of specific co-consumption patterns involving substances such as cocaine, heroin, prescription opioids, and cannabis, which were associated with higher rates of relapse. Additionally, shame-proneness (Abbasi et al., 2022; Iqbal et al., 2023) emerges as another potentially culture and social status-dependent factor among polysubstance users, influencing relapse rates but receiving scant scientific attention.

Limitations of studies on drug relapse after prolonged remission

Due to the unpredictable nature of this patient population and the expenses associated with long-term investigations, however, studies examining abstinence rarely extend beyond the 12-month mark (f. e., Branson et al., 2012; Daigre et al., 2021; Mutlu et al., 2016) and frequently exhibit poor operational design.

For example, Scott et al. (2011) conducted a nine-year study involving 1326 adults, primarily focusing on factors contributing to premature death. However, their approach based abstinence on interviews with the Addiction Severity Index and matching treatment documents, omitting critical information such as the frequency of urine sampling, intervention characteristics, and the time until relapse. Moreover, other studies heavily rely on self-reported abstinence or sporadic contact points with several months in between (Darke et al., 2015; Rolová et al., 2023; Termorshuizen et al., 2005). The frequency of measuring points in investigations with a duration of two or more years varies from every fourth week (with a study length of just 96 weeks) to every 600th week (for review see Moe et al., 2022). The terms “relapse” and “remission” also appear to be inadequately, inconsistently, or not at all defined, as highlighted by Moe and colleagues in their recent meta-analysis. Furthermore, the assessment of drug types used during relapse and any differentiation within SUDs is notably lacking (Moe et al., 2022).

Legal Frameworks and Treatment Strategies for Drug Addiction: Section 64 of the German Criminal Code

In Germany, the conceptualization of one branch of forensic psychiatric treatment offers a worldwide almost unique opportunity to investigate the treatment success of SUDs and long-term remission under standardized conditions. Section 64 of the German Criminal Code allows a standardized two-year multimodal (i.e., ergotherapy, sport therapy, occupational therapy, drug substitution therapy, individual and group psychotherapy) treatment for convicted patients with an SUD diagnosis in a forensic psychiatry unit under strictly controlled circumstances. It is important to note that most Sect. 64 governed clinics are gender-separated, with both investigation sides exclusively treating men. Therefore, female patients are not included in this study.

Over the two years, security levels are gradually decreased, and reintegration into society is started (Müller-Isberner et al., 2000). These measures allow the patients to take up work and to spend, on average, the last six months in their own flat. Approximately 60% of patients complete the mandatory two-year hospital treatment (Franke et al., 2020). Successful completion of the clinical treatment then leads to a probational release under up to five years of post-discharge supervision with regular contact to an outpatient care facility. Here, consistent contact and mandatory drug screenings are a vital part of the supervision. Any “relapse” (i.e., date of the first positive urine sample) is thoroughly therapeutically processed. This allows the collection of an immense amount of data regarding treatment success and relapse moderating factors, as well as the influence of social and clinical factors moderating long-term remission. This study exclusively focuses on individuals who have attained this milestone.

The Current Study

Hence, we endeavored to utilize the fair standardization of treatment, leveraging detailed case documentation conducted by forensic psychiatric ambulances in conjunction with regular urine sampling and therapeutic contact. It is important to note that we adhere to the definition provided in the systematic analysis of Moe and colleagues (2022), where we refer to extended periods of abstinence as “remission,” given that the observational period spans up to five years. Our aim was to address a myriad of questions that have hitherto only been partially explored but are considered very relevant in clinicians’ opinions, as reported through personal correspondence during working meetings and conferences. These questions pertain to various parameters such as social factors (e.g., employment or living situation) that are believed to be crucial for long-term treatment success without sound scientific basis. What is the percentage of patients who successfully sustain remission without relapse for up to five years? How does the class of drugs of abuse, as per the ICD-10, influence the duration of remission? In the event of relapse, under what circumstances and with which drug of choice does it occur? Does the presence of a personality disorder impact the overall outcome? To what extent do social factors such as employment, living situation, or relationship status after release from the clinic influence the outcome? Do variables like violent index crimes, age at hospital admission, or age at the first conviction play a role in maintaining remission? Can drug consumption during hospital admission serve as a negative predictor for the success of remission? Does dialectical behavioral therapy contribute to improving the length or rate of remission? Lastly, does the number of addiction pretreatments factor into the likelihood of a successful remission?

Methods

Definitions

All specific diagnoses refer to the International Classification of Diseases in its 10th version (ICD-10).

Patient Population

The study included all patients (N = 439) who were treated in the Clinics for Forensic Psychiatry and Psychotherapy at the District Hospital Günzburg (n = 129) and at the District Hospital Kaufbeuren (n = 310), Bavaria, Germany according to § 64 German Criminal Code and who were conditionally discharged between 2008 and 2021. Conditional discharge occurs when, from a therapeutic and legal perspective, the patient no longer poses a danger to society. The measure is suspended for a probationary period of at least two and a maximum of five years. A description of the socio-demographic and forensic-psychiatric characteristics of the patient sample can be found in Table 1.

Table 1 Sociodemographic and Forensic-Psychiatric Characteristics of the Patient Sample
Full size table

All patients were further supervised by an adjacent Forensic Aftercare Outpatient Clinic at each location for the duration of probationary supervision (M = 34.46 months, SD = 18.33). The frequency with which the patients underwent drug screening at that facility can be found in Table 2.

Table 2 Frequency of Drug Screenings
Full size table

Documentary Sources

To capture sociodemographic, medical, and legal information, a survey form was designed, which included the following variables: gender, age at the time of discharge, nationality, migration background, living situation before hospital admission, employment status before hospital admission, marital status, educational accomplishment, medical history (number of prior inpatient treatments, age at first inpatient treatment, psychiatric primary diagnosis and comorbidities according to ICD-10, somatic diagnoses), current inpatient stay (date of admission, date of leave, date of discharge, total duration of stay, substance use during hospital admission, violence during hospital admission, participation in Dialectical Behavioral Therapy), medication (excluding substitution), substitution (substitute, dosage, start date, and if applicable, end date), criminal backgrounds (number of previous placements under forensic orders (in a different court procedure), number of previous convictions (according to criminal record extract), age at first conviction, spectrum of offenses, index offense, total imprisonment duration), employment situation and living situation at discharge, Forensic Aftercare Outpatient Clinic: frequency of contact, relapse, cause of relapse, date of relapse, frequency of urine and hair tests, substances involved in the relapse, date of the end of probationary supervision, date of the end of forensic aftercare.

Study Procedure

To complete the items of the survey form, the following records at the two Clinics for Forensic Psychiatry and Psychotherapy at the District Hospitals in Günzburg and Kaufbeuren were reviewed: initial assessments, court judgments, available excerpts from the Federal Central Criminal Register, data from the Forensic Information System (FIS), records from admission and diagnostic conferences, treatment and rehabilitation plans, medication sheets, medical summaries (epicrises), and abstinence controls.

Ethical Approval

An ethical approval from the Ethics Committee of the University of Ulm (Application No. 466/22) has been obtained for this study.

Statistical Analysis

The data were analyzed using IBM SPSS Statistics for Windows Version 28 (Armonk, NY: IBM Corp.). Initially, descriptive statistics (mean, standard deviation, absolute and relative frequencies) were generated for sociodemographic and forensic variables. To investigate the research questions, correlations, cross-tabulations and Kaplan-Meier estimators were calculated. A Generalized Linear Model (GZLM) was employed to compare the influence of different predictors. The dependent variable was the binary variable "recidivism" (0=no, 1=yes). The following variables were inserted into the model as predictors: primary diagnosis F19 (0=no, 1=yes), substance use during placement (0=no, 1=yes), personality disorder (0=no, 1=yes), age at first conviction, and number of previous convictions. To predict the dichotomous criterion, the predictors were incorporated into a model with a binomial distribution and the Logit link function. The significance level was set at 5%.

Results

Overall Relapse Rates

While the patients were observed for a mean duration of 31.82 months (SD=16.40 months) following discharge from forensic psychiatry, 227 out of 439 patients (52%) experienced a relapse into substance use during this period. The patients’ mean survival time of 38.183 months, approximately 3 years (SE=3.251; Median: 27.630; SE=3.770) is illustrated in Figure 1 by a Kaplan-Meier survival function. Over the maximum observation period of nearly 60 months, 32% (=cumulative proportion of survivors) would remain abstinent.

Fig. 1
figure 1

Kaplan-Meier survival function for substance use relapses in the entire patient sample (N=439).

Full size image

Situational Factors Influencing Relapse

The relapses occurred with equal frequency both alone (n=89, 47%) and in social settings (n=85, 45%). The three most common triggers for relapse were lack of motivation for abstinence, a celebration/party, or an emotionally stressful situation (see Table 3).

Table 3 Circumstances Surrounding Relapse (n = 227)
Full size table

In instances of relapse, alcohol (n=70, 32%), multiple substances (n=48, 22%), and opioids (n=40, 18%) were consumed most frequently. The correlation between the type of substance consumed during relapse and the primary diagnosis of the patients varied across different substance use disorders, as outlined in Table 4. Patients diagnosed with alcohol use disorder (F10) were more likely to relapse with alcohol, and less likely to consume cannabis in such cases. Similarly, individuals with opioid use disorder (F11) tended to relapse with opioids, while those with cocaine dependence (F14) were more likely to relapse with cocaine. In contrast, patients diagnosed with other stimulant-related disorders (F15) exhibited equal likelihood of relapse with cocaine and other substances. Interestingly, individuals diagnosed with polysubstance use disorder (F19) were most likely to relapse with cannabis. Surprisingly, patients with cannabis use disorder (F12) did not exhibit a specific drug preference during relapse.

Table 4 Spearman Correlations Between the Primary Diagnosis (according to ICD-10) and the Substance Consumed During Relapse (n = 227)
Full size table

Social Factors Influencing Relapse

The patients' work situation upon discharge did not play a role in influencing substance use relapse. Similarly, the marital status at the time of discharge had no significant impact on relapse and the circumstances of living showed no association with relapse (see Table 5).

Table 5 Social, Treatment Related, Diagnostic and Criminalistic Factors Affecting Relapse and Sustained Remission in The Patient sample (N = 439)
Full size table

Therapeutic, Diagnostic, and Criminalistic Factors Influencing Relapse

Participation in Dialectical Behavioral Therapy - Forensic (DBT-F), specially developed to improve personal skills surrounding impulsivity and addiction, did not have a significant impact on relapse (see Table 5). The primary diagnosis, however, had an impact on relapse (see Table 5 and for graphical demonstration Figure 2). As indicated by the standardized residuals, patients with a disorder involving multiple substance use were more likely to experience relapse (+1.2), whereas patients with a disorder related to cannabinoids (-1.7) and other stimulants (-1.1) were less prone to relapse. Substance use during confinement significantly influenced relapse after discharge (see Table 5). Patients who engage in substance use during confinement were significantly more likely to experience relapse. Furthermore, patients with a comorbid personality disorder were significantly more likely to relapse compared to patients without a comorbid personality disorder (see Table 5). The nature of the index crime leading to the current treatment was found to be irrelevant regarding the maintenance of abstinence. Patients convicted of a violent crime did not experience relapse more frequently than patients without a violent index crime (see Table 5). The patient's age at discharge was found to be non-predictive regarding relapse and sustained long-term remission. However, the age at first conviction correlated with an increased probability of relapse. Younger patients at their first conviction tended to relapse more frequently (see Table 5). The analysis indicated that the number of previous clinical treatments for drug addiction did not affect the overall relapse rate after long-term remission. However, the number of previous convictions was found to increase the likelihood of relapse. Patients with more convictions had a higher relapse frequency (see Table 5).

Fig. 2
figure 2

Kaplan-Meier Survival Function for Substance Use Relapses Presented Separately for the Individual Primary Diagnosis (Curves for F12 and F15 are Overlapping; F10=105; F11=37; F12=35; F14=22; F15=11; F19=228; Missing = 1)

Full size image

In summary, our various analyses reveal that several factors are linked to a potential recurrence of substance relapse, including the type of the main ICD-10 SUD diagnosis, substance use during confinement, the presence of a personality disorder, age at first conviction, and the number of prior convictions. To compare the impact of these factors, we employed a generalized linear model. The direct comparison of predictors was assessed using the Wald statistic, with a larger test statistic indicating a more influential predictor. In this analysis, the variables of substance use during confinement, number of previous convictions, and the presence of a personality disorder, in descending order of significance, stand out as significant predictors for anticipating a relapse into addiction (see Table 6).

Table 6 Result of the Generalized Linear Model for Predicting Relapse in The Forensic Patient Sample (N = 439)
Full size table

Discussion

Factors Influencing Relapse After Long-Term Remission

In summary, during this study several key factors influencing relapse emerged: emotional stress, social activities, personality disorder diagnosis, inpatient treatment relapse, age at first conviction, and loss of motivation. Surprisingly, social factors like family and work status showed no impact, nor the way of living. In line with other studies (Abbasi et al., 2022; Heidebrecht et al., 2018; Iqbal et al., 2023), polysubstance use, however, increased relapse likelihood, unlike other specific substance addictions, while multiple previous inpatient treatments had little predictive value. Remarkably, the presence of violence in the index crime did not affect long-term remission, nor did dialectical behavioral therapy as an add-on therapy. Also, the number of inpatient pretreatments had little to no predictive value regarding the length of long-term remission or relapse probability. Finally, a generalized linear model allowed substance use during treatment, personality disorder, and numerous prior convictions to be the most central predictors of post-discharge relapse.

It's unsurprising that emotional stress and engagement in high-risk situations emerge as key mediators in relapse, since it is consistent with extensive research (Koob & Mason, 2016; Melemis, 2015). Furthermore, the co-occurrence rate of SUDs and personality disorders is notably high, with SUDs and borderline personality disorder co-occurring in 12 to 53%, and SUDs and antisocial personality disorder in 9 to 23% (Torgersen et al., 2001). However, studies on treatment success prognosis in SUD-personality disorder patients remain scarce, with only one study emphasizing a chronic course of alcohol use disorder in patients with an antisocial personality disorder (Hasin et al., 2011). Thus, in accordance with our results we would like to underscore the need for specialized treatment options (other than DBT), just like Stetsiv et al. did (2023).

Dialectic Behavioral Therapy – a Critical View

The ineffectiveness of dialectical behavioral therapy may surprise initially, given some evidence of its efficacy in relapse prevention (Kaminer et al., 2002; López et al., 2021). However, studies reporting efficacy for prolonged abstinence are limited to 12 months or shorter, highlighting a gap in understanding DBTs influence on long-term outcomes (Kaminer et al., 2002; López et al., 2021). Moreover, earlier randomized controlled trials focused on individuals with polysubstance dependence in community-based treatment settings, as demonstrated by Linehan et al. in 1999. While their study showed promising results with DBT as an add-on treatment, the low number of participants (12 in the DBT group versus 16 in the control group) makes it challenging to pinpoint the exact factors contributing to reduced substance consumption and increased treatment adherence. A subsequent study by Linehan et al. in 2002 provided more rigorous control conditions and validation therapy, but it still had limitations such as a small sample size (23 opiate-dependent individuals with borderline personality disorder) and a focus solely on consumption reduction and treatment adherence within the last four months of treatment. In 2018, Maffei et al. conducted an open-label study on 244 alcohol-dependent individuals, reporting a 73.2% abstinence rate after 3 months of DBT training for emotional regulation. However, the study did not provide information on sustained remission. The most comprehensive study to date was conducted by Cavicchioli and colleagues in 2019, which assessed consecutive days of abstinence through random drug testing twice a week in 108 individuals with Alcohol Use Disorder. While the study showed a significant increase in consecutive days of abstinence in the DBT group, changes in DBT-related skill training scales did not account for this increase. Moreover, the study was limited to the duration of the treatment period. In summary, the reported ineffectiveness of DBT on long-term remission as presented in the current work may not be surprising, particularly when considering the statement of a recent meta-analysis conducted by Warner and Murphy in 2022: “Despite offering preliminary support for DBT-ST for SUD, the lack of controls, small samples and inconsistent adaptations of DBT-ST across studies, limits capacity to draw causal conclusions or make specific recommendations.”

Personality Aspects and Recovery Capital

It is notable that the number of previous convictions and age at first conviction alongside (antisocial) personality disorders serve as strong relapse predictors. Despite no actual studies than ours assessing those two factors, their impact is evident.

While Recovery Capital theory (Cloud & Granfield, 2008) with limited empirical support of questionably scientific soundness (for critical review please see Hennessy, 2017; f.e., Weston et al., 2018) emphasizes social resources in SUD recovery, our study challenges this notion, showing that factors like family status, living circumstances, and employment status played no predictive role in relapse.

Limitations

Still our work is not above limitations. On being its retrospective, patient record-based approach, which relies solely on file accuracy. Additionally, these files were not specifically designed to address the questions raised by the researchers, which may introduce bias. Missing information could also potentially lead to false positive or false negative results, a possibility that cannot be entirely ruled out. Therefore, the factor "Age at first inpatient treatment," which was found to be insignificant for relapse but has the highest number of missing data, should be interpreted with caution. However, in all other cases, the amount of missing information is minimal, and in the authors’ opinion, the statistical conclusions appear robust.

Furthermore, while our abstinence control method is relatively strict, it has limitations, such as urine testing for ethanol detecting consumption only for 48 to 72 hours (Hiemke et al., 2018). Despite these challenges, our study reports higher abstinence rates than previously estimated (Fleury et al., 2016; Rolová et al., 2023), advocating for long-term inpatient treatment, and continued therapeutic support and control.

Finally, another limitation stems from the structural constraint, dictated by clinical circumstances, which restricts the study to male patients only. Nonetheless, the authors are keen on comparing factors influencing relapse across both sexes. Currently, they are in the process of conducting a comparable study in two Section 64 governed clinics exclusively treating female patients with SUDs.

Conclusions

The current study sheds light on new factors impacting relapse following extended periods of remission. While familiar elements like high-risk environments/situations (e.g. parties) and emotional stress remain significant contributors to relapse, it is noteworthy that social factors such as marital status, employment status, and living situation were not found to have any influence. Additionally, antisocial personality traits and behaviors, which have been traditionally rather overlooked in scientific research, emerged as key drivers in increasing the likelihood of relapse.

Clinical Implications

This study clearly emphasizes the critical need for evidence-based clinical treatment decisions over subjective viewpoints and anecdotal impressions. Future strategies for addressing substance use disorder may require innovative approaches to accommodate some of the here newly identified factors. For example, there should be a greater emphasis on strategies for sustaining motivation for abstinence rather than focusing on social factors related to the release setting. Moreover, enhancing patients’ understanding of addiction-related high-risk situations and providing them with behavioral strategies to navigate and overcome these scenarios may prove pivotal. Additionally, gaining insights into bodily stress reactions and developing a nuanced understanding of emotional states emerge as crucial elements in preventing relapse. Perhaps the most novel aspect appears to be recognizing the significance of antisocial personality traits and behaviors in relapse risk underscores the importance of tailoring treatment interventions to target these specific aspects. By doing so, we may pave the way for more effective and comprehensive approaches to long-term recovery.