P5 "Individual predictors of the development and course of problematic cannabis use in the general population and for patients undergoing psychotherapy"

Maria Neumann

Supervisor: Prof. Dr. Gerhard Bühringer

Duration: 10/2014 – 09/2017

Background. Cannabis is the most commonly used illegal psychoactive substance worldwide. While cannabis use (CU) remains unproblematic and transient in many cases, some users tend to regularly use cannabis, which can result in adverse consequences such as the onset of a can- nabis use disorder (short: cannabis disorder) and eventually require professional interventions. These facts raise two central questions that will be examined in-depth within this thesis:

• 1)  What are important individual predictors of the course of problematic cannabis use (PCU), including regular use and a cannabis disorder, in the general population?

• 2)  What are significant individual predictors of the success or failure of psychotherapy for PCU?

These questions will be addressed separately for the general population in the first part of the thesis and for patients undergoing psychotherapy due to PCU in the second part of the thesis. It is suggested that predictors of the course differ for both populations. Within the framework of this thesis a model of the course of PCU is being developed, which comprises regular use over a longer time period as a first stage. From there on, individuals may progress to further stages of a cannabis disorder onset and to either the chronification or the spontaneous or professionally supported cessation of PCU.

In the first part, individual factors are systematized with regard to their change and stability as predictors of stages of the PCU course in the general population. It is hypothesized that psycho- pathological and substance-use related predictors become more relevant while environmental factors decrease in relevance. So far, little is known about the relevance of these factors for the PCU cessation. In the second part of the thesis, predictors of achieving PCU treatment goals are examined in greater detail. Although the effectiveness of existing interventions for PCU is well documented in meta-analyses, the sustained abstinence rates are still rather low. Therefore, an examination of individual factors is conducted with regard to their prognosis for treatment suc- cess or failure. The hypothesis is that the use-related predictors withdrawal and illegal polysub- stance use predict lower abstinence and remission rates, while a high degree of refusal self- efficacy predicts higher rates and may act as a moderator between withdrawal and abstinence. The aim of the thesis is to provide a basis for the improvement of etiological models in order to stimulate further research and to improve prevention and treatment approaches for the reduc- tion of PCU.

Methods. To address the first question, longitudinal, epidemiological and representative cohort studies on predictors of the onset and course of PCU in the general population were systemati- cally reviewed according to PRISMA-guidelines. The databases Web of Knowledge, Scirus, PubMed, and PsyIndex were screened for publications in English and in the time from 1994 to 2014, which deal with predictors for the outcome variables increase, persistence, abstinence, and recurrence of cannabis use as well as onset, stability, remission and relapse of a cannabis disor- der.

To address the second question, data from the multicenter treatment study CANDIS II, a randomized-controlled trial of treatment seekers with regular cannabis use, served as a basis. Data from those 166 patients which fully participated in the cognitive-behavioral, manualized treatment program and in scientific investigations at treatment entrance and end were analyzed for this work.

Polysubstance use was examined through 13 indicators (including substance class, quantity, and frequency) as predictors of abstinence and remission, and secondary treatment outcomes. Pre- dictors were analyzed with crude regression models or with appropriate linear or gamma re- gression models, respectively. Similarly, adjusted regression models were calculated using em- pirically determined control variables of cannabis use patterns (including frequency, dose, and mode of administration), age, and gender. Refusal self-efficacy and withdrawal were examined as predictors of remission of an ICD-10 dependence at treatment end and for abstinence up to six months after treatment end using logistic regression models. A moderation analysis of re- fusal self-efficacy as a moderator between withdrawal duration and abstinence was performed.

Results. First question: 29 publications from 14 epidemiological studies with an observational time span of one to 30 years were analyzed. Some predictors were quite specific for certain stages. For example, alcohol use and cannabis use of peers were specific predictors for use increase and persistence. Positive life events specifically predicted use cessation. Conflicts with and between parents as well as certain psychological disorders were specific predictors of an onset of a can- nabis disorder. A major depression specifically predicted relapse. Some factors were predictive of several stages such as a high use frequency, other illegal substance use and externalizing symptoms. There was a lack of evidence on predictors of the natural remission of cannabis dis- orders.

Second question: Polysubstance use indicators did not predict lower remission rates. Mental health was even improved in patients with illegal PSU; however, this group experienced lower absti- nence rates. Amphetamine use predicted poorer outcomes regarding CU-related problems, and use of opiates, inhalants, and dissociative substances predicted poorer physical health outcomes. A high refusal self-efficacy and a withdrawal syndrome of short endurance both predicted higher rates of remission and sustained abstinence. A moderating effect of refusal self-efficacy on the relationship between withdrawal and abstinence could not be demonstrated, although some results suggest such a relation. In the multivariate model (cannabis use patterns, polysubstance use, duration of withdrawal, refusal self-efficacy, age, gender), refusal self-efficacy was the most relevant predictor of sustained abstinence.

Discussion and conclusion. The suggested increase in psychopathological predictors in the tran- sition from regular use to a cannabis disorder was empirically supported. A decreasing relevance of social predictors could mostly not be found, in part due to a lack of respective studies. Mental protective and risk factors are suggested to be more predictive for both the spontaneous and the therapeutically supported cessation of PCU than use-related risk factors such as withdrawal and polysubstance use.

Conclusions for practice and research

• Clients with PCU can be successfully treated in an ambulant setting also in the presence of moderate polysubstance use and also if withdrawal occurs.

• Illegal polysubstance use may be a more relevant predictor of the PCU course in a natural compared to a treatment setting. The results should be checked for robustness and ex- amined in further samples.

• A high degree of refusal self-efficacy may be a key predictor for the sustained abstinence from cannabis, especially in the event of a withdrawal syndrome.

  The stage model presented in this thesis allows the comparison and prediction of a variety of PCU courses. Hence, the thesis importantly contributes to the systematization of the clinical- psychological research on predictors of the course of PCU. This contribution may serve as a basis for advancements in the respective etiological research and for interventions that aim to reduce PCU.