Machine Learning Facial Emotion Classifiers in Psychotherapy Research: A Proof-of-Concept Study.

BACKGROUND: New advances in the field of machine learning make it possible to track facial emotional expression with high resolution, including micro-expressions. These advances have promising applications for psychotherapy research, since manual coding (e.g., the Facial Action Coding System), is time-consuming. PURPOSE: We tested whether this technology can reliably identify in-session emotional expression in a naturalistic treatment setting, and how these measures relate to the outcome of psychotherapy. METHOD: We applied a machine learning emotion classifier to video material from 389 psychotherapy sessions of 23 patients with borderline personality pathology. We validated the findings with human ratings according to the Clients Emotional Arousal Scale (CEAS) and explored associations with treatment outcomes. RESULTS: Overall, machine learning ratings showed significant agreement with human ratings. Machine learning emotion classifiers, particularly the display of positive emotions (smiling and happiness), showed medium effect size on median-split treatment outcome (d = 0.3) as well as continuous improvement (r = 0.49, p < 0.05). Patients who dropped out form psychotherapy, showed significantly more neutral expressions, and generally less social smiling, particularly at the beginning of psychotherapeutic sessions. CONCLUSIONS: Machine learning classifiers are a highly promising resource for research in psychotherapy. The results highlight differential associations of displayed positive and negative feelings with treatment outcomes. Machine learning emotion recognition may be used for the early identification of drop-out risks and clinically relevant interactions in psychotherapy.

Withdrawal ruptures in adolescents with borderline personality disorder psychotherapy are marked by increased speech pauses-can minimal responses be automatically detected?

Alliance ruptures of the withdrawal type are prevalent in adolescents with borderline personality disorder (BPD). Longer speech pauses are negatively perceived by these patients. Safran and Muran's rupture model is promising but its application is very work intensive. This workload makes research costly and limits clinical usage. We hypothesised that pauses can be used to automatically detect one of the markers of the rupture model i.e. the minimal response marker. Additionally, the association of withdrawal ruptures with pauses was investigated. A total of 516 ruptures occurring in 242 psychotherapy sessions collected in 22 psychotherapies of adolescent patients with BPD and subthreshold BPD were investigated. Trained observers detected ruptures based on video and audio recordings. In contrast, pauses were automatically marked in the audio-recordings of the psychotherapy sessions and automatic speaker diarisation was used to determine the speaker-switching patterns in which the pauses occur. A random forest classifier detected time frames in which ruptures with the minimal response marker occurred based on the quantity of pauses. Performance was very good with an area under the ROC curve of 0.89. Pauses which were both preceded and followed by therapist speech were the most important predictors for minimal response ruptures. Research costs can be reduced by using machine learning techniques instead of manual rating for rupture detection. In combination with other video and audio derived features like movement analysis or automatic facial emotion detection, more complete rupture detection might be possible in the future. These innovative machine learning techniques help to narrow down the mechanisms of change of psychotherapy, here specifically of the therapeutic alliance. They might also be used to technologically augment psychotherapy training and supervision.

The impact of outcome expectancy on therapy outcome in adolescents with borderline personality disorder.

BACKGROUND: Outcome expectancy has been found to be a significant predictor of psychotherapy outcome. However, given that severity, chronicity and comorbidity are moderators of outcome expectancy, it is important to provide evidence of whether the same holds true in clinical conditions marked by these attributes, such as in borderline personality disorder (BPD). The aim of the present study was to investigate the role of patients' outcome expectancy in adolescents undergoing early intervention for BPD using pre-post difference of psychosocial functioning as outcome. METHODS: Forty-four adolescent BPD patients were treated with Dialectical Behavior Therapy for Adolescents (DBT-A) or Adolescent Identity Treatment (AIT). We investigated the effect of outcome expectancy on outcome with type of treatment as moderator. Based on the relevant literature, we assess the correlation between outcome expectancy and pretreatment symptomatology, namely BPD severity, personality functioning, childhood trauma and depression. RESULTS: The results showed a significant effect of expectancy on outcome (stand. ß = 0.30, p = 0.020) above autoregression. ANOVA analysis revealed no difference between the two treatments. Further, results indicate that pretreatment symptomatology, i.e., depression, childhood trauma and personality functioning dimensions self-direction and intimacy, are associated with early treatment expectancy. CONCLUSION: Outcome expectancy as a common factor plays a key role in successful psychotherapy with adolescent BPD patients. Elevated pretreatment depression, childhood trauma and impairment in personality functioning dimensions self-direction and intimacy are risk factors associated with lower expectancy. Low outcome expectancy should be addressed in early psychotherapy to improve the therapeutical process.

A DSM-5 AMPD and ICD-11 compatible measure for an early identification of personality disorders in adolescence-LoPF-Q 12-18 latent structure and short form.

The LoPF-Q 12-18 (Levels of Personality Functioning Questionnaire) was designed for clinical use and to promote early detection of personality disorder (PD). It is a self-report measure with 97 items to assess personality functioning in adolescents from 12 years up. It operationalizes the dimensional concept of personality disorder (PD) severity used in the Alternative DSM-5 Model for Personality Disorders and the ICD-11. In this study, we investigated the factorial structure of the LoPF-Q 12-18. Additionally, a short version was developed to meet the need of efficient screening for PD in clinical and research applications. To investigate the factorial structure, several confirmatory factor analysis models were compared. A bifactor model with a strong general factor and four specific factors showed the best nominal fit (CFI = .91, RMSEA = .04, SRMR = .07). The short version was derived using the ant colony optimization algorithm. This procedure resulted in a 20-item version with excellent fit for a hierarchical model with four first order factors to represent the domains and a secondary higher order factor to represent personality functioning (CFI = .98, RMSEA = .05, SRMR = .04). Clinical validity (effect size d = 3.1 between PD patients and controls) and clinical utility (cutoff ≥ 36 providing 87.5% specificity and 80.2% sensitivity) for detecting patients with PD were high for the short version. Both, the long and short LoPF-Q 12-18 version are ready to be used for research and diagnostic purposes.

Mental disorders into adulthood among adolescents placed in residential care: A prospective 10-year follow-up study.

BACKGROUND: Child welfare and juvenile justice placed youths show high levels of psychosocial burden and high rates of mental disorders. It remains unclear how mental disorders develop into adulthood in these populations. The aim was to present the rates of mental disorders in adolescence and adulthood in child welfare and juvenile justice samples and to examine their mental health trajectories from adolescence into adulthood. METHODS: Seventy adolescents in shared residential care, placed by child welfare (n = 52, mean age = 15 years) or juvenile justice (n = 18, mean age = 17 years) authorities, were followed up into adulthood (child welfare: mean age = 25 years; juvenile justice: mean age = 27 years). Mental disorders were assessed based on the International Classification of Diseases 10th Revision diagnoses at baseline and at follow-up. Epidemiological information on mental disorders was presented for each group. Bivariate correlations and structural equation modeling for the relationship of mental disorders were performed. RESULTS: In the total sample, prevalence rates of 73% and 86% for any mental disorder were found in adolescence (child welfare: 70%; juvenile justice: 83%) and adulthood (child welfare: 83%; juvenile justice: 94%) respectively. General psychopathology was found to be stable from adolescence into adulthood in both samples. CONCLUSIONS: Our findings showed high prevalence rates and a high stability of general psychopathology into adulthood among child welfare and juvenile justice adolescents in Swiss residential care. Therefore, continuity of mental health care and well-prepared transitions into adulthood for such individuals is highly warranted.

Prevalence and 10-Year Stability of Personality Disorders From Adolescence to Young Adulthood in a High-Risk Sample.

Background: With the implementation of the 11th edition of the International Classification of Diseases (ICD-11) in early 2022, there will be a radical change in the framework and process for diagnosing personality disorders (PDs), indicating a transition from the categorical to the dimensional model. Despite increasing evidence that PDs are not as stable as previously assumed, the long-term stability of PDs remains under major debate. The aim of the current paper was to investigate the categorical and dimensional mean-level and rank-order stability of PDs from adolescence into young adulthood in a high-risk sample. Methods: In total, 115 young adults with a history of residential child welfare and juvenile-justice placements in Switzerland were included in the current study. PDs were assessed at baseline and at a 10-year follow-up. On a categorical level, mean-level stability was assessed through the proportion of enduring cases from baseline to follow-up. Rank-order stability was assessed through Cohen's κ and tetrachoric correlation coefficients. On a dimensional level, the magnitude of change between the PD trait scores at baseline and at follow-up was measured by Cohen's d. Rank-order stability was assessed through Spearman's ρ. Results: The prevalence rate for any PD was 20.0% at baseline and 30.4% at follow-up. The most frequently diagnosed disorders were antisocial, borderline, and obsessive-compulsive PDs, both at baseline and at follow-up. On a categorical level, the mean-level stability of any PD was only moderate, and the mean-level stability of specific PDs was low, except of schizoid PD. Likewise, the rank-order stability of any PD category was moderate, while ranging from low to high for individual PD diagnoses. On a dimensional level, scores increased significantly for most PDs, except for histrionic traits, which decreased significantly from baseline to follow-up. Effect sizes were generally low. The rank-order stability for dimensional scores ranged from low to moderate. Conclusion: The findings indicate low to moderate stability of Pds and Pd traits from adolescence to adulthood, which supports the growing evidence that categorical diagnoses of Pds are quite unstable. This in turn, emphasizes the use of the upcoming ICD-11 that Acknowledgments Pds to be only "relatively" stable.

Psychophysiological responses to sadness in girls and boys with conduct disorder.

Reduced responsiveness to emotions is hypothesized to contribute to the development of conduct disorder (CD) in children and adolescents. Accordingly, blunted psychophysiological responses to emotions have been observed in boys with CD, but this has never been tested in girls. Therefore, this study compared psychophysiological responses to sadness in girls and boys with and without CD, and different clinical phenotypes of CD: with versus without limited prosocial emotions (LPE), and with versus without comorbid internalizing disorders (INT). Nine-hundred and 27 girls (427 CD, 500 controls) and 519 boys (266 CD, 253 controls) aged 9-18 years participated. Psychophysiological responses were measured while participants watched two validated sad film clips, specifically: heart rate (HR), respiratory sinus arrhythmia (RSA; indexing parasympathetic activity), preejection period (PEP; indexing sympathetic activity). Girls and boys with CD showed larger HR responses to sadness than controls. This effect was rendered nonsignificant, however, after controlling for covariates. We observed aberrant RSA responses to sadness in CD compared with controls. Similarly, we found a significant positive association between RSA responsivity and antisocial behavior when assessed dimensionally. The effects were very small, though. Results were similar for boys and girls. We found no evidence for emotional underresponsiveness in CD in the largest psychophysiological study to date in this field. More research is needed to explore whether this is specific to sadness or generalizes to other emotions. Furthermore, we recommend that studies on emotion processing in CD assess different physiological measures to help disentangle CD-related effects on sympathetic and parasympathetic activity. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Therapists' emotional state after sessions in which suicidality is addressed: need for improved management of suicidal tendencies in patients with borderline personality pathology.

INTRODUCTION: Patient suicidality is a frequently experienced topic for psychotherapists. Especially adolescents with borderline personality pathology (BPP) often exhibit suicidal tendencies. Previous research which examined therapists' countertransference towards suicidal patients suggested that therapists are negatively affected and distressed by them. We hypothesize that this emotional response of the therapists is related to specific sessions in which suicidality came up as a topic. Accordingly, the objective of this study consists in examining therapists' emotional state on a session level of analysis. METHODS: The sample consisted of N = 21 adolescents (age 13-19 years) with BPD or subthreshold BPD. Therapists' emotional states were measured in n = 418 sessions using the Session Evaluation Questionnaire. Principal component analysis was used to reduce dimensionality of the therapist response. The emotional states were compared depending on whether suicidality has been addressed in the session (SS) or not (NSS). RESULTS: Two components could be identified. Firstly, therapists were more aroused, excited, afraid, angry and uncertain after SS than after NSS. Secondly, therapists were more aroused, excited, definite and pleased after SS than after NSS. DISCUSSION: Suicidality does not always have to be a burden for therapists: Both a "distress" and an "eustress" component occur in this context from which the latter is supposed to help clinicians master a difficult situation. Since countertransference feelings are often not fully conscious, it is necessary to do research on therapists' emotional states after sessions in which suicidality is addressed. This is crucial to both prevent the therapeutic process from being endangered and preserve clinicians' mental health. Clinical implications and limitations are discussed.

Movement Synchrony in the Psychotherapy of Adolescents With Borderline Personality Pathology - A Dyadic Trait Marker for Resilience?

Movement synchrony describes the coordination of body movements. In psychotherapy, higher movement synchrony between therapist and patient has been associated with higher levels of empathy, therapeutic alliance, better therapy outcome, and fewer drop-outs. The current study investigated movement synchrony during the psychotherapeutic treatment of female adolescents with borderline personality disorder. It was hypothesized that there are higher levels of movement synchrony in the analyzed therapy sessions compared to pseudo-interactions. Further, we tested whether higher levels of movement synchrony correlate with stronger patients' symptom reduction and whether higher movement synchrony predicts higher post-session ratings. A total of 356 sessions from 16 completed psychotherapies of adolescent patients with BPD were analyzed. Movement synchrony was assessed with motion energy analysis and an index of synchrony was calculated by lagged cross-correlation analysis. As hypothesized, the findings support higher levels of movement synchrony in therapy sessions compared to pseudo-interactions (Cohen's d = 0.85). Additionally, a correlation of movement synchrony with better therapy outcome was found (standardized beta = -0.43 indicating stronger personality functioning impairment reduction). The post-session ratings were negatively associated with higher levels of movement synchrony (standardized beta = -0.1). The relevance of movement synchrony and potential implications for clinical practice are discussed.

Positive and negative parenting in conduct disorder with high versus low levels of callous-unemotional traits.

Less is known about the relationship between conduct disorder (CD), callous-unemotional (CU) traits, and positive and negative parenting in youth compared to early childhood. We combined traditional univariate analyses with a novel machine learning classifier (Angle-based Generalized Matrix Learning Vector Quantization) to classify youth (N = 756; 9-18 years) into typically developing (TD) or CD groups with or without elevated CU traits (CD/HCU, CD/LCU, respectively) using youth- and parent-reports of parenting behavior. At the group level, both CD/HCU and CD/LCU were associated with high negative and low positive parenting relative to TD. However, only positive parenting differed between the CD/HCU and CD/LCU groups. In classification analyses, performance was best when distinguishing CD/HCU from TD groups and poorest when distinguishing CD/HCU from CD/LCU groups. Positive and negative parenting were both relevant when distinguishing CD/HCU from TD, negative parenting was most relevant when distinguishing between CD/LCU and TD, and positive parenting was most relevant when distinguishing CD/HCU from CD/LCU groups. These findings suggest that while positive parenting distinguishes between CD/HCU and CD/LCU, negative parenting is associated with both CD subtypes. These results highlight the importance of considering multiple parenting behaviors in CD with varying levels of CU traits in late childhood/adolescence.

Alliance Ruptures and Resolutions in Personality Disorders.

PURPOSE OF REVIEW: This review provides an overview of the state of research on alliance ruptures and resolutions in the treatment of personality disorders (PDs). We discuss frequently used instruments to measure alliance ruptures and resolutions. We discuss the effectiveness of rupture resolution processes and highlight possible avenues for research to explore. Innovative assessments with the potential to reveal the link of ruptures and resolutions and mechanisms of psychotherapeutic change are discussed. RECENT FINDINGS: The assessment of alliance rupture and resolutions is heterogeneous. Instruments vary largely with respect to a direct or indirect assessment, the time resolution of assessment (integral therapy, phase, session, event), session sampling strategy and perspectives (patient, therapist, observer). The heterogeneity in the instruments and study designs impedes comparability and interpretation of the findings. Results support the hypothesis that ruptures are more frequent in PD. Results also point towards beneficial effects of rupture resolution patterns, early alliance quality, and resolution complexity. Few studies control findings for pretreatment factors. Evidence points to the direction that rupture resolution processes can be considered a general principle of change in the treatment of PD. The concept of alliance ruptures and resolutions provides a useful tool for the management of the therapeutic alliance and its moments of deteriorations throughout the treatment course. Dimensional pretreatment personality functioning is considered a key variable in future studies to highlight what works for whom.

Supervised Speaker Diarization Using Random Forests: A Tool for Psychotherapy Process Research.

Speaker diarization is the practice of determining who speaks when in audio recordings. Psychotherapy research often relies on labor intensive manual diarization. Unsupervised methods are available but yield higher error rates. We present a method for supervised speaker diarization based on random forests. It can be considered a compromise between commonly used labor-intensive manual coding and fully automated procedures. The method is validated using the EMRAI synthetic speech corpus and is made publicly available. It yields low diarization error rates (M: 5.61%, STD: 2.19). Supervised speaker diarization is a promising method for psychotherapy research and similar fields.

Unresolved Attachment Mediates the Relationship Between Childhood Trauma and Impaired Personality Functioning in Adolescence.

The present study investigates the mediating effect of attachment trauma on the relationship between childhood maltreatment and personality functioning in 199 adolescents (12-18 years) using a novel approach of an in-depth analysis of attachment-related traumatic contents during an attachment interview (AAP). Our findings demonstrate that adolescents with a high amount of traumatic attachment-related material show a lower resilience when facing traumatic childhood experiences, resulting in a greater severity of personality dysfunction. In particular, the associations between emotional abuse and neglect and the domains of identity, empathy, self-direction, and intimacy were mediated by the severity of attachment trauma. These results advance our understanding of the different nuances of attachment-related traumatic material and how they might shape personality structure in an adolescent age group.

Family composition and age at menarche: Findings from the international Health Behaviour in School-aged Children study.

BACKGROUND: Early menarche has been associated with father absence, stepfather presence and adverse health consequences in later life. This article assesses the association of different family compositions with the age at menarche. Pathways are explored which may explain any association between family characteristics and pubertal timing. METHODS: Cross-sectional, international data on the age at menarche, family structure and covariates (age, psychosomatic complaints, media consumption, physical activity) were collected from the 2009-2010 Health Behaviour in School-aged Children (HBSC) survey. The sample focuses on 15-year old girls comprising 36,175 individuals across 40 countries in Europe and North America (N = 21,075 for age at menarche). The study examined the association of different family characteristics with age at menarche. Regression and path analyses were applied incorporating multilevel techniques to adjust for the nested nature of data within countries. RESULTS: Living with mother (Cohen's d = .12), father (d = .08), brothers (d = .04) and sisters (d = .06) are independently associated with later age at menarche. Living in a foster home (d = -.16), with 'someone else' (d = -.11), stepmother (d = -.10) or stepfather (d = -.06) was associated with earlier menarche. Path models show that up to 89% of these effects can be explained through lifestyle and psychological variables. CONCLUSIONS: Earlier menarche is reported amongst those with living conditions other than a family consisting of two biological parents. This can partly be explained by girls' higher Body Mass Index in these families which is a biological determinant of early menarche. Lower physical activity and elevated psychosomatic complaints were also more often found in girls in these family environments.

White Matter Microstructure in Youths With Conduct Disorder: Effects of Sex and Variation in Callous Traits.

OBJECTIVE: Studies using diffusion tensor imaging (DTI) to investigate white matter (WM) microstructure in youths with conduct disorder (CD) have reported disparate findings. We investigated WM alterations in a large sample of youths with CD, and examined the influence of sex and callous-unemotional (CU) traits. METHOD: DTI data were acquired from 124 youths with CD (59 female) and 174 typically developing (TD) youths (103 female) 9 to 18 years of age. Tract-based spatial statistics tested for effects of diagnosis and sex-by-diagnosis interactions. Associations with CD symptoms, CU traits, a task measuring impulsivity, and the impact of comorbidity, and age- and puberty-related effects were examined. RESULTS: Youths with CD exhibited higher axial diffusivity in the corpus callosum and lower radial diffusivity and mean diffusivity in the anterior thalamic radiation relative to TD youths. Female and male youths with CD exhibited opposite changes in the left hemisphere within the internal capsule, fornix, posterior thalamic radiation, and uncinate fasciculus. Within the CD group, CD symptoms and callous traits exerted opposing influences on corpus callosum axial diffusivity, with callous traits identified as the unique clinical feature predicting higher axial diffusivity and lower radial diffusivity within the corpus callosum and anterior thalamic radiation, respectively. In an exploratory analysis, corpus callosum axial diffusivity partially mediated the association between callous traits and impulsive responses to emotional faces. Results were not influenced by symptoms of comorbid disorders, and no age- or puberty-related interactions were observed. CONCLUSION: WM alterations within the corpus callosum represent a reliable neuroimaging marker of CD. Sex and callous traits are important factors to consider when examining WM in CD.

Callous-unemotional traits and brain structure: Sex-specific effects in anterior insula of typically-developing youths.

Callous-unemotional traits are characterized by a lack of empathy, a disregard for others' feelings and shallow or deficient affect, such as a lack of remorse or guilt. Neuroanatomical correlates of callous-unemotional traits have been demonstrated in clinical samples (i.e., adolescents with disruptive behavior disorders). However, it is unknown whether callous-unemotional traits are associated with neuroanatomical correlates within normative populations without clinical levels of aggression or antisocial behavior. Here we investigated the relationship between callous-unemotional traits and gray matter volume using voxel-based morphometry in a large sample of typically-developing boys and girls (N = 189). Whole-brain multiple regression analyses controlling for site, total intracranial volume, and age were conducted in the whole sample and in boys and girls individually. Results revealed that sex and callous-unemotional traits interacted to predict gray matter volume when considering the whole sample. This interaction was driven by a significant positive correlation between callous-unemotional traits and bilateral anterior insula volume in boys, but not girls. Insula gray matter volume explained 19% of the variance in callous-unemotional traits for boys. Our results demonstrate that callous-unemotional traits are related to variations in brain structure beyond psychiatric samples. This association was observed for boys only, underlining the importance of considering sex as a factor in future research designs. Future longitudinal studies should determine whether these findings hold over childhood and adolescence, and whether the neuroanatomical correlates of callous-unemotional traits are predictive of future psychiatric vulnerability. General scientific summary: This study suggests that callous-unemotional traits have a neuroanatomical correlate within typically developing boys, but not girls. Bilateral anterior insula volume explains up to 19% of the variance in callous-unemotional traits in boys.

Community Violence Exposure and Conduct Problems in Children and Adolescents with Conduct Disorder and Healthy Controls.

Exposure to community violence through witnessing or being directly victimized has been associated with conduct problems in a range of studies. However, the relationship between community violence exposure (CVE) and conduct problems has never been studied separately in healthy individuals and individuals with conduct disorder (CD). Therefore, it is not clear whether the association between CVE and conduct problems is due to confounding factors, because those with high conduct problems also tend to live in more violent neighborhoods, i.e., an ecological fallacy. Hence, the aim of the present study was: (1) to investigate whether the association between recent CVE and current conduct problems holds true for healthy controls as well as adolescents with a diagnosis of CD; (2) to examine whether the association is stable in both groups when including effects of aggression subtypes (proactive/reactive aggression), age, gender, site and socioeconomic status (SES); and (3) to test whether proactive or reactive aggression mediate the link between CVE and conduct problems. Data from 1178 children and adolescents (62% female; 44% CD) aged between 9 years and 18 years from seven European countries were analyzed. Conduct problems were assessed using the Kiddie-Schedule of Affective Disorders and Schizophrenia diagnostic interview. Information about CVE and aggression subtypes was obtained using self-report questionnaires (Social and Health Assessment and Reactive-Proactive aggression Questionnaire (RPQ), respectively). The association between witnessing community violence and conduct problems was significant in both groups (adolescents with CD and healthy controls). The association was also stable after examining the mediating effects of aggression subtypes while including moderating effects of age, gender and SES and controlling for effects of site in both groups. There were no clear differences between the groups in the strength of the association between witnessing violence and conduct problems. However, we found evidence for a ceiling effect, i.e., individuals with very high levels of conduct problems could not show a further increase if exposed to CVE and vice versa. Results indicate that there was no evidence for an ecological fallacy being the primary cause of the association, i.e., CVE must be considered a valid risk factor in the etiology of CD.

Are cannabis prevalence estimates comparable across countries and regions? A cross-cultural validation using search engine query data.

BACKGROUND: Prevalence estimation of cannabis use is usually based on self-report data. Although there is evidence on the reliability of this data source, its cross-cultural validity is still a major concern. External objective criteria are needed for this purpose. In this study, cannabis-related search engine query data are used as an external criterion. METHODS: Data on cannabis use were taken from the 2007 European School Survey Project on Alcohol and Other Drugs (ESPAD). Provincial data came from three Italian nation-wide studies using the same methodology (2006-2008; ESPAD-Italia). Information on cannabis-related search engine query data was based on Google search volume indices (GSI). (1) Reliability analysis was conducted for GSI. (2) Latent measurement models of "true" cannabis prevalence were tested using perceived availability, web-based cannabis searches and self-reported prevalence as indicators. (3) Structure models were set up to test the influences of response tendencies and geographical position (latitude, longitude). In order to test the stability of the models, analyses were conducted on country level (Europe, US) and on provincial level in Italy. RESULTS: Cannabis-related GSI were found to be highly reliable and constant over time. The overall measurement model was highly significant in both data sets. On country level, no significant effects of response bias indicators and geographical position on perceived availability, web-based cannabis searches and self-reported prevalence were found. On provincial level, latitude had a significant positive effect on availability indicating that perceived availability of cannabis in northern Italy was higher than expected from the other indicators. CONCLUSION: Although GSI showed weaker associations with cannabis use than perceived availability, the findings underline the external validity and usefulness of search engine query data as external criteria. The findings suggest an acceptable relative comparability of national (provincial) prevalence estimates of cannabis use that are based on a common survey methodology. Search engine query data are a too weak indicator to base prevalence estimations on this source only, but in combination with other sources (waste water analysis, sales of cigarette paper) they may provide satisfactory estimates.