Summary: The aim of this project is to validate the Questionnaire sur les Troubles d’Origine traumatiques Précoces (Q-TOP), a French-language questionnaire designed to screen for and contribute to the assessment of reactive attachment disorder (RAD) and disinhibited social engagement disorder (DSED) among school-age children (5–12 years old). The English-language version of the instrument is called the Early TRAuma-related Disorders Questionnaire (ETRAD-Q). Various statistical analyses have demonstrated the instrument’s reliability (internal consistency, test-retest) and validity (factor, divergent, convergent). The ETRAD-Q consists of 42 items regarding children’s behaviour to be rated on a four-point Likert scale (0 = not at all true, 1 = a little bit true, 2 = pretty much true, 3 = very much true). Its subscales are based on the diagnostic criteria of the DSM-5 (APA, 2103). There are two for DSED: 1) interactions with unfamiliar adults; and 2) social disinhibition. There are three for RAD: 1) low selective attachment; 2) low social and emotional responsiveness; and 3) emotional unpredictability. The ETRAD-Q has been completed by about 1,500 caregivers of children 6–12 years old (1,000 children from the community, 100 adopted, 200 in foster care, 200 in residential care). One validation study of the 16-item short version of the questionnaire has been published (Monette et al., 2018), as has one of the 42-item long version (Monette et al., 2022). The ETRAD-Q kit (questionnaire, technical manual and corrector) is available in the “Instruments” section of the site.

Co-investigators: Chantal Cyr, Miguel Terradas, Sophie Couture

Funding: project start-up assistance from Institut Universitaire Jeunes en difficulté (IUJD)

Duration: 2016–….

Progress: primary data collection completed, primary validation study published, additional data collection ongoing (clinical groups: ADHD, ASD, norms, etc.).

Summary: 1) The primary aim of this project is to identify the risk factors in maltreatment history (type, severity, frequency, chronicity) associated with presence of RAD or DSED among school-age children in foster care or residential care (n = 150 in foster care, n = 120 in residential care). Its secondary aims are the following: 2) To provide validation data (inter-rater reliability) for the Extended Modified Maltreatment Classification System (EMMCS), an updated version of the MMCS (English et al., 1997, 2005), which serves to code the various dimensions of maltreatment (type/subtype, severity, frequency, chronicity) in CPS narrative reports; 3) To examine the prevalence of RAD and DSED among children in foster care and in residential care; 4) To examine the impact of RAD and DSED symptoms on children’s functioning in daily life; 5) To explore the association between RAD and DSED symptoms and executive functions (e.g., working memory, inhibition, flexibility); 6) To explore the association between RAD and DSED symptoms and social cognition; 7) To explore the association between DSED symptoms and different dimensions of impulsivity (negative urgency, lack of perseverance, lack of planning, thrill seeking, positive urgency); and 8) To examine the psychometric properties of the RAD and DSED Assessment (RADA) semi-structured interview.

Co-investigators: Chantal Cyr, Miguel Terradas, Sophie Couture

Funding: SSHRC (Insight Development Grant)

Duration: 2018–….

Progress: Data collection completed, article preparation/knowledge transfer

Summary: This project is a survey on the profession of neuropsychologist in Canada. The first edition in 2020 focused on a sample of 282 neuropsychologists from all provinces of Canada. It documented sociodemographic characteristics, type of academic degrees, professional activities, patients, and issues faced by neuropsychologists in different regions of Canada. At that time, the number of neuropsychologists in Canada was estimated at 1,670, making Canada the country with the highest rate of neuropsychologists per 100,000 inhabitants in the world. This project was renewed in 2025, in collaboration with the Canadian Psychological Association (CPA).

See the article from the 2020 edition:
Monette, S., Bertrand, J.-A., Perreau-Linck, E., Ramos-Usuga, D., Rivera, D., & Arango-Lasprilla, J. C. (2023). The profession of neuropsychologist in Canada: Findings of a national survey. The Clinical Neuropsychologist, 37(1), 1-33. https://doi.org/10.1080/13854046.2021.2002934

Co-researchers (2025 edition): Nicolás F. Narvaez Linares

Funding: None

Duration: 2025-2030

Principal investigators: Sébastien Monette and Chantal Cyr

Summary: 1) The aim of this project is to identify the risk factors in maltreatment history (type, severity, frequency, chronicity) associated with children’s development and the risk factors associated with re-reporting and family reunification breakdown (sample: 400 to 600 children 1 to 5 years old in CPS care) from the data in parenting capacity assessments (PCA). Detailed objectives: PART 1 – Predicting attachment disorder (RAD-DSED) behaviours and developmental delays: 1) To verify the prevalence of RAD and DSED behaviours among children victims of maltreatment; 2) To identify the risk factors associated with presence of RAD and DSED behaviours among children victims of maltreatment; 3) To verify presence of developmental delays (cognitive) among preschool-age children in CPS care; 4) To identify the risk factors associated with developmental delays among preschool-age children in CPS care; 5) To identify the predictors of developmental delay catch-up among preschool-age children in CPS care. PART 2 – Predicting mid-term outcomes (re-reporting and re-placement/family reunification breakdown): 6) To estimate re-reporting risk among preschool-age children in CPS care; 7) To identify the risk factors (including parenting capacity level and parent potential for change, estimated in the course of the PCA) associated with re-reporting and to verify their predictive value up to three years following PCA completion; 8) To estimate risk for re-placement/family reunification breakdown among preschool-age children in CPS care placed in a substitute setting and then reunited with a parent; 9) To identify the risk factors for re-placement/family reunification breakdown and to verify their predictive value up to three years following PCA completion.

Co-investigators: Sonia Hélie, George Tarabulsy et Karine Dubois-Comptois

Funding: SSHRC (Insight Grant)

Duration: 2022–2028

Progress: Data collection ongoing

Summary: This research project explores the reliability and validity of the Rorschach Performance Assessment System (R-PAS) and of the Roberts-2 with children in the care of Child Protective Services (CPS). The R-PAS (Meyer et al., 2011) is a revised and improved coding system for the famous implicit personality inkblot tests developed by Hermann Rorschach in 1921. The Robert-2 (Roberts & Gruber, 2005) is an updated version of the Roberts Apperception Test for Children (RATC; McArthur & Roberts, 1982), which requires children to tell a story based on drawings of family scenes. While the reliability and validity of the R-PAS has been demonstrated empirically with adults (see meta-analysis by Mihura et al., 2013), the test’s psychometric properties with children are less well established. The psychometric properties of the Roberts-2, for their part, have not been investigated much (only one independent study). We propose to verify the inter-rater reliability and convergent validity of the R-PAS and the Roberts-2 using data (tests/questionnaires: WISC-V, Conners CBRS, ETRAD-Q, TSC-YC, etc.) extracted from the files of children who have undergone a psychological assessment by a psychologist from the Clinical Support and Specific Intervention Department (SISSC) of the Integrated University Health and Social Services Centre (CIUSSS) of South-Centre Montreal (CCSMTL). This project will allow us to gain a better understanding of the strengths and limitations of the R-PAS and the Roberts-2 with children and possibly to generate new interpretative hypotheses for these tests, which will serve to arrive at more precise psychological assessments and better-adapted recommendations.

The project’s specific objectives are the following: 1) To verify the inter-rater reliability of the R-PAS variables using data from the files of children and adolescents in CPS care; 2) To verify the convergent validity of the R-PAS “Engagement and Cognitive Processing” domain scores; 3) To verify the convergent validity of the R-PAS “Perception and Thinking Problems” domain scores; 4) To verify the convergent validity of the R-PAS “Stress and Distress” domain scores; 5) To verify the convergent validity of the R-PAS “Self and Other Representation” domain scores; 6) To verify the reliability of the Roberts-2 variables; 7) To verify the convergent validity of the Roberts-2 variables.

Co-investigators: none

Funding: none

Duration: 2023–….

Progress: ongoing

Principal investigators: Sonia Hélie and Sébastien Monette

Summary:  The Quebec Incidence Study of Reported Child Abuse and Neglect (QIS/EIQ) is conducted every five years with Quebec’s CPS caseworkers. Aside from estimating the annual incidence of child maltreatment reports investigated by CPS, the study produces the only reliable province-wide data to describe the maltreatment situations, the psychosocial profiles of the children and their parents, and their housing conditions. At every QIS cycle (1998, 2003, 2008, 2014), the approach has been essentially the same: A representative sample of child maltreatment reports assessed by CPS from October 1 to December 31 of the target year is constituted (sample size varies from one cycle to the next; in 2014, it totalled 4,011 reports) and, for each report in the sample, the primary caseworker completes a study form (QISF) after completing each report assessment.

Another way to measure maltreatment is to code CPS report using a standardized coding grid. The instrument most often used for this purpose in research is the Modified Maltreatment Classification System (MMCS; English et al., 1997). It has been used in more than 75 studies (Huffhines et al., 2016) and is considered the most complete instrument of its kind by experts in the field (Jackson et al., 2019). The MMCS is a revised and improved version of the Maltreatment Classification System (MCS; Barnett et al., 1993), which was updated as part of the well-known LONGSCAN longitudinal study in the United States. A revision of the MMCS was called for, however, to take account of the findings of the past 25 years of research, primarily regarding psychological maltreatment. Whereas the MMCS comprises only one global category for “emotional maltreatment”, research has shown that various subtypes of emotional or psychological maltreatment are identifiable (Glaser 2002, 2011; Hart et al., 2017) and numerous researchers have underscored the importance of refining how psychological maltreatment is measured (English et al., 2015a, 2015b; Tailleu et al., 2016). S. Monette completed an update of the MMCS, which we call the Extended Modified Maltreatment Classification System (EMMCS). The reason for this update was to define different subtypes of psychological maltreatment. The EMMCS has all of the strong suits of the MMCS, namely, it still covers the following dimensions: 1) types and subtypes (augmented) of maltreatment; 2) severity per event; and 3) frequency of maltreatment events.  However, it also covers the following new dimensions: 4) developmental periods of exposure; and 5) maltreatment chronicity. Some parental risk factors and some child adverse life event risk factors are also now documented.

The aim of this project is to compare different methods of measuring maltreatment based on data in CPS files. More specifically, three methods will be applied to 200 files and compared: 1) a QIS form completed by a CPS caseworker, which is the usual method used in the QIS; 2) a QIS form completed by a research assistant (based on CPS narrative reports); and 3) coding of CPS narrative reports with the EMMCS by two research assistants. The sub-objectives of this study include verifying whether rates (McNemar test) and agreement (kappa statistic) for maltreatment are similar: 1) between the QIS form completed by a caseworker and the QIS form completed by a research assistant; and 2) between the QIS form completed by a caseworker and the EMMCS coded by two research assistants (consensual coding). Another sub-objective is to validate the EMMCS by: 3) determining the instrument’s inter-rater reliability (200 files coded by two research assistants) and: 4) determining the instrument’s convergent validity by comparing the maltreatment rates obtained with the EMMCS and with the QIS form completed by a research assistant.

Co-investigators: Delphine Collin-Vézina, Tonino Esposito, Nico Trocmé

Funding: Public Health Agency of Canada (PHAC) and Institut Universitaire Jeunes en difficulté (IUJD)

Duration: 2021–2024

Progress: data collection completed, article preparation/knowledge transfer

Summary: In Quebec, when a child is placed in out-of-home care following an intervention by child protective services (CPS), an attempt is made in the majority of cases (80%) to reunify them with their parents. However, a recent research report (Hélie et al., 2020), which followed nearly 10,000 children over 10 years by CPS, indicates that 40% of these family reunifications result in a reunification failure, also called “replacement”, within just 12 months. Another Quebec study (INESSS, 2019) involving 28,242 reunified children shows a similar replacement rate (34%) over a one-year observation period. This situation is concerning because a replacement often implies that the child has again been a victim of maltreatment in their family during reunification, and furthermore, they are likely to be placed in a different resource (e.g., foster family, group home, etc.) than the previous one, contributing to a trajectory of instability that is harmful to the child’s development. Studies have identified several predictors of replacement, which can be grouped under child characteristics (psychopathology, attachment issues, etc.), parental characteristics (substance abuse, mental disorders, single parenting, domestic violence, neglect, abandonment, etc.), and service history (replacement history, duration and type of placement, etc.). Decisions regarding family reunification largely rely on “unaided” clinical judgment. The predictive validity of clinical judgment is systematically lower than that of actuarial tools. The main objective of this project is to initiate the development and validation of an actuarial tool to predict the risk of replacement among children aged 0 to 12 years.

Co-researchers: Sonia Hélie, Geneviève Parent, Vanessa Lecompte, Chantal Cyr

Funding: IUJD Startup Fund and SSHRC Insight Development Grants

Duration: 2025-2027

Progress: Data collection to take place in 2025

Summary: This research project aims to harness the potential of generative artificial intelligence (AI), in the form of a local large language model (LLM), to automate the coding of narrative reports related to maltreatment (anonymized narrative investigation reports) from child protection services (CPS). More specifically, the main objective is to verify whether AI can complete a maltreatment classification grid updated by our team, the Extended Modified Maltreatment Classification System (EMMCS), which extracts several dimensions of maltreatment: types, subtypes, frequency, severity, timing of exposure, chronicity, as well as certain parental risk factors and other adverse life events for the child. These narrative data (i.e., coding investigation narratives from CPS files using the EMMCS) allows us to obtain variables impossible to obtain from CPS administrative data (i.e., a directly usable database, but without text fields), such as the severity of parental maltreatment events/behaviors. Furthermore, the narrative data consistently produces maltreatment rates (for each type) nearly double those of administrative data, raising questions about the validity of administrative data as a source of information on maltreatment experienced by children. In this project, the method consists of testing the effectiveness of different language models, particularly those that can operate locally, in order to avoid the confidentiality issues associated with transferring data to the cloud. The project is based on a sample of 240 cases already coded by humans using the EMMCS, from a previous research project (project #6). This data will be used to train and validate the AI ​​by comparing its performance to that of two human assistants (consensus coding). The control engineering strategy will be developed to optimize the accuracy of the results provided by the AI ​​by testing various parameters (presence of contextual information, examples, thought processes). Evaluation criteria will include accuracy, recall, F1 score, and concordance with human coding (Kappa coefficient and ICC). The goal is for the AI, if it performs well, to significantly reduce the time required to code narrative reports from several hours to just a few minutes. Ultimately, this innovation could transform research on malteatment, guide best practices in assessment of maltreatment, and contribute to a more nuanced understanding of risk factors and phenomena, all while respecting data confidentiality.

Co-researchers: Michel Rousseau

Funding: none

Duration: 2025-2027

Progress: in preparation

Projects in preparation :

• Predicting foster care placement breakdown
• Parent-child contact/visitation during placement