Calibrating violence risk assessments for uncertainty.

Psychiatrists and other mental health clinicians are often tasked with assessing patients' risk of violence. Approaches to this vary and include both unstructured (based on individual clinicians' judgement) and structured methods (based on formalised scoring and algorithms with varying scope for clinicians' judgement). The end result is usually a categorisation of risk, which may, in turn, reference a probability estimate of violence over a certain time period. Research over recent decades has made considerable improvements in refining structured approaches and categorising patients' risk classifications at a group level. The ability, however, to apply these findings clinically to predict the outcomes of individual patients remains contested. In this article, we review methods of assessing violence risk and empirical findings on their predictive validity. We note, in particular, limitations in calibration (accuracy at predicting absolute risk) as distinct from discrimination (accuracy at separating patients by outcome). We also consider clinical applications of these findings, including challenges applying statistics to individual patients, and broader conceptual issues in distinguishing risk and uncertainty. Based on this, we argue that there remain significant limits to assessing violence risk for individuals and that this requires careful consideration in clinical and legal contexts.

Can machine-learning methods really help predict suicide?

PURPOSE OF REVIEW: In recent years there has been interest in the use of machine learning in suicide research in reaction to the failure of traditional statistical methods to produce clinically useful models of future suicide. The current review summarizes recent prediction studies in the suicide literature including those using machine learning approaches to understand what value these novel approaches add. RECENT FINDINGS: Studies using machine learning to predict suicide deaths report area under the curve that are only modestly greater than, and sensitivities that are equal to, those reported in studies using more conventional predictive methods. Positive predictive value remains around 1% among the cohort studies with a base rate that was not inflated by case-control methodology. SUMMARY: Machine learning or artificial intelligence may afford opportunities in mental health research and in the clinical care of suicidal patients. However, application of such techniques should be carefully considered to avoid repeating the mistakes of existing methodologies. Prediction studies using machine-learning methods have yet to make a major contribution to our understanding of the field and are unproven as clinically useful tools.

Prevalence of comorbid substance use in schizophrenia spectrum disorders in community and clinical settings, 1990-2017: Systematic review and meta-analysis.

BACKGROUND: Comorbidity is highly prevalent between substance use disorders (SUDs) and schizophrenia. This systematic review and meta-analysis estimated prevalence rates of SUDs in epidemiological and treatment-seeking patients diagnosed with schizophrenia or first episode psychosis. METHODS: A literature search of Medline, EMBASE, PsycINFO and CINAHL databases was conducted from 1990 to 2017 inclusive. Prevalence of co-morbid SUDs and schizophrenia were extracted and odds ratios (ORs) were calculated using random effects meta-analysis. Combining like studies was dictated how authors reported substance use. RESULTS: There were 123 included articles with a total sample size of 165,811 subjects that yielded six epidemiological studies, 11 national or state case-registry studies, 20 large cohort studies and 86 clinical studies using in- or out-patient samples. The prevalence of any SUD was 41.7%, followed by illicit drugs (27.5%), cannabis (26.2%), alcohol (24.3%) and stimulant use (7.3%). Meta-analysis showed the pooled variance of any SUD in males was 48% which was significantly higher than that for females with schizophrenia (22.1%, OR 3.43, 95% CI 3.01, 3.92). Patients with SUD had an earlier age of onset of schizophrenia. Meta-regression showed prevalence increased over time for illicit drugs but not for other substances, including alcohol. CONCLUSIONS: The meta-analysis revealed that SUDs in schizophrenia is highly prevalent and rates have not changed over time. This indicates SUD are difficult to treat in this patient population and there is an urgent need for more informative studies to help develop better prevention, detection and treatment of SUDs in persons with schizophrenia and co-morbid disorders.

Characteristics of people attending psychiatric clinics in inner Sydney homeless hostels.

OBJECTIVE: To describe the characteristics of people attending mental health clinics at shelters for the homeless in inner city Sydney. DESIGN: Retrospective review of medical records of homeless hostel clinic attenders. SETTING: Mental health clinics located in three inner city homeless hostels. PARTICIPANTS: Consecutive series of clinic attenders, 21 July 2008 - 31 December 2016. MAIN OUTCOME MEASURES: Demographic characteristics; social, medical and mental health histories of homeless people. RESULTS: 2388 individual patients were seen at the clinics during the 8.5-year study period. Their mean age was 42 years (standard deviation, 13 years), 93% were men, and 56% were receiving disability support pensions. 59% of attenders had been homeless for more than a year, and 34% of all attenders reported sleeping in the open. The most common diagnoses were substance use disorder (66%), psychotic illness (51%), acquired brain injury (14%), and intellectual disability (5%). Most patients had more than one diagnosis. Early life and recent trauma was reported by 42% of patients. Pathways to homelessness included release from prison (28% of the homeless), discharge from a psychiatric hospital (21%), loss of public housing tenancy (21%), and inability to pay rent because of problem gambling. CONCLUSIONS: The high rates of substance use and mental disorder among homeless people in inner Sydney confirms the need for increased access to treatment for these conditions in this setting. Homelessness among those with mental illness might be reduced by developing alternative housing models, and supporting people with multiple problems to retain tenancy.

La práctica de la evaluación del riesgo de violencia en España / The practice of violence risk assessment in Spain

Antecedentes. La valoración del riesgo de violencia es un requisito fundamental en la toma de decisiones profesionales que implican prevenir, intervenir o informar sobre la conducta de las personas. El uso de herramientas estructuradas de evaluación del riesgo de violencia ha mostrado mejoría en la precisión de las evaluaciones basadas exclusivamente en el juicio clínico o en la pericia de un experto en contextos psiquiátricos, penitenciarios y jurídicos. Objetivo. Este estudio presenta los resultados de la primera encuesta sobre las prácticas profesionales asociadas al uso de herramientas de evaluación del riesgo de violencia en España. Materiales y métodos. La información fue recogida mediante la administración de una encuesta en internet que fue distribuida por correo electrónico a los miembros de organizaciones profesionales. Resultados. De manera similar a los contextos profesionales del resto del mundo, las escalas de psicopatía de Robert Hare (Psychopathy Checklist-Revised y Psychopathy Checklist: Screening Version) y el Historical-Clinical-Risk Management-20 encabezaron la lista de las herramientas más usadas tanto por elección personal como por requisito institucional. Conclusiones. Se ofrecen datos novedosos sobre la prevalencia de uso y la utilidad percibida de las herramientas estructuradas, así como sobre otras cuestiones relacionadas a las prácticas profesionales de evaluación del riesgo de violencia en España que pueden orientar tanto a los profesionales de contextos sanitarios, correccionales y forenses, como a los responsables de las instituciones en la elección de las herramientas a implementar para asistirlos en la toma de decisiones.

Background. Violence risk assessment is a key requirement in professional decision making involving prevention, intervention or reporting on human behavior. The use of structured tools for violence risk assessment has shown to improve the accuracy of assessments based exclusively on clinical judgment or expertise in psychiatric, correctional and legal settings. Objectives. This study presents results of the first survey about professional practices associated with tools for violence risk assessment in Spain. Materials and methods. The information was collected by administering an online-based survey that was distributed by e-mail to members of professional organizations around the country. Results. As in professional contexts worldwide, the Robert Hare's psychopathy scales (Psychopathy Checklist-Revised and Psychopathy Checklist: Screening Version) and the Historical-Clinical-Risk Management-20 topped the list of the most used tools both by professional choice and institutional requirement. Conclusions. We provide novel data on the prevalence of use and the perceived utility of specific tools, as well as on other issues related to the professional practice of violence risk assessment in Spain, which can guide professional in the health care, correctional and forensic settings, as well as those responsible for decisions in institutions about choosing which tool to implement.

Stimulant use disorders in people with psychosis: a meta-analysis of rate and factors affecting variation.

OBJECTIVE: Stimulant abuse and dependence often complicate the care of people with psychotic disorders. This study systematically reviews the prevalence estimates reported for stimulant abuse and dependence in people with psychotic disorders, and examines personal, clinical, regional and methodological factors which explain variation in these rates. METHODS: PsychINFO, EMBASE and MEDLINE (1946-2013) were searched systematically for studies reporting on stimulant drug use disorders in representative samples of people with psychotic disorders. Random effects models estimated the pooled rate of a stimulant use disorder, defined to include stimulant abuse and stimulant dependence. Study characteristics associated with heterogeneity in rates of stimulant use disorder were examined by subgroup analyses for categorical variables, by meta-regression for continuous independent variables and by multiple meta-regression. RESULTS: Sixty-four studies provided 68 estimates of lifetime or recent stimulant use disorders in 22,500 people with psychosis. The pooled rate of stimulant use disorder was 8.9% (95% CI 7.4%, 10.5%). Higher rates of stimulant use disorders were reported in studies of affective psychosis, studies from inpatient settings, studies from the USA and Australia, and studies with higher rates of cannabis disorder; in multiple meta-regression analysis these factors explained 68% of between-study variance. Rates of stimulant use disorder were stable over time, and unrelated to age, sex, stage of psychosis, type of stimulant drug or study methodology factors. CONCLUSIONS: Reported rates of stimulant use disorder in people with psychosis are much higher than in the general population but vary widely and are associated with regional, service setting and clinical differences between studies. It is likely that stimulants contribute to the overall burden of psychosis, and that social and environmental factors combine with drug and illness-related factors to influence stimulant use in psychosis.

Suicide risk categorisation of psychiatric inpatients: what it might mean and why it is of no use.

OBJECTIVE: Existing suicide risk assessment policies and guidelines do not specify whether the risk to be assessed is absolute, relative to other patients or relative to the same patient at an earlier time. METHODS: We utilised the results of a recent study of inpatient suicide to explore the implications of each of the above types of risk. RESULTS: No interpretation of the current requirements for suicide risk categorisation of psychiatric inpatients can lead to information that might usefully guide clinical decision-making. CONCLUSIONS: Policies and guidelines requiring suicide risk categorisation of inpatients should be withdrawn. The College should delete the mandatory entrustable professional activity requiring risk assessment.

A 21-year retrospective outcome study of New South Wales forensic patients granted conditional and unconditional release.

AIM: To retrospectively examine the outcomes of forensic patients found not guilty by reason of mental illness (NGMI) in New South Wales (NSW) and subsequently released into the community, as measured by reoffending, conditional release revocation and psychiatric hospital readmission. METHOD: Data were collected from the NSW Mental Health Review Tribunal (MHRT) files for all patients who received an NGMI verdict between January 1990 and December 2010, and who were released into the community during this period. The outcome measures of conditional release revocation and psychiatric hospital readmission were extracted from these files. Information about subsequent criminal charges, convictions and penalties were obtained from the Bureau of Crime Statistics and Research's reoffending database. RESULTS: During the 21-year period studied, 364 offenders received an NGMI verdict and were placed under the supervision of the MHRT. Of these, 197 were released into the community, including 85 who were granted unconditional release. Over a follow-up period averaging 8.4 years, 18% of conditionally released patients reoffended, 11.8% were convicted of a further offence, 8.7% were charged with a violent offence, 3.1% were convicted of a violent offence and 3.7% were sentenced to a term of imprisonment. Five (3.1%) conditionally released forensic patients received a further NGMI verdict. One-quarter of the conditionally released patients had their conditional release revoked and half were readmitted to hospital. Of the forensic patients granted unconditional release, 12.5% were charged with an offence, 9.4% received convictions for an offence, 6.3% were charged with a violent offence and 4.7% were convicted of a violent offence, in a mean follow-up period of 7.6 years. None committed a further serious offence resulting in a term of imprisonment, nor a second NGMI verdict. CONCLUSIONS: The results of this study suggest that the treatment and rehabilitation of forensic patients in NSW, together with the decision-making procedures of the MHRT, is effective in protecting the community from further offending by forensic patients.