Relative to those with mental disorder alone, clients with co-occurring mental and substance abuse disorders are at double the risk of violence. Little theory and research are available to inform efforts to monitor these clients'violence risk state, even though clinicians may be sued for negligence when they fail to protect third parties from harm. The proposed research adapts a Conditional Model of Prediction (CMP) to enhance understanding of clients'risk state and how it can be monitored in the outpatient context. The adapted CMP shifts focus from clinical prediction to clients'self prediction. With a lifetime of experience in a wide variety of situations, individuals often forecast their own behavior more accurately than external evaluators. Although it may seem counterintuitive, individuals often provide accurate self-reports, even when there are incentives to deceive. According to the CMP, clients have experienced-based schemas that specify the kind of violence they might become involved in, given particular conditions (e.g., drinking). We posit that clients'knowledge of their own "if...then" patterns (Mischel &Shoda, 1995) equips them to assess their own risk state, and that clients'accuracy is enhanced with interviews that facilitate careful consideration of those patterns.
AIMS : Our primary aims are to (1) compare the accuracy of patients'self predictions of violence risk with that of clinical judgment and two clinically feasible actuarial tools;(2) assess whether the accuracy of patients'self predictions of violence risk is increased with "cognitive scaffolding" (i.e., a clinical interview about past violence-relevant experiences);(3) explore whether patients'accuracy is based on understanding of their own, risk-relevant "if...then" patterns;and, (4) determine whether patients make lower self-predicted violence risk assessments to clinicians than researchers, to explore generalizability. A secondary aim is to assess how antisocial and antagonistic traits - which increase risk of violence - affect the accuracy of self-prediction. METHOD:
These aims will be addressed via a prospective study of 554 patients with co-occurring disorders. Patients will be randomly assigned to one of two conditions: elicitation of self predictions either with- or without- cognitive scaffolding. We will assess patients'risk and elicit clinicians'risk judgments while patients are hospitalized (baseline), and then conduct a brief follow-up interview with patients 8 weeks after discharge (patient follow-up), and full interviews with patients and collateral informants 20 weeks after discharge (joint follow-up) to assess violence.
For Aim 1, the baseline and follow-up data permit a comparison of the accuracy of self-predictions, clinical judgment, and actuarial tools.
For Aim 2, the experimental manipulation permits a comparison of the accuracy of self-predictions when elicited with- or without- cognitive scaffolding.
For Aim 3, an assessment of situations related to self predictions (e.g., provocation) and situations experienced (or not) during the follow-up permit a test of whether self-predictions manifest greater accuracy when their evaluation takes situations into account.
Aim 4 will be addressed by comparing average levels of self predicted risk elicited in this research sample with a sample of self-predictions elicited by hospital clinicians. Our assessment of psychopathic, antisocial, and borderline traits will permit us to assess how these traits affect the level and accuracy of self predicted risk. SIGNIFICANCE: Policy makers and citizens are concerned about the adequacy of violence risk assessment and treatment services for high risk clients. If the proposed research supports the novel hypothesis that self prediction accurately captures violence risk state, its potential to assist clinicians in managing risk, promoting clients'integration in the community, and enhancing public safety is unparalleled.

Public Health Relevance

Assuming we validate our model and find support for its generalization, self-prediction has extraordinary potential to assist mental health clinicians in their ongoing efforts to assess and manage violence risk. For high risk clients with co-occurring disorders, self-prediction may promote treatment engagement, highlight risk factors to target in treatment, and promote community integration. These benefits to clinicians and clients translate directly into benefits for society. Given that high risk clients account for most of the violent incidents in which clients are involved, the availability of an empirically-informed, feasibly-implemented method of monitoring their violence risk can only enhance public health and safety.

Agency
National Institute of Health (NIH)
Institute
National Institute of Mental Health (NIMH)
Type
Research Project (R01)
Project #
7R01MH083799-05
Application #
8804075
Study Section
Adult Psychopathology and Disorders of Aging Study Section (APDA)
Program Officer
Shoham, Varda
Project Start
2010-03-16
Project End
2014-12-31
Budget Start
2013-01-08
Budget End
2014-12-31
Support Year
5
Fiscal Year
2014
Total Cost
$297,679
Indirect Cost
$107,469
Name
University of California Berkeley
Department
None
Type
Schools of Social Work
DUNS #
124726725
City
Berkeley
State
CA
Country
United States
Zip Code
94704