Empirical evidence of a relationship between insight and risk of violence in the mentally ill: a review of the literature

Author:
BJORKLY Stal
Journal article citation:
Aggression and Violent Behavior, 11(4), July 2006, pp.414-421.
Publisher:
Elsevier

Lack of insight has established itself both as an important element of models of risk of violence and as a clinical item in structured approaches to measuring risk of violence. The main purpose of this paper is to delineate findings from a review of the literature concerning the empirical basis for the existence of a relationship between insight and violence risk. A second aim is to examine whether certain dimensions of insight may have a higher violence-triggering potential, and, if so, to learn more about valid and reliable approaches for measuring these dimensions. Results of this review show that there are very few empirical studies that demonstrate a possible association between insight and violence risk. Five out of a total of nine investigations found a positive relationship between poor insight and violent behavior. Nonetheless, the strength and specific nature of the relationship remain unclear due to considerable methodological and conceptual shortcomings in the reviewed studies. Seven recommendations for future research are presented to address these shortcomings.

Extended abstract:
Author

BJORKLY Stal

Empirical evidence of a relationship between insight and risk of violence in the mentally ill: a review of the literature.

Journal citation/publication details

Aggression and Violent Behavior, 11(4), July-August 2006, pp.414-23.

Summary

It is widely assumed that insight into their condition will influence the danger that mentally ill people may present to others, and insight is included as an item in several structured approaches to measuring the risk of violence. However, this review of nine studies shows that there is very little robust evidence demonstrating a possible association between poor insight and violent behaviour, or its strength and specific nature. The author makes seven recommendations to improve the quality of future research in this area.

Context

Three reasons are given for conducting the review. First to discover whether there is a demonstrable link between insight and risk of violence that might be incorporated into the HCR-20 (Historical, Clinical and Risk Management) risk assessment tool. Second, to test the claim of a 1999 study that it was the first to demonstrate such a link. And third, as part of the process of planning a pilot study on the risk assessment of forensic patients.

Methods

What sources were used?

Searches were carried out on: PsycINFO, 1967-June 2005; Medline, 1966-June 2005; EMBASE (Excerpta Medica), 1908-June 2005; ProQuest, 1992-June 2005; Science Direct (1995-June 2005); and ISI Web of Science (1975-June 2005). ProQuest is an information provider rather than a database, while Science Direct gives access to journals and other material published by Elsevier. Updating searches were carried out in September 2005. The reference lists of all relevant papers were checked, as was a referenced, online bibliography about HCR-20 published in 2005.

 

What search terms/strategies were used?

Four separate searches were carried out, with the first combining 'review of' with sets of terms covering risk assessment, violence and psychiatric patients. Search two repeated the first search, excluding 'review of'. Search three combined 'review of' with terms covering risk assessment, violence, psychiatric patients and medication compliance. Search four repeated the third, with the exclusion of 'review of'. The terms used in each search are given, apparently in full, in the paper.

 

What criteria were used to decide on which studies to include?

English language papers on the association between insight and violence towards another person were sought, primarily prospective empirical studies in which a measurement of insight was followed up by the recording of aggressive acts. However, case control studies (retrospective investigations) were also included. Multi-dimensional definitions and measurement had to be used. This ruled out studies that used treatment compliance alone as an indicator of insight; those that relied only on indirect measures of indicators of insight; and those that used single-item indicators of insight. One exception to the last exclusion - the Positive and Negative Syndrome Scale - was allowed because its single insight items cover the aggregate assessment of three dimensions of insight.

 

Who decided on their relevance and quality?

The total number of hits delivered by the initial searches, and subsequent relevance and quality assessment, are not discussed.

 

How many studies were included and where were they from?

Nine studies (two prospective, seven retrospective) published between 1997 and 2005 met the inclusion criteria. Two each were from Australia , Great Britain and the USA , and one each from Spain , Sweden and Taiwan . They are summarised in Table 1, which gives details of samples, methods of measuring insight and violence, types of violence addressed and findings,

 

How were the study findings combined?

The variability of the studies precluded a meta-analysis. The criteria for rating a relationship between insight and violence were based on a significance level of p <.05 for differences in the presence of insight when comparing violent or non-violent groups; and/or a significance level of p <.05 for correlation between the presence of insight and violent behaviour; and/or an odds ratio for violence of at least 2 (within 95% confidence interval) in patients with lack of insight compared to those with insight.

Findings of the review

Empirical support for the existence of a relationship between insight and the risk of violence is found to be 'inconclusive', with five of the nine studies reporting a positive relationship between lack of insight and violence, and four reporting no relationship. Psychiatrists have suggested that insight in psychosis has three components - recognition of illness, awareness of the need for treatment, and ability to correctly label abnormal symptoms such as delusions - and one aim of the review was to identify which, if any, of these had the strongest predictive validity of violence. None of the studies provided information to help answer this question because of both conceptual and methodological shortcomings.

 

Nonetheless, the author identifies some interesting methodological differences between the positive and negative studies that provide useful information for the design of future research in this area.

 

Four of the positive studies, but only one of the negative studies, had a prospective or pseudo-prospective design, the latter defined as studies in which insight assessments were completed on the basis of information available when patients were discharged but where the data were not coded until later.

 

Three of the positive studies, and only one of the negative studies, used specialised instruments for the measurement of aggression.

 

The base rate of aggression was higher in the positive than in the negative studies (43% and 25% respectively).

 

Authors' conclusions

'Assumption of the presence of an insight-violence link is often cited in the literature. However, according to the findings of this review, systematic research on this association is still in its infancy, and the strength and specific nature of the relationship remain unclear.'

 

The author's recommendations thus relate to improving the quality of the evidence base. He calls for future research to: be prospective in nature; include more accurate measurement and analysis of the different components of insight; present differentiated measures and analysis relating to different degrees of aggression; and provide statistical power and effect size estimates to clarify the strength of a possible insight-violence link.

 

In addition, he advises researchers to focus in particular on the dynamics of aggression as a component of insight, to compare the relationship between insight and psychotic symptoms in patients with and without concomitant emotional distress, and to address the impact of risk management strategies.

Implications for policy or practice

None are discussed.

Subject terms:
literature reviews, mental health problems, risk, violence;
Content type:
systematic review
Link:
Journal home page
ISSN print:
1359-1789

Key to icons

  • Free resource Free resource
  • Journal article Journal article
  • Book Book
  • Digital media Digital media
  • Journal Journal

Give us your feedback

Social Care Online continues to be developed in response to user feedback.

Contact us with your comments and for any problems using the website.

Sign up/login for more

Register/login to access resource links, advanced search and email alerts