How common is violence by psychiatric patients? (Lancet Psych 2021) 


How common is violence by psychiatric patients? (Lancet Psych 2021) 

The Question

What do evidence suggest as the risk of violence by individuals with different diagnoses? Violence is a broad term encompassing a variety of behaviours. In this narrative review, Daniel Whiting, Paul Lichtenstein, Seena Fazel looked at officially registered violent crimes (arrests / conviction) ie in the context of contact with criminal justice system and its associations with mental illnesses.

The public perceive violence in people with mental illness. This perception contributes to stigma towards mental illness. Stigma obviously is associated negative consequences on those with mental illness. Violence risk is a factor that matters a lot in decisions of compulsory admissions and as such the notions of violence can have significant impact on freedom of those with mental illness. There is also significant concern about violence towards those with mental illness.


There is an increased risk of violence in people with schizophrenia spectrum disorders even after adjusting for substance misuse. Odds of homicide is higher than for other violent crimes. The absolute life time risk of perpretation is very low (o.3%). A consistent finding that we all know from our clinical work is the mediating role of comorbid substance misuse in such crimes. Substance use typically doubles the risk of violence. Sibling studies (where you compare affected sibling against unaffected siblings to reduce confounding) show that unaffected sibling has an OR of 1.89 as against the general population. Affected sibling’s OR is 4·2 (95% CI 3·8–4·5).  It is worth noting that unaffected siblings are also showing more violent behaviours. 

What about life time risk?

Over a lifetime, 23% of individuals with a diagnosis of schizophrenia in Sweden had a violent conviction.

When is risk more?

The five-year incidence of violent conviction after first diagnosis was 11% in men and 3% in women.In first- episode psychosis, around one in ten individuals perpetrated physical interpersonal violence in the 1–3 years after first contact with clinical services. Over half of patients with violent offending histories committed their first violent crime before illness onset or first diagnosis. In individuals with psychotic illnesses who commit homicide, around 40% do so during the first episode of illness before treatment initiation. This highlights the need for early assessment and monitoring of all at risk symptoms. How much of this PRE ILLNESS/ PRE DIAGNOSIS/ PRE CONTACT violence is due to substance use or socio economic / familial factors or pre-symptoms (like behavioural disinhibition, social dislocation , affective disconnect, traumatic distress etc) of the illness is not known. 

If we look at the big picture of all homicides in a population, individuals with schizophrenia spectrum disorders typically represent less than 10% of all homicide convictions. It is possible that individuals with psychotic disorders are more likely to be convicted than others perhaps due to them revealing intent / leaving more evidences for conviction. 

Stranger Homicide & Schizophrenia

The risk of stranger homicide by those with schizophrenia is very very rare. It is estimated to be one in 14 million. (i.e. 0.06 per 100 000 general population). Compare this with death by lightning in the UK (0·05 per 100 000 population)!


Individuals with bipolar disorder had a 5-fold increased risk of violence compared with the general population. This OR comes down to below 3 when we adjust for sociodemographic factors and substance misuse (adjusted OR 2·8, 95% CI 2·5–3·1). Sibling studies show that the risk of violent crime in unaffected siblings (compared with the general population) is higher, suggesting familial confounding (1·2, 1·1–1·4).Absolute rates of violent crime are estimated at 8% in men and 2% in women with bipolar disorder, of which 70% occur in the 5 years after diagnosis,7 and 11% when using a lifetime outcome.


It is surprising to see that depressed individuals have a three times increased risk of violent offending after diagnosis, compared with the general population (adjusted OR 3·0, 95% CI 2·8–3·3)

Other diagnoses 

ADHD before the age of 18 years show an increased odds of violence-related arrest of 3·6 (95% CI 2·3–5·7) compared with controls who did not have ADHD.Unaffected siblings also show more violence. 

Autistic spectrum disorders: No consistent association is found. Nearly 30% of those with ASD has comorbid ADHD or Conduct Disorder and this may account for high risk shown in some studies.

Personality disorder is associated with a three times increased risk of violence. As we know, antisocial personality disorder is the most important one in this regard. In terms of absolute risks 8% of individuals with a diagnosis of any personality disorder perpetrated a violent offence in up to 10 years of follow-up.

Substance misuse: A key longitudinal study show Hazard Ratio for violence as 6·2 for men (95% CI 14·6–17·9) and 36·0 for women (27·0–48·0). Evidence for independent associations with violence for individual drugs are more uncertain. Hazard Ratio for violent offending in alcohol use disorders is 9·0 for men [95% CI 8·2–9·9] and 19·8 for women (14·6–26·7)


Some psychiatric disorders are clearly associated with increased relative risk of violence. Absolute risks (up to 10% each in schizophrenia and personality disorders, and over 10% in substance use disorders) suggest that these are still very high priority areas for mental health practice. Risk assessment, Risk communication and Risk reduction are integral to good clinical practice. The findings of this review highlight some key areas with in clinical pathways were we might want to invest more to reduce risk of violence.

Article: Whiting D, Lichtenstein P, Fazel S. Violence and mental disorders: a structured review of associations by individual diagnoses, risk factors, and risk assessment. Lancet Psychiatry. 2021 Feb;8(2):150-161. doi: 10.1016/S2215-0366(20)30262-5. Epub 2020 Oct 20. PMID: 33096045. DOI: 10.1016/S2215-0366(20)30262-5


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: