Preventing serious adverse outcomes in schizophrenia

Jul 10 2014

A very positive new development in clinical academic psychiatry has been the recent launch of a new publication by The Lancet, called The Lancet Psychiatry. This improves the visibility of mental health research and will hopefully increase the translation of such research into clinical practice.

The first issue contained a particularly important paper by Seena Fazel and colleagues that explored the rates of violent crime, suicide, and premature death in people with schizophrenia, and the factors that increased the risk of these outcomes (Fazel et al, 2014). Understanding the risk factors is critical to performing a risk assessment, to establish whether they are suitable for preventive intervention, and to see to what extent the same factors predict all outcomes.

The paper itself is the kind to turn some researchers (like me!) green with envy, because of the enormous sample with excellent follow-up; nearly 25,000 patients with schizophrenia and almost half a million matched controls! The reason such a large, high-quality dataset was available was because the research took advantage of the longitudinal nationwide population registers that exist in Sweden. These registers cover all inpatient admissions (and since 2001 also outpatients), cause of death, health insurance, crime and labour market statistics, and details about assessments conducted at conscription to national service, which is compulsory for all male Swedes. It is also possible, through the Multi-generation Register, to link people to their parents and siblings and thus examine the contribution of parental and familial factors on adverse outcomes.

The huge high-quality dataset for this study comes from the Swedish longitudinal nationwide population register, but how applicable is this to other countries?

Outcome measures

Three major adverse outcomes were selected for investigation:

Conviction for a violent offense (including homicide, attempted homicide, aggravated assault, common assault, robbery, arson, any sexual offence, illegal threats/intimidation)
Premature mortality (defined as death before age 56 – while this is obviously an early death it is not clear why 56 specifically should be chosen)
Death by suicide (this included undetermined deaths – and was able to include more than 99% of deaths in Swedish residents, even if the death occurred outside Sweden)

Results

Severe outcomes

Severe outcomes are much more likely in patients diagnosed with schizophrenia (unsurprisingly). Indeed, they are:

7.4 times more likely to be convicted of violent offences
8.1 times more likely to die prematurely, and
20.7 times more likely to kill themselves

Risk for these severe outcomes was also significantly higher in the siblings of patients as compared to the general populations, although the risk was only between 1.5 and 2.5 times that of the control group. This suggests that at least part of the increased risk is the result of familial factors, not the diagnosis itself.

This study provides further evidence that the future is bleak for many people living with psychosis.

Risk factors

In terms of risk factors, three important predictors of all outcomes were:

Drug and alcohol use disorders
A history of previous conviction for violence
A history of self-harm

Although other risk factors were significant (for example, marital status predicted suicide and self-harm, and low family income was important for predicting violent offenses), these three are distinguishable in being present before the diagnosis and are potentially modifiable.

Importantly, these risk factors are also the same ones that predict these outcomes in siblings and in the general population. Indeed, the association between the risk factors and the outcomes (the hazard ratios) are higher in the general population, indicating that if interventions could be implemented at the population level, they would reduce a number of these severe outcomes across the board, and not just in those with schizophrenia.

Major risk factors were drug and alcohol use disorders and a history of violence or self-harm.

Caveats

There are some issues that could affect the study:

Controls and siblings were free of schizophrenia/non-affective psychosis, but may have had other disorders
There is an assumption that data from Sweden can be generalised to other western countries, but this may well not be the case (the commentary in The Lancet Psychiatry points out that the level of unmet need in Sweden is significantly lower than other countries, for example)
There was a general lack of information about the way in which some variables were obtained. For example, no detail was provided about how self harm was detected through the registers
There is a high degree of missing data for the most novel risk predictors (IQ and birth complications). In the case of IQ they are only available in a little over half the males with schizophrenia (from the draft registry), and for birth complications they are only available for about a quarter of the total patient sample (males and females)
There was no investigation of the role of treatment variables (particularly antipsychotic medication) in predicting violent offences, suicide or premature death

Summary

This large population-cohort study clearly demonstrates that people with schizophrenia are at greatly increased risk of violent offences, premature death, and suicide.

Dealing with these issues, particularly violence, in those with mental health disorders, must always be done carefully because of the stigma and discrimination that can result. Three aspects of the paper need emphasis:

First, that association between schizophrenia and suicide is much more powerful than that between schizophrenia and violence
Second, that the vast majority of patients with schizophrenia are neither violent nor suicidal
Finally, that the main policy implications of these findings are that interventions to reduce these severe outcomes should be targeting at the general population, rather than at patients specifically