In May 2016, Winkler et al published a systematic review on cohort studies following up patients after discharge from long-term psychiatric hospital care. ^{Reference Winkler, Barrett, McCrone, Csémy, Janousková and Höschl1} The study did not show relevant numbers for imprisonment or homelessness after discharge. The authors concluded that the study contradicted ecological studies reporting a relationship between prison population rates and psychiatric bed numbers. They propose ecological fallacies as a possible explanation. In the related editorial, the ecological studies are referred to as arguing against deinstitutionalisation. ^{Reference Salisbury and Thornicroft2}

As an author of one of those ecological studies, I would like to comment. Rather than arguing against deinstitutionalisation of mentally ill people, the studies express concern that deinstitutionalisation does not occur in societies with massively increasing prison populations ^{Reference Mundt, Chow, Arduino, Barrionuevo, Fritsch and Girala3} and very high rates of severe mental illness among prisoners. ^{Reference Mundt, Kastner, Larrain, Fritsch and Priebe4} Although the relationship between psychiatric bed numbers and prison population rates in South America was rather strong, findings from Europe were less robust. ^{Reference Chow and Priebe5} In quantitative terms, prisons have become the most important facilities institutionalising mentally ill people in the Americas. Mentally ill people in prisons cause much more concern with respect to human rights than those in psychiatric hospitals. A way forward could be to improve care for people in prison, as well as improving community care for mentally ill people at risk of criminal justice involvement to prevent imprisonment. There is a broad consensus that short-term hospital admission is more efficient than longer stays, and that psychiatric hospital admission should be linked with community services in care systems. In the ecological studies, all types of psychiatric hospital beds were acknowledged. ^{Reference Mundt, Chow, Arduino, Barrionuevo, Fritsch and Girala3} The majority of beds nowadays are used to provide short-term care, including in low- and middle-income settings. Long-term hospital admission is no longer a common type of service provision in general psychiatry in the countries in which the studies that Winkler et al included in the review were conducted. ^{Reference Winkler, Barrett, McCrone, Csémy, Janousková and Höschl1} Therefore, the study seems rather of historical value.

It is not surprising that elderly people, after decades spent in hospital, have low criminogenic energy. Young people with severe mental illness and comorbid substance use disorders are of much more concern. For understanding the interdependence of penal justice systems and psychiatric in-patient care systems, recently published large linkage studies of registries are more relevant. These show very high rates of psychiatric hospital admission prior to imprisonment and in the year after release from imprisonment. They also show markedly elevated risks for people with mental disorders to commit violent crimes and to be victims of violence compared with the general population. What contribution short-term hospital care can make to postponing or preventing criminal justice involvement and protecting people with mental disorders is still unresolved. However, to reject findings from ecological studies, as in Winkler et al's review, may be a fallacy of categories.