From a Contributing Editor, Colleague and Friend of the Editor

This week’s reading is a provocative companion piece to the recent review of efforts to reduce involuntary admission to hospital. It is the environmental expression of the legal deprivation of freedom of movement: the locked door. Locked doors have a powerful symbolic meaning in psychiatry; outpatients coming for elective consultations sometimes tell me they are afraid if they “say the wrong thing” that I will “lock them up”. Asylum superintendents carried large rings of keys that embodied power and control.

Locked doors, better outcomes?

Having spent half my career working on inpatient units, I am, like almost all of you, familiar with the locked doors that distinguish our wards from all others found in a hospital. And I know the reasons for their justification: prevention of elopement by people at risk of harm to themselves and others. And that prevention is intended to serve not only the patient and family but also the clinicians and the institution in terms of risk management. And yet…people do elope. Sometimes they return and sometimes they do not. Sometimes they attempt or complete suicide and sometimes they do not.

There is, as always, a tension between safety and risk, between freedom and protection, between autonomy and control. Locks are ubiquitous but not universal on psychiatric wards. What do we know about whether they make a difference? And what would be the ethically acceptable methodology for determining it?

– David Goldbloom, OC, MD, FRCP(C)

Locked Doors and Risks

“Suicide risk and absconding in psychiatric hospitals with and without open door policies: a 15 year, observational study”

Christian G. Huber, Andres R. Schneeberger, Eva Kowlinowski, Daniela Frohlich, Stephanie von Felten, Marc Walter, Martin Zinkler, Karl Beine, Andreas Heinz, Stefan Borgwardt, Undine E. Lang

The Lancet Psychiatry, published online July 28, 2016


Inpatient suicide and absconding of inpatients at risk of self-endangering behaviour are important challenges for all medical disciplines. Patients at risk are often admitted to closed psychiatric wards. Increasingly, more modern inpatient wards in the UK are being kept permanently locked, despite this seeming contrary to the UK’s current Mental Health Act Code of Practice. Furthermore, in a census investigation in five European countries, only a small correlation existed between involuntary commitment and treatment on a locked ward. In Austria and Hungary, committed patients were more frequently treated on open wards than on locked wards, whereas in Slovakia, voluntarily admitted patients were more often treated on closed wards than on open wards. In line with these findings, locking policies seem to be mainly based on local tradition and are highly variable between countries, hospitals, and wards.

The main rationale for locking psychiatric wards is safety; i.e., the prevention of suicides and of absconding with possible self-harm or harm to others after the patients have left the ward. To our knowledge, no study has examined whether the presence of locked wards actually leads to a reduction in suicide and absconding compared with hospitals with an open-door policy.

Conversely, the therapeutic atmosphere on locked wards is known to be restrictive and inferior to that on open wards, which could potentially interfere with therapy outcomes and lead to a greater motivation to abscond. Results from a 2015 meta-analysis showed high suicide rates in psychiatric inpatient settings; the suicide rate was 147 deaths (95% CI 138–156) per 100 000 inpatient years, with a tendency for suicides to occur soon after admission.

To address the safety of locking psychiatric units, we have studied the occurrence of suicide, suicide attempts, and absconding in 21 German psychiatric hospitals, which have either open-door or locked-door policies.

Christian G. Huber

While Huber and his colleagues completed this study at a university psychiatric clinic in Switzerland, it is derived from a German database of 21 hospitals, 16 of which had at least one locked ward for the study period at the beginning of 1998 to the end of 2012, while four hospitals had no locked ward throughout that study period. The resulting dataset included 177,295 patients and 349,574 hospitalizations. Of note, all the participating hospitals had to provide care to all individuals from their catchment area, which means they could not cherry-pick or refuse patients based on risk.

The primary outcome measure was completed suicide, and the secondary outcome measures were suicide attempts during treatment, elopement with return, and elopement without return – all the outcomes locked doors are intended to prevent.

The 271,128 patients on locked wards were compared with the 78,446 patients on unlocked wards in terms of age, gender, housing, occupation, diagnosis, comorbidity, self-harm prior to admission, and legal admission status. While some differences emerged that reached statistical significance given the large sample size in a non-randomized study, they strike me as clinically of low significance (e.g., a substance use comorbidity of 36.9% in the locked ward group was statistically significantly different than 36.2% in the unlocked ward group). What is more striking on examining the table of variables is how similar the two groups appear.

To deal with the reality that this was not – and never could be – a randomized clinical trial, the investigators developed a propensity score – a measure of the probability, based on clinical profile, that the patient would have been admitted to an unlocked rather than a locked ward. They then used this and other statistical methods “to achieve the strongest causal inference possible without an experimental design”.

What did they find?

Initially, the results make sense. People admitted to unlocked wards were less likely to have contemplated or attempted suicide beforehand, although self-harm behaviour rates were the same, and were more likely to have been voluntarily admitted. They were more likely to receive medication and to stay longer.

But rates of self-harm behaviours, suicide attempts, and completed suicides during treatment were the same in unlocked and locked wards. Elopement with return occurred more frequently in unlocked wards (1.5%) than locked wards (1.2%), while elopement without return occurred less frequently in unlocked wards (0.6%) than locked wards (0.7%). Further, using odds ratios for hospital type as a predictor of the primary and secondary outcomes, the presence or absence of a locked door was not a predictor for any of the outcomes of the study design.

What does this study tell us?

1. You need big data sets to study the frequency of relatively rare events.

2. You need this type of study design because it is unlikely there will be ethics approval for a randomized controlled study of locked versus unlocked wards.

3. We are reminded that some of things we do because they make sense to us in the moment or are rooted in tradition may not actually make the difference we hope for.

But here are some challenges that are not addressed:

1. How are medicolegal risks managed for the clinicians and the institution? Inpatient elopements and suicides are devastating for families, regular fodder for the media and can also be the subject of court proceedings and inquests. Passing muster with the editorial board of the Lancet is not the same as facing a coroner, a panel of judges, or the editorial board of a newspaper.

2. How are unlocked wards managed, given the current severity and acuity of illness, the high threshold for admission, and the responsibilities for care and custody when the Mental Health Act is invoked to abrogate the freedom of movement?

There are no easy answers. When I ran an inpatient unit, and the door was by default locked, I tried to change the default by asking at team rounds each morning, with regard to the current inpatients, why we needed the door locked. There were many days when, by consensus of the team, the door was unlocked. Not every day – but it changed (for a while) the default assumption.

In a thoughtful editorial on this research in the same issue, senior British psychiatrist Tom Burns addressed another aspect of this issue:

“Locked Doors or Therapeutic Relationships?”

Tom Burns

The Lancet Psychiatry, published online July 28, 2016


On June 1, 1972 I started my first job in psychiatry. Dingleton Hospital in the Scottish Borders was renowned for its family approach and outreach (all patients were assessed and treated in their own homes). But its greater claim to fame was as the first unselective catchment area psychiatric hospital in the UK that had permanently opened all of its doors. It did this in 1948, long before the introduction of antipsychotics, and they remained open until the hospital closed in 2001.

Dingleton was not some isolated Alcatraz, situated on a bleak moor or surrounded by impenetrable marshes. What it had was a culture of strong, confident nurses who prided themselves on really knowing their patients and fully engaging with them, no matter how unwell they were. There was no separate occupational therapy service nor was there a locked nursing station. Relationships were strengthened by shared engagement in day-to-day tasks, such as cooking meals on the ward (more settled patients queued and ate the same food in the same canteen as the staff —the only difference was that they did not have to pay).

How utterly different this is to the psychiatric units of today.

Tom Burns

He recognizes this was an earlier time, with different exigencies than now. But perhaps he has identified something – by no means specific to the discipline of nursing – that has been lost in the wave of advances and replaced by a door that does not open without authorization.

This week’s selection, a companion piece to the previous week’s, is not intended to invalidate clinical tradition and fling all doors open; rather, its purpose is to trigger reflection on the gap that can exist between good intention and desired outcome. A wag once defined clinical experience as making the same mistakes year after year with increasing levels of confidence (O’Donnell M. A Sceptic’s Medical Dictionary. London: BMJ Books, 1997). It needs to be balanced by careful research of high standard that challenges our assumptions.

Reading of the Week. Every week we pick articles and papers from the world of Psychiatry.