Suicide Risk After Hospital Discharge

Inpatient psychiatrists in the United States deal with three problems: suicide risk, aggression risk, and the inability care for oneself or impaired functional capacity. Those have always been the primary reasons why people are admitted to psychiatric hospitals. Over the past 30 years there has been a problem with business intruding on these medical indications. Businesses and or more specifically managed care companies eventually adopted a single codeword “dangerousness”. Dangerousness was supposed to encompass all three of the dimensions but eventually it developed a life of its own. As an example, I have been asked by insurance company reviewers “Where’s the dangerousness?” whenever they tried to throw one of my patients out of the hospital and onto the street. Science or medical principles were not involved, just the economics of being able to use a word to make money.

The reality of inpatient work is that the people there are very high-risk for suicide, aggression, and premature death from multiple causes. There was a study done in Germany about 20 years ago where they looked at all-cause mortality of people discharged from psychiatric units five years later and the number was very high. Post discharge suicide rates are much higher than suicide rates in the general population. These high numbers are expected because patients and inpatient units are selected for these traits that predispose to higher mortality and morbidity. The trends have been complicated by much shorter lengths of stay and bed limitations that means patients with severe mental illness may be refused admission even if they clearly need it. As example, since leaving the inpatient setting about 10 years ago, I have attempted to refer severely ill patients to psychiatric hospitals and they were turned away at the emergency department. In some cases they were turned away without being seen by a physician. There are probably a handful of psychiatrists in the United States who know the type of problem that needs to be treated that inpatient units and I am one of them.

That situation makes a recent study on risk of suicide after discharge from inpatient psychiatric care and interesting one. The study was done in Sweden. It encompassed the years 1973 to 2009. During that time there were 2,883,088 admissions and presumed discharges. 690,937 patients were discharged more than once. Most the discharges were men (57.6%). There were no explicit indications for admission. Patients were followed up and it was determined that there were 3695 suicides within 30 days of discharge. The authors calculated a suicide rate of 181/10,000.  The discharge diagnosis most associated with suicide with depression. They gave some rough estimates of the prevalence of disorders in this population: 34% alcohol use disorder, 15.5% had mood disorders and (bipolar disorder or depression) and 9.9% had schizophrenia. 

Looking at the results according to diagnosis depression was followed by reaction to crisis or what is probably called an adjustment disorder in the US, but any specific psychiatric disorder and elevated hazard ratio for suicide within the first 30 days of discharge relative to the diagnosis of alcohol use disorder. 

Suicide risk was also examined relative to recent suicidal behavior. The suicidal behavior was considered to be any deliberate self-harm less than 30 days prior to admission. That was noted to have a hazard ratio of 4.75. The diagnoses were re-examined in the context of deliberate self-harm prior to admission and the risks were significantly higher in schizophrenia (HR = 8.94) and other nonorganic psychosis (HR =6.82).  Interested readers are referred to the full text which is available free online for the specific details including hazard ratios and confidence intervals for those hazard ratios.

The main findings of this study include the association of relatively high risk at discharge for most diagnoses and much higher risk if a specific diagnostic category was associated with a self-harm event 30 days prior  to admission.  This confirms clinical risk assessments that are typically done and also the fact that this is a high-risk population.

The authors do state that they regarded principal diagnosis at discharge to be the best available information on the reason for admission. I contacted the corresponding author about this and he did confirm that the reasons for admission in Sweden are very similar to what they are in the United States and that is suicide risk, aggression risk, and ability to care for oneself - but those specific metrics were not listed in the paper. 

The authors speculate on why the suicide risk is high. They describe the slow recovery from depression and the clearing of psychomotor retardation prior to the resolution of depressogenic thinking.  In the US, psychiatrists are generally taught that psychomotor retardation may reduce the risk of acting on suicidal thoughts so that during treatment there may be a point where activation may put the person at risk for acting on unresolved suicidal thoughts. Given the characteristics of suicide particularly the impulsivity associated with it this progression of events has never been proven and remains highly speculative. The authors also had the interesting observation that crisis events or negative life events are expected to offer good prognosis but this study showed men with the diagnosis of reaction to a crisis were at high risk after discharge irrespective of whether there was any recent suicidal behavior.

The authors review the strengths and limitations of the study. The main strength is that it is a large-scale study with a significant number of suicides. They also point out how their study has similarities with other studies of suicide in hospitalized patients. On the limitation side most of the limitations had to do with a lack of granularity in the data. I pointed out the lack of specific admission indication in addition to diagnosis. In current databases there may be metrics having to do with the level of suicide or aggression risk. There are some large-scale studies being done on an outpatient basis looking at those metrics as well as supportive interventions based on risk scores.

The biological side was not discussed in this study even though the database used probably contained all of the admission and discharge medications. It would be interesting to know if certain pharmacological interventions were more or less associated with suicides after discharge. It would also be very useful to know if there were any protective factors from inpatient treatment that could be discerned from the data both from the standpoint of psychosocial interventions and biological interventions like electroconvulsive therapy, transcranial magnetic stimulation, or treatment with ketamine.

I have included a reference to another study of post discharge suicide rates done in a Medicare sample and with a slightly different methodology (2).  In this case the researchers looked at a population of 770,643 patients with mental disorders compared with a 1,090,551 patients in a cohort with no mental disorders and compared suicide rates in the first 90 days after discharge. They found suicide rates that were 10-20 times higher in the mental disorder cohort.  The non-mental disorder cohort had a rate that was lower than the baseline suicide rate in the US at the time (11.6 versus 14.2 per 100,00 person years).  Like the Swedish study rates were the highest for mood disorders. Comparing both of these studies would be an interesting seminar for residents or journal club for staff psychiatrists interested in different epidemiological approaches to the same clinical problem.  I have included two references by Chittaranjan Andrade, MD and encourage the use of his series on statistics and epidemiological concepts in the Journal of Clinical Psychiatry.

From a clinical standpoint, the take-home message for clinicians is to make sure that deliberate self-harm prior to admission and the diagnosis are carefully explored. It is fairly common practice to consider adjustment disorders to be low risk in terms of brief hospital stays and discharge plans that do not include intensive outpatient treatment. This study suggests that at least some of those patients need more intensive intervention but there is no guidance on how to identify that group. The other high-risk groups of patients with psychosis that had deliberate self-harm prior to admission and any trained inpatient psychiatrist should admit those patients and treat them until there is clinical improvement that results in decreased risk.

My only concern about the current patient flow through emergency departments and onto psychiatric inpatient units is that many of these patients never get admitted and if they do they are discharged in a short period of time with the same symptoms that they presented with.  The inpatient environments in the US are also deteriorated to the point that they resemble correctional settings and patients want to leave as soon as possible.

These are not good ways to address the issue of post hospitalization suicide risk in a high risk population.

George Dawson, MD, DFAPA

Reference:

1:  Haglund A, Lysell H, Larsson H, Lichtenstein P, Runeson B. Suicide Immediately After Discharge From Psychiatric Inpatient Care: A Cohort Study of Nearly 2.9 Million Discharges. J Clin Psychiatry. 2019 Feb 12;80(2). pii: 18m12172. doi: 10.4088/JCP.18m12172. PubMed PMID: 30758922. (full text)

2: Olfson M, Wall M, Wang S, Crystal S, Liu SM, Gerhard T, Blanco C. Short-termSuicide Risk After Psychiatric Hospital Discharge. JAMA Psychiatry. 2016 Nov 1;73(11):1119-1126. doi: 10.1001/jamapsychiatry.2016.2035. PubMed PMID: 27654151. (full text)

3: Andrade C. Why odds ratios can be tricky statistics: the case of finasteride, dutasteride, and sexual dysfunction. J Clin Psychiatry.2018;79(6): 18f12641. Link

4: Andrade, Chittaranjan. Drug interactions in the treatment of depression in patients with ischemic heart disease. The Journal of Clinical Psychiatry 73.12 (2012): 1475-1477.