Introduction Suicide is the 10th leading cause of death in the USA. Inpatient suicide is the fourth most common sentinel event reported to the Joint Commission. This study reviewed root cause analysis (RCA) reports of suicide events by hospital unit to provide suicide prevention recommendations for each area.
Methods This is a retrospective analysis of reported suicide deaths and attempts in the US Veterans Health Administration (VHA) hospitals. We searched the VHA National Center for Patient Safety RCA database for suicide deaths and attempts on inpatient units, outpatient clinics and hospital grounds, between December 1999 and December 2018.
Results We found 847 RCA reports of suicide attempts (n=758) and deaths (n=89) in VHA hospitals, hanging accounted for 71% of deaths on mental health units and 50% of deaths on medical units. Overdose accounted for 55% of deaths and 68% of attempts in residential units and the only method resulting in death in emergency departments. In VHA community living centres, hanging, overdose and asphyxiation accounted for 64% of deaths. Gunshot accounted for 59% of deaths on hospital grounds and 100% of deaths in clinic areas. All inpatient locations cited issues in assessment and treatment of suicidal patients and environmental risk evaluation.
Conclusions Inpatient mental health and medical units should remove anchor points for hanging where possible. On residential units and emergency departments, assessing suicide risk, conducting thorough contraband searches and maintaining observation of suicidal patients is critical. In community living centres, suicidal patients should be under supervision in an environment free of anchor points, medications and means of asphyxiation. Suicide prevention on hospital grounds and outpatient clinics can be achieved through the control of firearms.
- patient safety
- mental health
- incident reporting
- root cause analysis
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. All relevant data to the study are included in the study. No other data are available.
Statistics from Altmetric.com
Suicide remains the 10th leading cause of death for all age groups in the USA, with 47 000 deaths reported in 2017, accounting for 2.3% of all deaths.1 The Joint Commission, an independent not-for-profit organisation that accredits >22 000 healthcare organisations and programmes in the USA,2 identifies inpatient suicide death as the fourth most common sentinel event type reported in US healthcare organisations, with 75% of suicide deaths occurring on inpatient mental health units.3–5 The incidence of suicide deaths on inpatient mental health units is reported to be between 31.0 and 51.7 per year in the USA, a rate of approximately 3.2 deaths per 100 000 admissions.6 In the Veterans Heath Administration (VHA), the largest integrated healthcare system in the USA, providing comprehensive healthcare services to over 9 million veterans at 170 VHA medical centres; the reported rate of suicide is 0.74 deaths per 100 000 admissions on psychiatric units and 0.6 deaths per million admissions on medical units.7–9
Hanging is the most common method of suicide reported on inpatient mental health units, accounting for >70% of suicide deaths. The most commonly reported ligature points were doors, door handles and door hinges.6 10 Suicide events on inpatient mental health units are well studied, but less is known about other areas of the hospital. In previous studies looking at suicide in US VHA hospitals, we found that the primary methods were cutting and overdose on medical units,9 hanging and cutting in emergency departments,11 cutting and overdose in community living centres12 and gunshot, overdose and cutting on hospital grounds and outpatient clinics.13 Inoue et al found that while hanging was the most common method for suicide on both mental health and medical units, methods such as jumping from heights and cutting were more common on medical units than mental health units.14
Suicide risk and prevention on mental health units has been a focus for many years. Recently, there has been increased focus on suicide risk in other areas of the hospital. In 2019, The Joint Commission produced guidelines outlining steps for suicide prevention for all areas of the hospital.15 The guidance focuses on reducing environmental suicide hazards on all hospital units and outpatient clinics. Since hanging is the most common means of suicide on mental health units, it tends to be the focus of suicide prevention efforts in other areas. However, it is not clear if strategies for suicide prevention on mental health units are effective or apply to other areas of the hospital. While other areas of the hospital have been studied by our team individually, we have not combined and updated this information for a clear understanding of the best suicide prevention strategies by unit.7 8 10–12 The National Action Alliance for Suicide Prevention (an alliance of 250 national partners from the public and private sectors in the USA focused on suicide prevention) made recommendations for treating people with suicide risk in primary care, emergency departments, outpatient behavioural health clinics and mental health units that include screening, safety planning, means reduction and follow-up mental healthcare.16 They recommend removing lethal means in the environment in the emergency department and outpatient clinics, when feasible.16 These earlier studies do not provide specific environmental safety guidance for suicide prevention that tailors the recommendations based on existing evidence of the most common methods of suicide or prevalence rate of suicide in each area. In this study, we evaluate the methods of suicide attempts and deaths and the root causes of these events for multiple areas of the hospital and provide tailored suicide prevention recommendations.
This is a retrospective analysis of suicide deaths and attempts on VHA campuses and inpatient units. We searched the National Center for Patient Safety root cause analysis (RCA) system for suicide deaths and attempts between 1 December 1999 (when the RCA reporting database began) and 31 December 2018. All data were deidentified.
The National Center for Patient Safety root cause analysis adverse event reporting system
All 170 VHA medical centres have inpatient units, outpatient clinics and hospital campuses. Patient Safety Managers at VHA medical centres investigate adverse events such as suicide death and attempts.17 When adverse medical event report is received it is reviewed and rated for harm to the patient (on a 4-point scale from minor to catastrophic) and for the probability of the event occurring again (on a 4-point scale from remote to frequent). Harm and probability are combined to get a score from 1 to 3 called the Safety Assessment Code.17 18 A rating of 1 represents the lowest level of priority for RCA investigation, while 3 represents the highest level of priority. All events with a safety assessment code of 3 are investigated using the RCA process. The patient safety manager pulls together a team of subject matter experts, who were not involved in the event, to carry out the RCA. The team interviews staff, conducts chart reviews and delves deeply into the event to understand what, how and why it happened and what changes can be made to the system to prevent similar future events.
The RCA report includes the event location, including inpatient mental health units, emergency departments, medical and surgical units, community living centres, residential units, outpatient clinics and hospital grounds (defined as any area on the campus other inpatient or outpatient areas, including parking areas). Residential units in the VHA provide residential rehabilitative and clinical care to veterans who have a wide range of care needs (medical, psychiatric, rehabilitation, substance use, homelessness, vocational, educational or social). The residential programmes are located on VHA campuses and provide care in home-like settings where patients come and go, generally keep their medication and are responsible for attending treatment programming. Community living centres, previously referred to as nursing home care units, provide short-term rehabilitation and specialised services as well as long-term care services, and are located on or near VHA campuses.
The patient safety manager reports the event type, including suicide attempts and deaths. Suicide attempts include physical attempts at self-harm and do not include thoughts of suicide. A suicide death is a self-inflicted death and with belief there was intent to die. Suicide attempts and deaths may be reported by staff, patient or family. RCA reports include a narrative describing what happened before and during the event and the underlying causes as determined by the RCA team.18 RCA investigations focus on system issues associated with the event and provide limited specific information about the patients or clinicians involved. The RCA process is described in further detail in other studies .17–19 RCA reports are reported to a central database managed by the National Center for Patient Safety.
Data processing and analysis
We conducted a search for RCA reports involving suicide deaths or attempts between 1 December 1999 and 31 December 2018 that occurred on an inpatient unit, emergency department, outpatient clinics or on hospital grounds. The search was completed by searching for the reported event location, as coded in the RCA, and use of natural language processing software (PolyAnalyst, Megaputer, Bloomington, Indiana, USA) to identify the terms related to suicide or suicide attempt anywhere in the RCA text. Once the RCAs were identified one author (PDM) reviewed each report for inclusion or exclusion. Reports of suicide attempt or death as earlier defined that occurred on a VHA campus (inpatient, outpatient clinics, grounds) were included. A validated codebook used in previous studies was used for coding the data.13 19 The RCA reports contained clear information about the location, type of event, method of suicide and final root causes. The location, event type and method of suicide were coded for all 19 years of RCA data and information on the root causes was coded for the years 2017–2018 by one author (PDM). We have reviewed the root causes of suicide events in these locations in previous work and wanted to discern the most recent causes of suicide event in VHA.8–13 RCA reports identifying suicide by hanging were further coded for anchor point, and lanyard type used.
Admission data extraction
We were able to obtain valid and reliable inpatient admission data by unit type from 2011 to 2018. These data were collected from the VHA Support Service Center, a large data repository of clinical and administrative data. The rates of death by suicide were calculated based on the number of suicide deaths by inpatient unit type, over the total number of admissions nationally by inpatient unit type during the time period. We were able to obtain the total number of outpatient clinic visits between 2011 and 2018. The rates of suicide deaths in outpatient clinics was calculated as the total number of suicide deaths in outpatient clinics over the total number of outpatient visits during the same time period. Rates were not calculated for events that took place on VHA grounds (outside of units or clinics).
Our initial search of the RCA database yielded 21 075 RCA reports. We included reports about suicide attempts or deaths (3425 reports), then further screened reports and included reports that occurred on inpatient units, emergency departments, outpatient clinics or on the hospital grounds and excluded those that did not meet these criteria. Our screening resulted in 847 RCA reports of suicide attempts and deaths on VHA campuses and inpatient units, including 89 deaths and 758 attempts (table 1). Table 2 displays the number of deaths per year by unit type or clinic area and the number of admissions to the unit or visits to outpatient clinics for the years 2011 through 2018 (the years for which we were able to obtain admission data). The rate of death by suicide for these areas is at the bottom of each column.
Characteristics of event types by location
Hanging accounted for 70.9% of the deaths on mental health units, 50.0% of the deaths on medical units, 36.4% of the deaths on residential units and 27.3% of the deaths in community living centres. Hanging accounted for 38.1% of the attempts on mental health units, 25.9% of the attempts on medical units, 19.6% of the attempts in emergency departments and 22.9% of the attempts in community living centres. Cutting with a sharp object accounted for high percentages of attempts on mental health units (24.6), medical units (22.4%), emergency departments (22.4%) and community living centres (39.6%) but caused no deaths. Strangulation accounted for a large percentage of attempts on mental health units (20.9%), medical units (15.3%) and emergency departments (29.0%), but accounted for very few deaths. Overdose accounted for 54.5% of the deaths in residential units and all the deaths in emergency departments. There were high percentages of attempts by overdose in emergency departments (21.5%), medical units (18.8%) and residential units (68.1%). Gunshot accounted for 59.1% of the deaths on hospital grounds (outside of inpatient or clinic area) and all the deaths in outpatient clinics. The number of deaths by suicide on hospital grounds accounted for 43.9% of all the on-campus deaths by suicide between 2011 and 2018 (18 of 41 total suicide deaths). The rate of death by suicide varied from a low of 0.28 per 10 million visits in our outpatient clinics to a high of 2.8 per 100 000 admissions in our residential units.
Root causes for suicide attempts and deaths by unit
Root causes of suicide attempts and deaths by unit for 2017 and 2018 are displayed in table 3. On mental health units the primary root causes were problems with treatment and assessment of suicidal patients, environmental hazards, staff knowledge and communication of risk. In emergency departments, the primary root causes included problems with observing suicidal patients, poor communication of risk among staff, environmental hazards, need for improved assessment, knowledge of suicide risk and problems with contraband checks. On medical units, root causes included problems with treatment of suicidal patients, environmental risk factors and problems with contraband checks. Root causes in community living centres included problems with treatment and assessment of suicidal patients, and poor communication of risk. On residential units there were issues with assessment of suicide risk, poor communication of risk among staff and service to service as well as a need for improved treatment for special populations and suicidal patients. On hospital grounds root causes included poor communication of risk and problems with assessment and treatment of suicidal patients. In outpatient clinics, root causes included inadequate controls of access to clinic buildings and the need for staff education on suicide risk factors.
In this retrospective analysis of suicide deaths and attempts reported to the US VHA National Center for Patient Safety RCA database, we found that the method and rate of death by suicide varies by location. We were not able to find literature on the rates of death by suicide on medical units, emergency departments or other areas of the hospital; however, our current rate of death by suicide on mental health units in the VHA of 0.31 per 100 000 admissions is lower than national estimates of 3.2 deaths per 100 000 admissions.6 This lower rate on mental health units in VHA is likely due to the use of the Mental Health Environment of Care Checklist, which has significantly reduced the rate of suicide on these unit since its deployment in VHA in 2008.6 In addition, the incident of suicide in nursing homes in the USA was estimated to be 14.16 per 100 000,20 while the rate of suicide in VHA community living centres between 2011 and 2018 was 2.11 per 100 000 admissions. This lower rate in community living centres in VHA may be due to the strong priority of suicide reduction in VHA since 2008.
We found that on mental health units and to a lesser extent, medical units, hanging was the most frequently used method for suicide attempts and deaths. Hanging accounted for 70.9% and 50.0% of the deaths by suicide on mental health and medical units, respectively and 36.6% of the deaths on residential units. While the most lethal method of suicide in community living centres was hanging, a range of methods were used on these units including cutting, overdose, gunshot, jumping, self-stabbing. These results mirrored Inoue et al, 14 who found that hanging was the most common method for suicide on both mental health and medical units. In an earlier study of medical units in VHA, we found that hanging was less common on medical units than on mental health units,9 but since that time we have seen more suicide attempts by hanging on our medical units.
On VHA residential units and in emergency departments, overdose was the cause of most of the deaths by suicide (54.5% and 100.0%, respectively), while on hospital grounds and outpatient clinics most deaths were caused by gunshot (59.1% and 100.0%, respectively). Gunshot deaths in clinic areas were extremely rare and the only method of suicide death identified in clinic areas (0.0028 per 100 000 visits). Gunshot suicide deaths on hospital grounds, however, represent 31.7% of all VHA on-campus suicide deaths. Our data reveal that between 2011 and 2018 the number of suicide deaths on inpatient units and outpatient clinics was between 2 and 6, while on hospital grounds there were 18 deaths, 13 of those by gunshot. Jumping was seen more on hospital grounds than on inpatient units, however, we found in this study and a previous study, that most jumping events did not resulting in death.21 We hypothesise that in the most controlled and observed environments, such as mental health and medical units, the opportunities and means for suicide are rarer, so methods using available means such as hanging and cutting are most common. In less well-observed units such as community living centres, residential units and emergency departments, medications, illicit drugs and sharp objects are more easily obtained so overdose, cutting and asphyxiation are more common. In more uncontrolled environments such as hospital grounds and clinic areas, the method of suicide mirrors the methods used in the USA more generally, namely firearms.22
Prevention strategies by area
Mental health units: to prevent suicide on mental health units it is important to eliminate all anchor points that can be used for hanging and remove other lethal means, such as drugs, plastic bags (that can be used for asphyxiation) and sharps off the unit. We have seen a dramatic decrease in the rate of death by suicide on VHA mental health units since implementing our Mental Health Environment of Care Checklist in 2008.7 The checklist helps staff find and eliminate anchor points, sharp edges, plastic or elastic that could be used for asphyxiation, and modify all lights and outlets on the unit so that they cannot be used for self-harm. In addition, we have found some evidence to support the use of over-the-door alarms to prevent suicide by hanging on the door in inpatient mental health units.23 Facilities should continue to make environmental safety-rounds on mental health units an integral part of their process of care and ensure staff are providing evidenced-based assessment and treatment for suicidal patients.24
Emergency departments: in the emergency department overdoses were the only method of suicide resulting in death. The overdoses involved drugs or medication brought into the emergency department by the patient and used in the bathroom when not under supervision.11 At point of care an evidenced-based assessment of suicide risk, such as the Columbia suicide risk assessment should be conducted to determine patient’s suicidality and recommended observation or interventions while in the emergency department.25 Poor observation of suicidal patients was the number one root cause for these events in the emergency department. These patients should be kept under one-to-one observation, including in the bathroom. In addition, equipment or hazardous items that could be used for self-harm should be removed when feasible. In a recent study of adverse events in emergency departments in VHA hospitals, Gill et al 24 found that 60% of the suicide attempts involved using material that the patient had brought into the emergency department for self-harm. While there is no evidence in the literature of the effectiveness of contraband searches at preventing suicide, it does constitute a form of ‘means reduction’ to prevent suicide which does have support in the literature.26 27 There were a significant number of suicide attempts by hanging, cutting and strangulation. We agree with the National Action Alliance for Suicide Prevention’s recommendations to remove all lethal means from hospital and emergency department rooms16 including sharps, tubing, medications and anchor points for hanging where possible.
Medical units: on medical units it can be more difficult to eliminate all lethal means because medical equipment is needed to provide care (such as intravenous tubing). In such situations, it is critical to provide one-to-one observation for suicidal patients by a staff member who is aware of the patient’s suicide status and trained to observe the patient at all times (including in the bathroom), and to remove any items that can be used for self-harm. While the literature on one-to-one observation is mixed and there are reports of patients killing themselves while on one-to-one observation,28 it is difficult to imagine allowing an actively suicidal patient to remain on a medical unit without continuous observation; and Janofsky has outlined a method of using failure mode and effects analysis to improve the observation process of suicidal patients.28 As the Joint Commission recommends,10 it is important to conduct a thorough environmental assessment to understand the potential hazards and remove risks when possible. Providing mental health treatment for suicidal patients on medical units is an important part of suicide prevention, as problems with treatment for suicidal patients was the most common root cause; and there is some evidence to suggest that treatment for suicidal behaviours can reduce the incident of suicide on inpatient units.29 Overdoses were reported on medical units, so a careful inspection for hazardous items that may be brought in by the patient or family is important. Additional actions include ensuring that suicidal patients do not have access to guns (eg, in their car) or access to high places for jumping. It is important that all staff interacting with patients be aware of their suicide risk status so that items are not left in the room or provided to the patient that can be used for self-harm (eg, eating utensils/knives, bleach solution).10
Community living centres: in these nursing home like settings the method of suicide varied and included hanging, cutting strangulation, overdose and asphyxiation. It is important to eliminate anchor points for hanging where possible and to ensure suicidal patients do not have access to medications for overdose, plastic bags for asphyxiation or sharps for cutting. Patients at risk for suicide should be placed on one-to-one observation by a trained observer until the patient can be moved to a more secure location.28 Looking at our RCA data it appears that better assessment and treatment of risk factors for suicide is needed as is improved communication about suicide risk. While there is not strong evidence in the literature that supports specific interventions to reduce suicide in nursing homes,30 the US Department of Health and Human Services Substance Abuse and Mental Health Services Administration published a tool kit for well-being, strategies for ensuring that staff properly identify and effectively treat residents at risk of suicide, and procedures for appropriate responses to suicide.31
Residential units: in our residential units, we see a different pattern of suicide attempts and deaths. In residential units, we found that overdoses are the most common method of suicide attempts and deaths, so it is critical to keep contraband medication and drugs out of the unit. The environment of care in these units is home-like and may contain many suicidal hazards. It is critical to evaluate all patients coming into the programme for suicide risk and only allow those patients at lower risk for suicide to enter the programmes. There is a need for improved assessment of suicidal patients and improved communication about suicide risk for patients entering the programmes.
Hospital grounds: on hospital grounds, we see suicide attempts from a variety of means including ingestion of chemicals (27.7%), jumping (27.7%) and cutting (19.1%). Gunshot was the cause of 59.1% of suicide deaths; however, we did see suicide deaths from hanging and jumping. While is it difficult to eliminate all means of hanging on hospital grounds, it is possible to make parking garages and high places safer by limiting access, installing barriers and providing signage with emergency contacts.20
Outpatient clinic areas: in outpatient clinic areas, suicide attempts and deaths are extremely rare events. Suicide attempts use a variety of methods, but all deaths were from gunshot. Eliminating the ability of a veteran to bring a gun into a clinic should be considered. The VHA should continue to raise awareness in the veteran population of how to recognise and report unusual behaviour and continue to train VHA police on addressing suicidal veterans. Firearms are the primary means of suicide deaths on grounds and in clinics and in approximately 70.7% of all outpatient suicide deaths among US veterans.32 In response, the VHA has made gun-locks available to all patients and is investigating strategies for integrating firearm safety into the overall suicide prevention programme33 including lethal-means safety education, firearm safety interventions targeted to specific patient populations, improved understanding of firearm-related risk perception, firearm-related data collection, and continued partnership with firearm owners and advocacy organisations.
Not all environmental hazards can be eliminated on all units, therefore contraband searches should be conducted, and patients should be screened for suicide risk to recognise which patients require 1:1 supervision and mental health treatment.
This study has several limitations. First, our adverse event reporting system relies on self-reporting from staff members into the National Center for Patient Safety reporting system, so it is likely that some suicide events have not been captured. It is unlikely however, that inpatient deaths from suicide were not reported. There is evidence to suggest that some overdose suicide deaths could have been misidentified as ‘accidental overdoses’ rather than suicide and so left out of our dataset.30 This may be particularly misidentified on residential units as patients are not under continuous supervision and it is difficult to determine if an overdose was intentional or not. The focus of RCA reports is on the system rather than individual patient or clinician characteristics, so data about the patients involved in these events are limited. Finally, all data were collected in the VHA system, all patients are US veterans and mostly males. The VHA system may be different from other hospitals in the USA and different from other hospital system in other countries, consequently generalisability may be limited.
Review of suicide attempts and deaths by unit type gives more specific information regarding suicide prevention effort in those areas.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. All relevant data to the study are included in the study. No other data are available.
This project has been approved by the Institutional Review Board, VAMC, White River Junction, Vermont, USA.
Contributors Contribution of data coding, analysis, writing, revision.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.