Objectives The overall aim of this study was to aggregate the conclusions of all investigations conducted after suicides reported to the supervisory authority in Sweden in 2015, and to identify deficiencies in healthcare found in these investigations; the actions proposed to deal with the deficiencies; the level of the organisational hierarchy (micro–meso–macro) in which the deficiencies and actions were situated; and outcomes of the supervisory authority’s decisions.
Design and setting This is a retrospective study of all reports from Swedish primary and secondary healthcare after suicide to the regulatory authority in Sweden in 2015.
Results In 55% (n=240) of cases, healthcare providers reported healthcare deficiencies that contributed to suicide; these deficiencies were primarily in ‘suicide risk assessment’ and ‘treatment’. Actions aimed at preventing new suicides were proposed in 80% of cases (n=347). By far, the most frequent actions were ‘education and competence’, present in 52% of cases (n=227) and did not much correspond with identified deficiencies. Sixty-five per cent of the deficiencies and actions were at microlevel, while the remainders were at mesolevel. In 65% (n=284) of cases, the supervisory authority approved the investigation without further requirements.
Conclusions The most common identified deficiencies were related to care in the immediate interface between patient and staff. Actions proposed to prevent new suicides were centred on single educational interventions without distinctive sustainable effects in the organisations and usually did not correspond with the identified deficiencies. Future research should examine if application of a framework based on knowledge of the suicide process, suicide prevention strategies and patient safety would enable more sophisticated investigations that could facilitate progress on suicide prevention.
- health & safety
- quality in health care
- risk management
- adult psychiatry
- suicide & self-harm
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
This is the first national aggregated analysis of the outcomes of investigations following suicides in Sweden.
The categorisation of deficiencies and actions for improvements was done by a single person to improve consistency.
All data were based on the healthcare providers’ reports of suicide to the supervisory authority, reports performed in different contexts by different persons with a large spectrum of disparities in experiences resulting in variegated quality.
Close to 800 000 people die by suicide worldwide every year.1 Studies show that ~9 out of 10 individuals who die by suicide have a psychiatric disorder at the time of death, and a large proportion of suicide deaths occur among individuals receiving ongoing psychiatric care or who have contact with other healthcare providers.2–5 There is some evidence that suicide prevention strategies diminish suicide rates6 7; however, despite intensified efforts to improve the healthcare safety for suicidal patients, the suicide rate has remained essentially the same in Sweden, at ~1200 deaths every year.8 In recent decades, awareness and knowledge of patient safety has increased. Many countries have established an incident reporting system, meaning that serious adverse events are to be investigated and reported to a supervisory authority. To better understand if failures in any area of the healthcare system have contributed to suicide, all suicides that occurred while a victim was receiving healthcare or within 4 weeks after healthcare contact were required to be reported by the healthcare provider to the supervisory authority for healthcare in Sweden in 2006–2017. A review conducted 1 year after this obligation was implemented showed that the supervisory authority criticised healthcare providers for healthcare deficiencies in 53% of cases, with the most frequent deficiencies being in routines and risk assessments.9 Since that report, no further national aggregated analysis of the outcomes of the investigations following suicides has been done. To our knowledge, there are neither any international aggregated analyses nor other analysis of this kind published.
Investigations based on root cause analysis (RCA) have become wide-spread tools in healthcare services efforts to understand and prevent adverse events.10 11 The principle of RCA is to identify and rectify underlying system vulnerabilities that allow human errors to cause harm to patients.12 This approach assumes that adverse outcomes can be explained by linear cause-effect chains and have causes that can be found and fixed, and that the actions preceding adverse events differ from those that precede ordinary, successful care.13 The actual value of incident reporting systems and the RCA approach in healthcare is subject to debate.14–18 Single analyses usually provide little learning beyond the involved staff and unit. Rather, aggregation of data from multiple analyses should generate more meaningful action plans for improvement and better facilitate the learning processes in organisations.
Swedish law states that when an adverse event has resulted or could have resulted in severe patient harm, this should be reported to the supervisory authority, the Health and Social Care Inspectorate (HaSCI). The role of HaSCI is to ‘…ensure that reported adverse events have been investigated to a necessary extent, and that appropriate actions have been taken by the healthcare provider to reach a high level of patient safety’.19 The report to the authority is to be preceded by an investigation of the healthcare services provided to the patient before the adverse event, conducted by the healthcare providing organisation. The head of the departments are formally responsible for the investigation and investigators can be any type of healthcare professional. The investigations aim to identify the causes and contributory causes of the incident and to identify improvements that should prevent the same incident from happening again. A distinction is made in investigations between actions performed immediately after an incident and non-immediate actions proposed or taken some time afterwards. The authority then examines the investigation and decides if the healthcare provider has fulfilled their legislated role of investigating the incident and taking actions to ensure patient safety. If there are shortcomings in the investigation, the HaSCI calls for additions or conducts a site visit to inspect the healthcare provider.
The overall aim of this study was to aggregate the conclusions of all investigations conducted after suicides reported to the supervisory authority in Sweden in 2015, and to identify deficiencies in healthcare found in these investigations; the actions proposed to deal with the deficiencies; the level of the organisational hierarchy (micro–meso–macro) in which the deficiencies and actions were situated; and outcomes of the supervisory authority’s decisions.
All suicide cases (n=436) reported to the HaSCI in 2015 were included. Complete incident investigations from healthcare providers with associated patient records and decisions of the supervisory authority were obtained from the supervisory authority. Every individual suicide was given a code number and the patient’s demographic data, contact with all areas of healthcare and received treatment in the 3 months before death were registered. Major diagnoses documented and coded in accordance with the International Statistical Classification of Diseases and Related Health Problems - Tenth Revision (ICD-10) coding system in the records were registered.
Categorisation of data
A coding scheme was used to categorise the causes and contributory causes of the suicide, as well as the immediately performed actions and non-immediate actions reported in the investigations. The coding scheme was based on the general categories of the most widespread method of investigating adverse events in Swedish healthcare, which is based on RCA.20 The categories were as follows: education and competence, communication and information, organisation and management, technics and equipment, and policies and procedures. To make the categorisation more specific, four of the major categories were divided into additional subcategories. Every category was described and exemplified and a category of ‘others’ was added in case none of the other categories was considered appropriate (table 1). Since the providers rarely made a distinction between causes and contributory causes in the investigations, these are reported as deficiencies in this paper. In this study, an action (immediate or non-immediate) was defined as an intervention that aimed to prevent new suicides. Therefore, actions taken to prevent reported suicides (telephone calls, resuscitations) or actions aimed at informing family or staff that a suicide has occurred were not registered as actions in this study. Separate notes were made when a deficiency or action was related to routines and if patient-related factors were reported. In cases where different providers reported the same suicide case, the outcomes of the investigations were grouped. Identical deficiencies or actions reported by different providers regarding the same patient were excluded, thus ensuring that every factor was counted only once. How learning from the investigation was described; inside the department, outside the department, irrelevant or not mentioned, was registered. All data collection and categorisation was conducted by only one researcher, an experienced psychiatrist, to achieve consistency.
A classification of the organisational levels of deficiencies and actions was conducted to better understand where in the organisational system the identified deficiencies and actions were situated. The deficiencies and actions were coded according to a micro–meso–macro perspective.21 Microsystems were defined as the basic building blocks of all healthcare systems formed around the patient and family, such as the inpatient or outpatient care unit. The mesosystem encompassed interactions between different microsystem units, such as cooperation between clinics or healthcare providers. The macrosystem involved the whole system of healthcare, such as legislation, political prioritisations and national policies on healthcare. The highest organisational level for each deficiency, immediate action and non-immediate action for each case was coded.
The decisions of the supervisory authority were coded as follows: ‘immediate approval’, ‘request for one or more additions’ or ‘inspection’.
Summary statistics were calculated for deficiencies, immediate actions, non-immediate actions and decisions of the supervisory authority. Frequencies for each category and organisational hierarchal level in deficiencies, immediate actions and non-immediate actions were analysed per individual and aggregated.
χ2 tests of independence were used to compare the number of deficiencies and non-immediate actions in the same category. We considered a two-sided p value of <0.005 to be statistically significant. Fisher’s exact test was used in cases where 20% of the analysed groups had an expected count of <5. The statistical analyses were performed using IBM SPSS Statistics V.24.
According to the Swedish Act Concerning the Ethical Review of Research Involving Humans (2003:460) and an advisory opinion from the Regional Ethical Review Board (no. 2017/234), this study did not require ethical review as it did not include human participants.
Patient and public involvement
Patients or public were not involved in this study.
In total, 1179 suicides were registered in Sweden in 2015.8 The supervisory authority received 473 reports. In 35 cases, the same suicide was reported by two different healthcare providers, regarding different parts and perspectives of the care process, and for one case, the same suicide was reported by three providers, resulting in 436 unique suicide cases. Characteristics of the cases and healthcare received in the last 3 months before suicide are presented in table 2.
Deficiencies in healthcare before suicide
In 55% (n=240) of suicide cases, the healthcare provider identified deficiencies in the healthcare that were considered to have contributed to the suicide. Among all cases, a total of 952 deficiencies were identified. The number of deficiencies per case ranged from 1 to 21, with a median of 3.
The most frequent deficiencies were in ‘treatment’ and ‘suicide risk assessment’. Examples were inadequate or delayed pharmacological treatment, non-adherence to existing guidelines, inadequacies in doctors’ prescribing, a misleading suicide risk assessment and non-adherence to local guidelines for suicide risk assessment. Deficiencies in ‘external communication’ were the third most frequent. Examples were shortcomings in communication between a somatic and psychiatric clinic and a lack of important information being handed over from one healthcare provider to another. For further details, see tables 3 and 4. In seven cases, identical deficiencies for the same case were reported by different providers, categorised as external communication, treatment, suicide risk assessment and ‘care plan’.
All reported deficiencies were at the microlevel in 65% (n=157) of cases (table 5). An example of a deficiency at the microlevel was inadequacies in doctors’ prescribing or in suicide risk assessment. The remaining 35% (n=83) had at least one deficiency at the mesolevel, such as shortcomings in cooperation between a psychiatric clinic and somatic clinic or inadequacies in communication between hospital and municipality. No deficiencies were considered to be at the macrolevel.
Deficiencies in routines were reported in 20% (n=96) of all cases. These often reflected non-adherence to existing routines. Missing or defective routines were reported in 11% (n=49) of cases. Deficiencies in routines could occur in any category.
In 31% (n=135) of cases, patient-related factors were reported to have contributed to the suicide. Examples were changes in the patient’s private relationships or life conditions, or circumstances the provider considered to be outside the influence of healthcare.
Immediately performed actions
Immediately performed actions were reported in 6% (n=26) of cases. In these, 45 immediate actions were described. The number of immediate actions per case ranged from 1 to 7, with a median of 1. The most frequent immediate actions taken were categorised as ‘human resources’, usually recruitment of physicians (tables 3 and 4). In one case, there was an action at the mesolevel; the remainders were all at the microlevel (table 5).
Non-immediate actions aiming to prevent new suicides were taken or proposed in 80% (n=347) of all cases. In these, a total of 1330 interventions were described. The number of actions per case ranged from 1 to 20, with a median of 3.
The most frequent non-immediate actions were in the category of ‘education and competence not specified’. Examples were case report discussions at staff meetings, lectures about affective disorders and reminding staff about existing local guidelines. The second most frequently reported non-immediate action category was ‘education and competence in suicide risk assessment’. Examples were lectures for staff about suicide risk assessment and reminding staff about existing guidelines for suicide risk assessment. Together, non-immediate actions in either of these two categories were described in 52% (n=227) of all cases, corresponding to 32% of all reported non-immediate actions.
The third most frequent non-immediate action category was changes in ‘work process’. Examples were new checklists and changes in the intern system of reporting adverse events. For further details, see tables 3 and 4. Identical actions regarding the same case were reported by different providers in 12 cases and were in the categories of external communication, education and competence not specified, suicide risk assessment, care plan, work process and education and competence in suicide risk assessment.
The organisational levels of the non-immediate actions were equal to those of the deficiencies; in 65% (n=225) of the cases, all actions were at the microlevel and in 35% (n=120) there was at least one action at the mesolevel (table 5). Examples of actions at the microlevel were case discussions at staff meetings, lectures and new checklists. Examples of actions at the mesolevel were changed procedures for communication or cooperation between different healthcare providers. Only one proposal was at the macrolevel, and this involved the possibility of the prescribing doctor checking what medications a patient received from pharmacies throughout the country.
Learning from the investigations were described to be inside the department in 56% (n=266) of the reports. In only 4% (n=20) of the reports, sharing of the experiences and conclusions outside the own department were described. In all other reports, nothing was mentioned about the learning or considered not being relevant.
Changes in routines were proposed in 35% (n=152) of all cases, and these actions could be in any category.
Decisions of the supervisory authority
In 65% (n=284) of cases, the supervisory authority approved the report from the healthcare provider without further requirements. In 29% (n=126), the supervisory authority called for one or more additions to the investigation before approval. In 6% (n=25), an inspection took place at the healthcare provider before the decision, and in these cases the supervisory authority usually called for additional actions before their decision. Of the 36 cases with more than one investigation, the decisions of the authority differed in 16 cases.
This study describes the aggregate results of healthcare provider investigations made after suicides in Sweden in 2015. In more than half of the studied cases, there were deficiencies in the healthcare provided before suicide that were considered by the providers to be of significance to the death. The majority of the deficiencies were at the micro organisational level, and no deficiency was found at the macrolevel. The most common deficiencies involved care delivered in the immediate interface between patient and staff, which were relatively easy for the investigators to identify. Actions to deal with the deficiencies were substantially more frequent than the number of described deficiencies and were dominated by educational actions. The majority of the actions were at the microlevel, and only one proposed action was at the macrolevel.
The most frequently reported deficiencies were related to treatment. Four out of five patients in this study were prescribed psychotropic drugs, most commonly sleeping pills and antidepressants. Pharmacological treatment of psychiatric disorders is regarded as a central and evidence-based component of the prevention of suicide.7 22 To deliver the right treatment for the patient, correct diagnoses are essential: diagnostic errors are known to be common causes of adverse events in all areas of healthcare.23 24 A majority of the patients in this study had at least one documented psychiatric diagnosis, although less than half had a diagnosis of depression. The deficiencies in ‘diagnosis’ category were lower than would be expected, given the known outcome of suicide, the fact that all cases had contact with healthcare shortly before death, and the fact that a vast majority of suicide deaths involve individuals who meet the diagnostic criteria for depression at time of death.5 Many investigations were performed without the participation of a physician, which could help explain the low number of reported diagnostic errors.
Admission to inpatient care is a common choice of treatment for those at risk of suicide. One-third of the patients in this study were admitted to the hospital in the 3 months before their death; however, only 8% of the suicide deaths involved inpatients, which is notably lower than the 24% found in a review of suicides in Sweden in 2007.9 This decrease could be a result of safer inpatient care; however, it could also reflect a shift of suicides from inpatient care to the postdischarge period, mirroring the reduction in the number of beds in psychiatric care during the last few decades.25 However, investigators in the present study did not reach this conclusion, as the number of hospital beds was reported as contributing to suicides in only 2% of cases. At the same time, it is not clear if this low frequency resulted because investigators considered this to be an issue outside their mandate.
Deficiencies in suicide risk assessment were frequently reported, as exemplified by inadequate performance of risk assessment or insufficient supervision of patients assessed to be at high risk for suicide at psychiatric inpatient units. All cases in this study were in contact with healthcare services during the 3 months before their suicide, and 90% were in contact more than once. Documentation of suicide risk in patients’ records during the last 3 months before suicide was absent in 25% of cases and regarded as low/non-existent in 39%. Suicide is usually the final outcome of a process over time and involves the interaction of several factors. As suicide intentions also fluctuate rapidly, assessments must be repeated to catch suicidal crises.6 The small number of cases in this study where suicide risk was assessed as high might reflect difficulties in assessments. However, it could also indicate success of healthcare in cases when suicide risk was assessed as high and then followed by preventive actions. Further research is needed to confirm this hypothesis.
Substantially, more actions to prevent new suicides were reported compared with the number of identified deficiencies, possibly reflecting insights into the weaknesses of the healthcare system that confer risk to patient safety. The proposed actions centred on educational interventions: these actions were proposed for half of cases and corresponded to one-third of all reported actions. In comparison, deficiencies in ‘education and competence’ were reported in only 10% of cases, indicated that providers aimed to solve deficiencies in different categories with educational actions. Most of the proposed educational actions represented a single case discussion or reminder of a routine in staff meetings, suggesting that the deficiencies were being simplified and quick fixes were being applied. Evidence that educational interventions reduce suicide rates relies on studies of extensive education programme.26–30 In order to reach successful implementation and sustainable behaviour change, considerable work—including long-term multifaceted interventions—is usually needed. Macrae emphasises the importance of active reflection, mindful participation and emotional engagement.31 32 If this kind of reflection is not part of how healthcare providers promote learning, the large amount of single educational actions can create a false sense of security without making the organisation safer. Strong leadership with visible engagement in patient safety at all levels is of high importance in shaping and maintaining safe structures in organisations.32–36 Very few deficiencies regarding management were reported in this study, probably reflecting the investigators’ lack of understanding of this issue rather than an absence of management shortcomings.
Even though missing or defective routines seldom were reported as contributing to suicides, new or changed routines were proposed to prevent new suicides in one-third of the investigations, often in the category of work process. This focus on routines in patient harm investigations has been shown before.9 35 37 Well-functioning work processes and adherence to routines are indisputably of high importance for ensuring safe healthcare. However, the large number of changes without corresponding shortcomings shown in this study might result in insecurity, rather than safety, among staff. This suggests that providers oversimplify the challenges of patient safety at the frontlines of healthcare.
Immediate action was taken in only a few cases, which probably reflects the absence of obvious deficiencies possible to be fixed. Compared with non-immediate actions, a larger share of immediate actions concerned ‘technics and equipment’, usually the removal of ligature points such as hooks and doors.
A majority of identified deficiencies and actions were at the organisational microlevel—they were usually within the care unit where the patient had their last contact with healthcare services. These findings were similar to those of a prior Swedish study.18 The results probably reflect the investigators’ knowledge and understanding of suicide and what they consider can be fixed more than the actual circumstances. The real purpose of investigations of healthcare after adverse events should be to reveal gaps and inadequacies in the healthcare system and to find effective and meaningful actions leading to sustainable improvement of healthcare.38 To succeed in this, we need to develop methods appropriate to current healthcare services and to improve the ability of healthcare organisations to learn from and recall incidents and investigation outcomes.10 31 32 In this study, learning from the investigations were in most cases described to be inside the own department, sharing of the experiences and conclusions outside the own department were described in only a few cases. Past studies have shown that the results and conclusions of investigations are rarely passed down to the organisation and that there is an absence of formalised organisational memory, even though many patient safety activities that arise from the investigations after incidents are based on such memory-making activities.18 39 Vincent suggests the use of a ‘safety analysis of the patient journey’ to identify the series of events and combinations of errors and system vulnerabilities that in combination and gradually unfold over time.32 Analyses over a longer period of time would enable identification of successful recovery from suicidal crises, which is necessary knowledge to progress in work on suicide prevention. This approach also requires investigators to view care through the eyes of patients, understand the patient’s journey in the care system, and to grasp the reality of the complex healthcare system the patient and next of kin have to navigate. Attention to interactions between different levels of the organisation is also needed. What happens at the microlevel, such as in personal meetings with patients, reflects decisions and management at the top of the healthcare organisation; as well what happens at the microlevel influences top-level decisions.40 These reflections on time, patient perspectives and organisations were generally non-existent in the investigations in this study but appear necessary to achieve progress in the care of suicidal patients.
The deficiencies in healthcare reported by the healthcare providers were in their investigations considered to be contributing factors to the completed suicide. This way of describing contributing factors is according to Swedish law and the RCA method. Healthcare and the suicide process both are complex processes, and such a linear approach might not be appropriate. This study illustrates how suicide as a possible patient harm is investigated in a nation where a RCA-inspired method is the recommended method, and what kind of learning and change in the healthcare systems that are possible with that approach. The result implies that sharper methods of investigation are needed to achieve progress in patient safety.
Limitations and strengths
All data were based on the healthcare providers’ reports of suicide to the supervisory authority. The contents in these reports are regulated by law; however, there still may have been shortcomings and inadequacies not pointed out and that the authority did not observe. The investigations were performed in different contexts by different persons with a large spectrum of disparities in experiences resulting in variegated quality. The investigations were performed after suicides, which often upset and strongly affect involved staff, and an awareness of external supervision might have biassed the outcomes. Furthermore, there is no national taxonomy for categorisation of deficiencies and actions; a coding scheme was therefore created and used in this study. The category of others was used only in a few cases, suggesting the categories in the coding scheme covered most of the reported deficiencies and actions.
The strengths of this study are that the data collection and categorisation were conducted by only one researcher, an experienced psychiatrist, to achieve consistency, and that the data were population based. This study was performed almost a decade after the obligation to report suicides was implemented and most providers and investigators would have been familiar with the procedure. Therefore, the cases in the study are expected to match the actual numbers to a good extent and the investigations are expected to be representative for suicides completed by patients in contact with healthcare within 4 weeks before death.
Many of the individuals who died by suicide were in contact with healthcare services shortly before death, and deficiencies in healthcare considered to be of significance to these deaths were reported for more than half of these patients. The majority of reported deficiencies and actions were at the organisational microlevel and the most common deficiencies related to care delivered in the immediate interface between patient and involved staff, which was easy for the investigators to identify. Actions proposed to prevent new suicides were centred on single educational interventions without distinctive sustainable effects in the organisations and usually did not correspond with the identified deficiencies. Conclusions from the investigations usually stayed inside the own department, systematic sharing and learning from experiences should be a future possibility to improve healthcare in a wider way and facilitate learning in practice.
Generally, the investigations lacked the perspectives of the patients and an analysis of the suicide process over time in connection with the complexity of healthcare organisations. Future research should examine if application of a framework based on knowledge of the suicide process, strategies of suicide prevention and patient safety would enable more sophisticated investigations facilitating progress in work on the prevention of suicide.
The authors are grateful to Region Jönköpings county and Futurum for funding and to Public Health Agency of Sweden for support.
Contributors ERaH designed the study, collected and registered the data, made the first analyses and wrote the manuscript. BAG, AR and ÅW contributed to the study design, analyses of the data and revisions of the manuscript. All authors read and approved the final manuscript.
Funding This study was funded by Futurum, the research centre at Region Jönköping county.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.
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.