Application of an accident approach to the study of acute suicidal episodes through repeated in-depth interviews

Abstract

We modified an accident analysis model for the study of acute suicidal episodes (ASEs). Our aim was to use this model (SCREAM, Suicidal Cognitions’ Reliability and Error Analysis Method) as a lens to understand the worst-ever ASEs of nine patients who took part in repeated in-depth interviews. Guided by the theory of SCREAM including four predetermined categories, nine themes were identified. ASEs were triggered by interactions with the interpersonal and physical environment and spiraled into a state of lost control. Timing and the availability of promoters and barriers in the environment were salient features. Findings may aid person-centered safety planning.

Assessing and treating suicidal persons constitute challenging tasks for mental health professionals (Waern et al., 2016). Numerous models have been introduced to aid our understanding of processes that lead to suicide. However, as pointed out in a recent review (Ajdacic-Gross et al., 2019) these are often based on theoretical grounds that emanate from specific disciplines, and may thus lack compatibility with suicide prevention approaches, which tend to be interdisciplinary. In their review, Ajdacic-Gross and colleagues stressed that only a very few of those who think about suicide actually die by suicide, and suggested that suicides could be construed as mental accidents, i.e. “failures to withstand temporary suicidal impulses” (Ajdacic-Gross et al., 2019).

The accident paradigm is not new to Sweden. In fact, it was clearly reflected in the National Action Plan for Suicide Prevention, proposed in 2006 (Socialstyrelsen, 2006) and approved by the Swedish parliament in 2008, which suggests that suicides can be construed as “psychological accidents”. The National Plan includes Sweden’s Zero Vision which emphasizes the situational context, stressing that no person should find themselves in a situation in which suicide appears to be the only alternative. The accident paradigm was introduced to the Swedish suicide prevention community by the late Jan Beskow (fourth author), a suicidologist who recognized similarities between suicides and accidents early on (Beskow et al., 1994). Beskow emphasized that both phenomena can involve loss of cognitive control in overwhelming situations. He was inspired by an error analysis method long applied in industries and air traffic in Sweden, the Cognitive Reliability and Error Analysis Method (CREAM) (Hollnagel, 2004), which was later modified for the analysis of traffic accidents (DREAM) (Ljung et al., 2007). Beskow (2010) suggested that a similar approach might be applied to the analysis of acute suicidal episodes (ASEs, ie delimited time periods in which suicide risk is imminent). He proposed that such a model could be called SCREAM (Suicidal Cognitions and Reliability Error Analysis Method), the acronym alluding to the intense psychic pain experienced by the suicidal person. In the initial stage of the current project, we identified major components of the above-mentioned traffic accident model (Ljung et al., 2007) and modified these to develop the SCREAM model. Paralleling the DREAM model, SCREAM identifies underlying long-standing “Latent conditions” as well as “Acute situational conditions” with an imminent risk, hours or days before the suicide attempt. “Triggering conditions” elicit the transition between latent and acute phases. The acute situation is the brief period of time in which the individual may experience a threat to such a degree that it leads to loss of control and compromised decision-making. Paralleling phenomena described in the resolution of traffic accidents, we suggest that the potentially fatal outcome of an ASE can be averted if “Recovery conditions” are present.

The aim of this study was to apply the SCREAM model as a lens to explore personal narratives of worst-ever ASEs. Each study participant took part in a series of in-depth interviews, which provided detailed descriptions of their lived experiences.

Material and methods

Participants

Participants were recruited from adult psychiatric outpatient services in Western Sweden. Purposeful selection provided variation in terms of sex, age and diagnosis. Inclusion criteria were (1) discernible episodes of acute suicidality, (2) ongoing treatment contact with a mental health professional and (3) ability to speak about own psychological matters, as determined by the health professional. Exclusion criteria included incapacity to participate in interviews due to severe psychosis or substance abuse, dementia, or insufficient language proficiency. Interviews were scheduled to ensure they did not take place during a state of elevated suicide risk, allowing participation of persons with intermittent high suicide risk. Ten potential participants received study information; six women and three men accepted to take part. Table 1 shows participant characteristics. Clinical diagnoses reflect our purposive sampling; comorbid conditions were common. While none had a primary diagnosis of alcohol use disorder, all but one had AUDIT scores indicating problematic consumption (Saunders et al., 1993). All were on antidepressant medication.

Table 1. Participant characteristics at study inclusion (n = 9).

Characteristic	n
Age (years)
19–24	3
25–34	2
35–45	2
46–66	2
Living arrangements
Alone	4
With parents	2
With spouse	1
With ex-husband/ex-boyfriend	2
Marital status
Married/partner	2
Single	7
Post-secondary education	4
Employment
Sick leave	5
Employed	2
Retired	2
Primary clinical diagnosis
Emotionally unstable personality disorder	3
ADHD /ADD	2
Bipolar disorder (I/II)	2
Recurrent depression	2
AUDIT score (Saunders et al., 1993)
0	1
8–12	7
≥13	1
Chronic physical conditions	6
Prescribed medication
Antidepressant	9
Mood stabilizing	3
Disulfiram	2
ADHD medication	1
Sedatives/hypnotics	6

Procedures

The study was designed as a series of interviews with each participant. The first author carried out all interviews. Participants could choose to meet with the interviewer at their own outpatient clinic or at the hospital research facility. A baseline interview initiated each series to establish contact, to ensure that study criteria were fulfilled and to collect background data. A series of three in-depth interviews followed to elucidate the course of events and the context of ASEs. Brought into the interviews was the concept of time-limited ASEs in accordance with Rudd’s modal theory (Rudd, 2000). An interview guide included initial open-ended questions: “Can you tell me about the time when you experienced an ASE? What happened during this episode? Describe what happened, from the beginning to the end, as detailed as you can, sequence for sequence”. Targeted probe questions followed. Milder episodes were requested first, allowing a connection to develop between the interviewer and the participant before moving on to more severe experiences. The “worst-ever” episode was specifically requested and provided the material for the current analysis. Such episodes, as rated by the SSI-W, have shown to be associated with suicide in a four-year follow-up of psychiatric outpatients (Beck et al., 1999).

The course of events was reconstructed through collaborative case conceptualization (Kuyken et al., 2008), a method from cognitive psychotherapy. Description and analysis of ASEs, including identification of useful coping strategies were jointly made with mutual respect. The participant, with unique knowledge of her/his own experience, actively narrated the story, while the interviewer asked open questions to aid reflection. The story grew through the dialogue, a “guided discovery” (Kazantzis et al., 2018). The five-factor model (Padesky, 1990), a cognitive theory model describing the interactions in a specific situation between cognitions, emotions, physiology and behavior, aided in systematically capturing the rapid interactions during the ASE or “suicidal mode” (Rudd, 2000).

Eight of the participants took part in all three in-depth interviews. One left the study after the second in-depth interview in connection with a relapse of severe substance use. Each in-depth interview lasted 60 − 150 min, was audio-recorded and transcribed verbatim. Several weeks (median 5) passed between interviews. During this time, the interviewer reviewed previous interviews in preparation for the next. In a final interview, the interviewer summarized what had been learned over the repeated interviews and requested validation. Mean total participation time was 9 h (range 4−14). All participants gave written informed consent. The Regional Ethics Board in Gothenburg (333-10) approved the study on 12 October 2010. Interviews were conducted 2012–2015.

Analysis

Textual data consisted of 26 transcripts with rich descriptions of the sequence of events in ASEs. The analysis in the present study is based solely on data pertaining to each participant’s worst-ever episode. Directed content analysis (Hsieh & Shannon, 2005) was applied. A directed approach is used to validate, refine, or extend an existing theory in a new context. The theory predicts concepts of interest and guides development of categories into a formative coding scheme. Initial coding categories are thus deductively derived from existing theory, which in this study comprised the four components derived from the traffic accident model and modified for ASEs (Latent conditions, Triggering conditions, Acute situational conditions, and Recovery conditions). A preliminary analysis was made after the first three in-depth interviews. Thereafter, data were analyzed continuously, parallel to data collection. Transcripts were reviewed several times to capture the essence. The first author identified meaning units related to the aim and these were coded in collaboration with the second author, who participated in all stages of the analysis. Preliminary codes were grouped and categorized, and themes were identified. Throughout the data analysis, a constant comparison technique was applied, comparing themes with categories, developing links, and creating themes. Interpretation was made by relating the narratives to the SCREAM model. All authors were involved in the completion of the analysis.

Results

The median Montgomery Åsberg Depression Rating Scale (Montgomery & Asberg, 1979) score at the initial interview was 19, with a range of 9–33, indicating mild-moderate depression. The median score on the suicide item of the Hamilton Depression Scale (Hamilton, 1960) was 0 (range 0–2). All participants scored high on worst-point suicidal ideation, ranging from 25 to 36 on the Scale for Suicide Ideation at Worst-Point (SSI-W), well above the standard cutoff at 16 (Beck et al., 1999). Additional clinical data is detailed in Table 2. The first suicide attempt occurred in childhood or young adulthood for all but two. The worst-ever episodes occurred within the past few years in all but one participant.

Table 2. Characteristics of suicidal behavior in study participants (n = 9).

Characteristic	N
Number of suicide attempts at baseline
1–2	3
3–5	4
>9	2
Age at first suicide attempt
4–12	3
13–21	4
≥40	2
Non-suicidal self-injury
Ever	6
Past year	3
Age at worst episode
18–25	4
26–35	1
36–45	3
46–65	1
Method of suicide attempt at worst episode
Poisoning (prescription medication)	5
Cutting	2
Gassing, by car in garage	1
Jump from balcony (aborted)	1
Time since worst episode
Past year	4
1.5−2 years	3
5 years	1
25 years	1

Each participant described the worst-ever episode as a brief episode that culminated in a suicidal action. Findings (themes) relevant to these worst-ever ASEs are elaborated below under the headings of the predetermined categories (Latent Conditions, Triggering Conditions, Acute Situational Conditions and Recovery Conditions).

Latent conditions

This category describes conditions distant in time, that shaped the context in which the ASE took place, in the form of one comprehensive theme.

Long-term negative experiences

This theme involves experiences, most of which are carried since early childhood, and often impacted on identity including loss of health and function and consequences of a mental disorder. Loneliness is central in this suffering. Isolation from others was present early in life. Participants experienced abandonment, rejection, and betrayal in close relationships. Formation of attachments had been disrupted. One participant had serious burns as a 1-year-old with consequences throughout childhood:

The first six months she (mother) could not stay with me because I was in isolation… then she could stay with me but was not allowed to hold me. And the continual separations, before each operation, when you had to say good-bye. (P5)

Narratives included interpersonal violence, sexual abuse or bullying. Some were exposed to suicidal behavior in family or friends. Others experienced acute physical illness or loss of a significant other. Negative societal attitudes toward mental disorder constituted a source of shame, and a barrier to help-seeking.

Triggering conditions

This category describes interacting conditions that triggered the development of an ASE at a specific point in time, captured in a single theme.

Threatening environment

Relational problems were prominent features within this theme. Exposure to psychological and physical violence in close relationships could constitute a powerful trigger.

He (boyfriend) tugged me into the room and beat me, my ear was bleeding. I have been beaten so many times before (by others) and this time I felt - I am going to die. This is it. There is nothing left. (P1)

The breakdown of a relationship could mean humiliation, abandonment, and disconnection from others. Participants felt hurt and insignificant. One person discovered that her boyfriend had made a fool of her on social media.

I meant nothing to him. He said that everything was my fault… I don’t deserve better. I’d rather disappear and get rid of it. (P9)

In the threatening environment, support from family, friends, and caregivers was insufficient. Feeling abandoned by professionals could trigger an ASE.

My doctor and psychologist gave up. No one has hope. There is no more help. (P3)

Triggering was facilitated by the influence of substances such as alcohol, which intensified anxiety and negative thinking. Also, the availability of a potential suicidal method could trigger an ASE.

Then it came to me quite suddenly when I was cutting fruit… I held the knife in my hand, and it came like a voice… It is terribly scary… But these are thoughts that I cannot control … It started something that stayed. (P7)

Acute situational conditions

This category comprises conditions that affect the outcome of the ASE by interacting directly “here and now” within the acute situation, presented below in three themes.

Looming and cognitive insufficiency

This theme highlights how multiple conditions interact almost simultaneously, accelerating into a looming state that is hard to exit and difficult to endure. Looming is described as an experience of intrusive, threatening cognitions that approach at a rapidly accelerating speed (Riskind, 1997). The perceptions and possibilities of choice and action are vague and difficult to handle which intensifies mental pain and fear. The demands become overwhelming, resulting in powerlessness.

My whole life came like a fast film… there was… only pain… It was like a struggle between living or dying, a battle between two monsters within me. One side wanted to live and the other one wanted to die… I felt sad, angry at myself, and tired. (P6)

Anxiety increases rapidly in the suicidal spiral and feelings, including shame, are overwhelming. Mental pain becomes unbearable. Participants experienced chaos. Exhaustion, as well as intake of alcohol and tranquilizers contributed to an inability to think clearly and to see alternatives, resulting in confusion. At this point, participants were stuck in a state far from normal, a state of cognitive failure. They saw themselves as malfunctioning or unworthy, blaming all failure on themselves. There could be a firm conviction that the future would never improve, or that there was no future.

I’m a broken article that can be sent to the scrap yard… I was torn up from within, having pain in my chest that was squeezed, and I was gasping for air. It was a claustrophobic feeling and panic… I felt empty, soulless. (P7)

Participants were convinced that no one could help. They felt caught in a trap with no exit; they were not able to reach others. Their wish to protect next-of-kin could mean that they hid how they really felt. This, in addition to previous negative experiences of health care, could hinder help-seeking. Dysfunctional avoidant behavior appeared, involving sedation by alcohol, and self-harm through cutting.

Point of no return

This theme is derived from experiences of how the chaotic suicidal spiral culminated in a loss of control; mental pain was incessant, and suicide the only available tool that remained. Participants felt they were in a hurry, which one illustrated by snapping his fingers.

It is nothing you plan… The pain grew and grew; I couldn’t control it… As if there are two people in one person, one gives strength to do it, the other tries to give reason not to. It becomes a big fight between them and when that one wins - It comes immediately, on a click. (P6)

Exceptional state of carrying out a suicidal act

This theme refers to experiences of implementing a suicidal act while in a special state of mind and environment. The entire consciousness could be preoccupied with the suicidal focus –

Everything revolved around suicide… It was as if one were staring for hours at a piece of paper with the word “suicide” written on it, even if you had a paintbrush in your hand, you can’t figure out that you could cover it up (with the paint), it’s that physical. (P7)

Participants felt dissociated from feelings or body as if they had already ceased to exist. Behavior became more automatic.

When it’s not thoughts anymore but actually becomes action of it… Then I was just absorbed… I did not feel scared… I don’t think I knew what it would mean, how it would feel… I probably would have been more afraid if I had known what I was facing, but I didn’t think about that. (P3)

I didn’t think much, I was just thinking that I would have time to swallow the pills before he sees …. I don’t think at all. (P1)

Circumstances could enable the suicidal act. Partners or family might be away, the participant could be at home alone without anyone knowing how bad they felt. The presence of guests at home might prevent the act temporarily, but when the guests were gone, so was this barrier. Being away from home could mean having no access to the usual coping strategies, e.g., the company of a pet. Harmful methods accessible then and there were employed.

I went into the garage and closed the door behind me, the pump for our sailboat was in the car and I arranged it so it would be as effective as possible. (P4)

The environment provided an opportunity in time and space: suddenly the appropriate opportunity appeared. Hence, the exceptional state implied an unconscious and automatic behavior, allowing cognitive errors and the influence of environmental conditions to dominate.

Recovery conditions

Four themes were identified as components of importance for the resolution of the ASE.

Recovery by the suicidal person

This theme is derived from the strategies used by participants in the moment. Narratives suggested that the participants were not particularly cognizant of these strategies, but these could be identified during the collaborative interviews. Cognitive errors during the suicidal act (e.g., forgetting to lock the bathroom door) could make room for recovery in the short-term. Other strategies could allow emotional regulation and regained access to problem-solving ability. Thoughts could be focused on something concrete, e.g., body sensations that might be consequences of the suicidal act.

Seeing my body destroyed by wounds and scars made me feel regret and the death wish dissolved. (P6)

Distractions could include focusing on basal needs. Upon waking up in the hotel room after a suicide attempt, one participant focused on the thought of having a nice hotel breakfast.

Compensation by others

This theme highlights the significance of external recovery conditions, in terms of immediate help and support. To be able to listen to the person and take responsibility, helpers needed appropriate skills.

My wife got worried when I didn’t answer the phone… She called our son. He could feel that something was wrong and called an ambulance. (P2)

A friend I trusted started to chat with me, noticed that I was quiet, that I felt bad, and it came out that I was thinking of killing myself. He understood, knew it was serious, comforted me. He saved my life…. He dressed the wound. I felt less worthless when he showed he cared. (P9)

Compensation by the physical environment

This theme is derived from the experiences of suicide being prevented by conditions related to the physical environment. Something in the environment could distract, moving focus away from the suicidal act. Also, access to appropriate rescue means in the environment could create the conditions for others to be able to help.

The keys that I forgot. It was the keys that I left on the outside of the garage door. (P4)

Physical distance could be important; it impacted on the timing of potential compensation. The temporal aspect or timing could matter for recovery, i.e., to intervene at the right moment. There was a strong thread of getting help by chance or coincidence, e.g, someone arrived unexpectedly.

My wife noticed when we talked on the phone that something didn’t make sense and she called my ex-wife who lives nearby, asking her to go to check on me. (P6)

Intrinsic mechanisms

This theme reflects a self-limiting property of the ASE when the acute state is switched off. As time passed, the person (and the situation) could calm down and cognitive ability was reset, allowing orientation toward coping. One person, whose worst ASE involved a discontinued attempt, described a sudden reset.

It struck me, if I jump from the balcony… there is no time to regret when I’m in the air… At last, I could see that I had a paintbrush in my hand, able to paint over the suicidal thoughts. (P7)

Discussion

Our research approach stimulated active participation, yielding rich descriptions of each individual’s lived experiences. Triggering conditions, relating primarily to the proximal interpersonal and physical environment, interacted with latent conditions. The ASEs then evolved as complex spirals of arousal, looming and cognitive insufficiency, peaking at a point of lost control and the exceptional state of carrying out a suicidal act, during which the person appeared to act rather automatically and was highly influenced by the environment, which, at that specific time, provided an accessible suicide method and no available barriers.

Numerous constructs of contemporary suicide models were apparent in the participants’ narratives. The threatening environment could involve situations leading to thwarted belongingness, a salient component of the interpersonal theory of suicide (Chu et al., 2017). Central to our participants’ narratives was the phenomenon of unbearable mental pain, originally described by Shneidman’s “psychache” (Shneidman, 1985), and highlighted in the Three-Step Theory (Klonsky & May, 2010), which also stresses hopelessness. In our SCREAM analysis, pain and hopelessness were apparent in the long-term negative experiences. Symptoms escalated to an unbearable extent in the experience of looming. Looming vulnerability (Riskind, 1997) elicits anxiety which in turn may evoke suicidality. The transition into the exceptional state, as described in the SCREAM model, parallels the transition from ideation to behavior, central to the Integrated Motivational Volitional model (O’Connor & Kirtley, 2018). In SCREAM, the transition is marked by a situation in which the suicidal acts tend to be carried out while in an exceptional state, where rational behavior is shut-off in favor of an automatic unconscious behavior. This enables errors in decision-making, which can culminate in a suicidal act. The intrinsic mechanism by which the ASE may be resolved was one of the four themes that emerged in the category Recovery conditions. This included a sudden change of thought when a state of dissociation/autopilot behavior is switched off and the reflective system is restored. These phenomena relate well to previous findings of a switch on/off mechanism in acute suicidal states (Orbach et al., 2003).

Considering the many commonalities described above, what does our research add to the understanding of acute suicidal states? The proposed SCREAM model emphasizes the complexity of interactions at play. The ASE is seen as a process; focus is shifted from individual factors that elicit suicidal behavior to an understanding of the situation that enabled a suicidal act. The SCREAM approach provides a person-centered perspective in which the suicidal person is seen as a competent individual who temporarily loses control in an overwhelming situation. Losing control in the presence of an easily accessible suicide method, rather than the intent to die per se, seemed a primary driver of the suicidal act. The SCREAM model places emphasis on the situation that brought on the act, and we believe that this can reduce the individual’s feelings of shame and blame. The exceptional state described by participants aligns well with “a predominance of automated, Pavlovian processes” in situations of social stress as described by the authors of a recent review of the decision neuroscience literature on suicidal behavior (Dombrovski & Hallquist, 2017). Factors that elicit the transition from the exceptional state back to a focus on living constitute important components of the SCREAM approach. Again, the context (the interpersonal and physical environment) is emphasized. By listening to the person’s own narrative, both the patient and the clinician can learn how and why ASEs develop. The theory of SCREAM might help to explain why interventions such as ASSIP (Gysin-Maillart et al., 2016; Michel et al., 2017) and CAMS (Tyndal et al., 2021) are effective. Both these interventions involve working side by side while uncovering the details of situations that elicit suicidality. Such an approach puts a spotlight on the individual’s own perspective, which is highly valued by patients (Michaud et al., 2021).

Implications for the clinician

Increased understanding on the part of both the patient and the clinician can also reduce the clinician’s own anxiety. Focus is shifted from the demands of a health care organization that expects the clinician to save the patient at all costs (Smith et al., 2015) to a collaborative effort to make the system surrounding the patient as safe as possible. The suggestion that a model derived from accident research might provide new perspectives on ASEs does not imply that suicidal behavior is random or uncontrollable, nor does it suggest that suicidal acts are lacking in intent. When the function of the suicidal thoughts and behaviors is clarified, possible problem-solving strategies appear. New insights and coping strategies can be collaboratively consolidated into potential safety measures. The SCREAM approach might aid suicidal individuals and clinicians alike in the identification of targets for person-centered safety planning. Individually tailored safety planning is central in the care of persons with suicidal behavior (Stanley et al., 2018). Increased knowledge about what elicits ASEs can empower the individual, who can find ways to avoid getting into that state in the first place. Learning more about compensating mechanisms (“recovery conditions”) provides insight into how factors that can increase the chance of positive outcomes might be strengthened, contributing to the “adaptive toolboxes” described by persons with lived experience of suicidal behavior (Scarth et al., 2021). As pointed out by Deisenhammar and colleagues (Deisenhammer et al., 2009) the acute suicidal process tends to be very brief. A short window for intervention during an ASE means that others (friends/relatives/inpatient staff) need to be aware of how they might intervene in acute situations.

Methodological considerations

Strengths of the study include the use of repeated interviews and the collaborative approach which facilitated data collection. Participants could tell their stories to an attentive listener who validated experiences, helped to maintain focus and elicited details. All nine had experienced chaos and cognitive insufficiency in connection with their “worst-ever” episodes. Despite this, upon telling and being listened to, all demonstrated outstanding narrative competence. The interview guide facilitated the collaborative process of comprehending “what led to what”. The qualitative approach provides insight into subjective experiences and contexts (Hjelmeland & Knizek, 2017) providing a broad understanding of each individual process. We chose directed content analysis in order to achieve a systematic exploration of the model that we modified from traffic accident research, and thus approached data from an informed position. In addition to the theoretical framework presented in the introduction section, the interviewer’s pre-understanding is built on clinical experience, as well as active participation in a community-based suicide prevention NGO. One researcher conducted all interviews which contributes to trustworthiness of data collection by minimizing risk of divergent foci. Trustworthiness was strengthened by the reflective and systematic discussion of interpretation during final analysis with coauthors, who represent different professional perspectives. Some limitations need mentioning. The sample was small and persons under 18 and over 66 were not included. The SCREAM model might be less applicable in older adults, whose suicidal behavior may be characterized by lower levels of anxiety and greater intent (Wiktorsson et al., 2021). While vivid descriptions of worst-ever episodes were provided, even when distant in time, recall bias remains an issue.

Conclusion and suggestions for future research

This exploratory analysis suggests that the SCREAM model might have potential for use in clinical suicide prevention. However, the findings herein represent just a first step, and further research is needed to elucidate the potential role of cognitive insufficiency and the complex interactions within ASEs in diverse patient groups and settings. Examples of specific groups that could be involved in future studies include older adults, survivors of suicide attempts characterized by high medical lethality, as well as survivors of attempts characterized by a relatively slow transition between the decision to die and the actual act (Paashaus et al., 2021). Biometric assessments could make it possible to objectively explore interactions between suicidal persons and their interpersonal and physical environments in real time.

Acknowledgments

The authors are grateful to all study participants who so generously shared their experiences.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The study was supported by grants from the Swedish Research Council [5212011-299, 2016–01590], the Swedish state through the ALF agreement [ALFGBG-433511, ALFGBG-715841], the Gothenburg Center for Person-centred Care, and the Gunnar and Märtha Bergendahl Foundation.