Imagery about suicide in depression—“Flash-forwards”?

1Suicide is now among the three leading causes of death worldwide among those aged 15–44 years. Suicide attempts are up to 20 times more frequent than completed suicide (World Health Organization, n.d.). The scale of the problem is illustrated in that “worldwide more people kill themselves than die in all wars, terrorist acts and interpersonal violence combined” (International Association for Suicide Prevention, 2006). Effective treatments for suicidality are woefully lacking. Suicidal behavior occurs most frequently during a depressive episode. Recommended evidence-based treatments for depression are psychological approaches such as cognitive behavior therapies, also known as CBT (e.g. ). CBT is also highly promising for patients who have recently attempted suicide. However, few effective treatments exist specifically for those recurrently depressed patients who show suicidal features (e.g. suicidal thinking) and are known to be at increased risk of eventual suicide. To support innovative treatment development, we need to better understand the underlying cognitive processes.

2Mental imagery has been shown to be causal in determining future behavior, for example imagining voting and actually doing so later. Imagining an event not only increases the likelihood of engaging in what is imagined (e.g. signing up for cable TV; or doing exam revision; ) but also makes it appear more probable (e.g. winning an election or donating blood; ). Imagery is well known to enhance memory, and is also proposed to be important in driving personal goals. Further, imagery is associated with strong emotion-imagining short emotion-laden scenarios has a more powerful impact on affect than verbal processing of the same material.

3If the imagined event is useful (e.g. navigating to a desired destination), the above consequences of mental imagery (i.e. on action, perceived probability, memorability and emotion) have clear benefits. However, if the imagined event relates to suicidal behavior, these properties have very different implications. Translated into the domain of suicidality, the potential impact of mental imagery to increase the likelihood of action, increase perceived event probability, enhance memorability, goal-focus and intensify emotionality, would become particularly concerning if about suicide. Cognitive therapy has long held that cognitions can take the form of images as well as verbal thoughts. We know that intrusive, affect-laden imagery is key to posttraumatic stress disorder and social phobia, where such imagery can feel very real, compelling and distressing to patients. We know patients with depression can experience negative intrusive imagery of past trauma. However, to our knowledge, no previous work has examined whether suicidal patients report experiencing mental imagery related to suicide, such as planning a future suicide attempt.

4Mental health clinicians routinely ask patients about “suicidal thoughts” during risk assessment, but tend to focus on content rather than form (e.g. does the patient have specific plans about ending their life? If so, how lethal is the plan?). Such questions may fail to elicit the presence of suicidal images. Whilst there has been some previous reference to imagery in suicidality, the issue has received scant attention. suggest that patients may have difficulty fighting off detailed images of previous suicide attempts and describe a patient who said “I imagine I just end it all” but not how. developed a CBT intervention for patients with previous suicide attempts, which included a component where patients were asked to imagine a scenario likely to evoke suicidal thoughts to mind and then imagine an adaptive response. However, the utility of this component has not been evaluated and we neither know whether patients had spontaneously occurring images related to suicide nor what they consisted of.

5In summary, the findings from cognitive psychology discussed above inspired us to explore whether images also play a role in suicide attempts. Both the cognitive and clinical literature, to our knowledge, suggest that this may be a useful but under-researched topic. This study aimed to examine: (a) whether recovered patients with a history of suicidal depression report experiencing any suicide-related imagery during crisis (including of previous suicide attempts, future plans, or being dead) and (b) the nature and content of any imagery reported, including distress/comfort, preoccupation and realness. We sought to test whether a clinical measure of suicidal ideation severity was associated with either the time spent thinking about such images or perceived image realness. As it was the first study of this type, we focused on patients in recovery, recently treated to provide skills to deal with the distress evoked by discussing suicidal episodes in the interview.

6Patients with a history of depression and suicidality (ideation and/or behavior) who had recently completed treatment in randomized controlled trial were invited to participate. This treatment was a form of cognitive therapy recommended for relapse prevention in patients with recurrent depression (mindfulness-based cognitive therapy, MBCT;; ). All participants were in recovery at the time of participation (defined according to NIMH criteria;; ). Self-reports of suicidal ideation were operationalized according to the Beck Scale for Suicide Ideation—Worst-Ever Version (BSSw). Participants with current psychosis, eating disorder, OCD, severe alcohol or substance misuse, ongoing habitual self-cutting or insufficient written or spoken English were excluded. Patients with bipolar disorder were included if they had been free from manic symptoms for 6 months.

7MINI is a well-established, structured psychiatric interview used to assess current and past diagnostic status.

8BDI-II is a well-established 21-item questionnaire measure assessing depressive symptomatology over the preceding 2 weeks.

9The BSSw uses the first 19 items of the Beck Scale for Suicide Ideation (BSS;;; ), but is worded in the past tense and requires respondents to report on the time they felt at their most down and depressed about life. The first five items produce a screening score and the remainder are only completed if the participant reports that they had an active desire to kill themselves or intended to take a chance on life or death in a life-threatening situation. suggest that worst-point (rather than current) suicidality represents a significant predictor of an individual's future risk of death by suicide. The current minimal criteria for defining suicidality were a BSSw score indicating a “weak/no wish to live” in conjunction with a “weak/moderate or strong wish to die”.

10A structured interview was developed based on previous clinical mental imagery studies. The terms “verbal thought” and “mental imagery” were explained using examples and feedback elicited to confirm comprehension. A checklist was used to assess content of cognitions when participants were at their most despairing. Participants were asked whether they had experienced any images or verbal thoughts for each of nine topics (see Table 1), and to describe examples to check comprehension. All participants were able to readily identify distinct images. Participants identified the most significant image they experienced when at their most despairing or suicidal, and described it in detail. They were asked for any associated affect or meaning. Pilot work indicated that suicide-related images may be comforting as well as distressing so both were explored. Participants rated image distress and comfort on nine-point scales (1=not at all to 9=extremely). Participants rated how real these images felt (when at their most despairing) and the amount of time they experienced images specifically related to suicide (to assess preoccupation). Ratings were made on nine-point scales (1=not at all to 9=as if it were reality/all the time). Interviews were conducted by the first two authors.

11At entry to treatment, participants completed the diagnostic interview (MINI), BSSw and BDI-II. Following treatment (delivered by J.M.G.W. and M.J.V.F.), patients completed the suicidal cognitions interview. Participants were thoroughly debriefed. A clinical psychologist (E.A.H.) discussed how participants might use skills acquired during treatment if distress re-occurred.

12Fifteen of the 18 invited individuals participated (nine women, six men). The mean age was 41.10 (S.D.=8.41) and mean of years in education was 15.33 (S.D.=2.89). All participants were Caucasian.

13The median number of prior depressive episodes was 4 (M=5.54, S.D.=5.06, range 1–20); two individuals reported episodes too numerous to count. All patients were in recovery with a mean BDI-II score of 15.27 (S.D.=12.77). While four participants reported active suicidal ideation at entry to treatment, none did at the time of the interview. Participants reported the following axis I disorders: alcohol abuse (3), social phobia (3), generalized anxiety disorder (2) and agoraphobia without panic disorder (2). History of co-morbid lifetime disorders was as follows: alcohol dependence (3), substance dependence (2), substance induced psychosis (1), bipolar I disorder (1), manic episode (1), psychotic symptoms (1), panic disorder with agoraphobia (3), panic disorder without agoraphobia (1), agoraphobia (1) and dysthymia (1).

14In addition to MBCT, participants had previously received the following treatment: antidepressant medication (7), lithium and an antipsychotic drug (1), inpatient treatment for depression or mania (2), inpatient detoxification (1), and counseling or psychotherapy (12).

15All participants reported previous suicidal ideation. Severity of the worst episode was assessed using the BSSw. The median time since this period was 66 months (M=83.20, S.D.=64.45, range 10–240 months). The mean screening score (first five BSSw items, maximum=10) was 7.40 (S.D.=2.95). Thirteen participants completed the whole BSSw, i.e. reported an active desire to kill themselves. The mean total score was 22.34 (S.D.=6.59) of a maximum of 38—greater than the cut off of 16 used to identify high-risk patients who had 14 times the risk of committing suicide of those in the low-risk range. This is also higher than in a study (mean total score 13.3, S.D.=7.4) where patients were demonstrated to have cognitive vulnerability related to suicidality. Four participants had made prior suicide attempts. A further two reported previous habitual self-cutting. The remainder had experienced suicidal ideation in the absence of suicidal behavior (see Table 2).

16The following results refer to cognitions during participants’ most despairing or suicidal period. Table 1 shows participants’ endorsement of the nine checklist items for content, for both imagery and verbal thought separately. Imagery was reported in addition to verbal thoughts for all nine topics. For two categories, significantly more images than verbal thoughts were reported. Using a Wilcoxon signed ranked test, images of “what might happen if you died” (10/15 participants) were more frequent than verbal thoughts (4/15 participants), Z=2.45, N=15, p=.014. Second, images of “planning or preparing to make a (new) suicide attempt or harm yourself” (10/15 participants) were endorsed more frequently than verbal thoughts (6/15 participants), Z=2.00, N=15, p=.046. No other comparisons were significant, Z<1.13. Thus, it was observed that for categories related to the future (e.g. attempts, other people's reactions) imagery appeared more prevalent than verbal thoughts.

17All patients reported experiencing intrusive, repetitive suicide-related images when at their most depressed and despairing. For a summary (following ), including image descriptions, affect and meaning, see Table 2. Three examples follow.

18Example 1: Image of future suicidal action with behavioral consequence. The patient reported repeatedly seeing themselves jumping from a specific cliff and deciding whether to leave a final message or phone call. They described “looking at my feet and grass on one side and sea and rocks on the other side … and I just thought I’d jump”, describing this as resulting in “blank and black and nothing … it was like the ultimate salvation, my own choice”. This patient had a history of absconding from hospital, and attempting to travel to the actual cliff with the intention of acting on the image as described. They reported this image occurred even when mildly depressed.

19Example 2: Sensory image of dead self and funeral. This patient described “I could feel it, that cold damp feeling and being in a coffin”. They also reported visual images of the funeral with details of the specific church. They noted that they did not imagine how other people would feel as this “did not comfort” them. During the descriptions, they began “reliving” the imagery, saying “the images are coming into my head now, I can see myself in the coffin, I’m looking in”.

20Example 3: Comforting image of location providing opportunity for suicide. The repetitive image was of having the opportunity (rather than the aftermath which they did not want to visualize) of deliberately crashing their car at a particular spot on their everyday route. “It's like um, the road kind of comes like here and then the trees stop, but they grow over the top, it was sort of like there was almost a tunnel the trees going over the tunnel. And I kind of felt like I was being drawn towards the tunnel all the time. Which just dropped down …”. The patient reported that the image came “quite often when I was struggling, it looked like quite a welcoming site really … quite comforting in a way … An opportunity to escape … I think it made me more relaxed actually”. This patient had made plans including arranging accident cover and life insurance policies.

21All participants rated their suicide-related image when experienced at times of crisis as distressing (7/15 rating them as moderately distressing or greater). However, 12/15 participants also rated them as comforting (with 10/15 rating them as moderately comforting or greater). Ratings of distress (M=5.13, S.D.=3.38) and comfort (M=5.00, S.D.=2.75) were similar.

22About 10/15 patients reported being preoccupied with suicide-related mental images more than half the time during the crisis (M=6.07, S.D.=2.80). Increased preoccupation with images related to suicide was associated with higher scores on BSSw screening items (rs=.56, p=.029, N=15).

23About 10/15 participants reported experiencing their suicide-related images as feeling at least half-real (rather than just a mental event) (M=6.70, S.D.=2.94). Higher levels of image reality were associated with higher “worst-ever” suicidal ideation (BSSw screening items, rs=.54, p=.036, N=15).

24We aimed to explore: (a) whether recovered patients with a history of depression and suicidality reported experiencing any suicide-related imagery (as well as verbal thoughts) during crisis and (b) the nature and content of such imagery, and whether a clinical measure of suicidal ideation severity was associated with either the time spent thinking about such images or perceived image realness. Our results suggested that when suicidal, participants did indeed experience mental imagery in addition to verbal thoughts for all content categories covered. They were more likely to experience imagery than verbal cognitions for two categories (what might happen if they died and for suicidal plans); interestingly, both of which were future-oriented. While all images were rated as distressing, most participants reported that images were also comforting. High scores on the clinical worst-ever suicidal ideation measure (BSSw) were significantly associated with higher levels of image reality and increased preoccupation with suicide-related images. Collectively these findings indicate the importance of understanding imagery in suicidal patients, and that this previously neglected area is an exciting avenue for future research.

25All participants reported experiencing suicide-related mental images during crisis, including acting out future suicidal plans or being dead. As we argued in Section 1, imagery can have implications for future behavior (cf. also prospective cognitions; ). We suggest that the term “flash-forwards” to suicide may be used to describe these suicide-related images. This term relates to “flashbacks” to past trauma—a hallmark of PTSD, but to thoughts of future suicide rather than to past trauma. Flashback memories are rich sensory-perceptual images, rather than verbal thoughts, and are affect-laden, accompanied by a sense of reality or “nowness”—as if the past trauma is really happening. The “flash-forward” images reported here were also described as possessing sensory qualities, of being real and compelling, and rich in detail, as illustrated in Table 2. Inspection of the content of the most significant images related to suicide described in Table 2, alongside the imagery interview, indicates that the majority of participants’ suicide-related images (13/15) were of the future rather than only of the past.

26Clinical risk assessment and the identification of cognitive (and potentially modifiable) predictors for suicide are notoriously difficult. Our clinical measure of suicidal ideation (which does not refer to imagery) was associated with both image preoccupation and realness, though the large size of these correlations should be treated with caution given the sample size. The associations provide at least a good indication that asking about imagery during suicide risk assessment might be a useful inclusion, since this is not currently part of current clinical screening interviews. Interestingly, several participants commented that clinicians had never directly asked about suicidal imagery. To the non-clinician it may appear odd that patients do not spontaneously report suicide-related imagery. However, across psychological disorders (including PTSD and social phobia), patients frequently only report verbal cognitions if asked about “thoughts”. To elicit information about emotional and potentially shameful imagery (including about flashbacks), specific prompting is typically needed.

27Focusing on imagery has informed the development of highly successful and improved cognitive-behavioral treatments for posttraumatic stress disorder and social phobia. We propose that considering flash-forward suicidal imagery may also aid treatment innovation. Tackling imagery can require different cognitive therapy strategies than for verbal thoughts, proving new opportunities as in PTSD and social phobia. For example, cognitive therapy for PTSD includes “imaginal reliving”, where patients imagine their trauma in detail and add information to update traumatic meaning of specific hotspot images (e.g. “I’m about to be shot” to “I wasn’t shot”).

28Our data indicate several imagery features that therapy could target: changing the flash-forward outcome to an alternative to suicide, reducing imagery realness and preoccupation, and reducing comfort. Akin to work in PTSD and borderline personality disorder, “imagery rescripting” might be used to produce an alternative flash-forward outcome, e.g. a flash-forward to overdose could be rescripted to an image of disposing of the tablets (cf.; ). In the context of this Special Issue, it is indeed of particular interest to consider the potential of imagery rescripting techniques. used imagery with rescripting to change the meaning of schematic representations of the past. The majority of the images reported in the current paper were not of the past per se (though some may have reflected past suicide attempts), but of images of harming/killing the self in the future. It is speculated that perhaps just as we can rescript imagery of past moments, we may learn to rescript future-oriented images. For example, Arntz et al. ask the patient to enter the image as a “bystander”, and the therapist can then interview the bystander about his/her feelings, thoughts and inclinations, and thus learn how best to rescript the image. Similar techniques could be applied to suicidal imagery to incorporate alternative meaning and outcomes necessary for (and informed by) the individual patient, e.g. outcomes of safety rather than harm, alternative sources of comfort and so on.

29“Comfort” associated with imagery may be positively reinforcing (cf. craving imagery; ) and imagery techniques may be useful to break this circle of reinforcement. In our data, imagery reality (“nowness”) and preoccupation were associated with higher scores on the BSSw, so addressing these features could be beneficial. Interestingly, the feature of “nowness” has been related to severity of depression in intrusions of past events in a student sample, and this remains to be tested for suicidal “flash-forwards”. Mindfulness techniques may, for example, help patients relate to their images as “images” rather than as reality, allowing them to explore alternative strategies to either suppressing them or becoming attached to them.

30This novel data is limited by a small sample, requires replication, and should be extended to non-treated sample with active suicidal ideation (if adequate clinical support can be ensured post-interview). Also, the data do not bear on whether suicide-related imagery is specific to depression as opposed to other disorders. Further imagery properties need to be explored, e.g. observer perspective, vividness and so forth. However, in the clinical context of individual case histories, the data are extremely compelling. We have argued that the presence of mental imagery (as compared to verbal cognition) about an event can increase emotion, perceived likelihood, goal focus and even action, and that in the context of suicide this may pose significant risk. As a cognitive process not yet directly tackled in the suicidality literature, our data suggest that more extensive consideration of mental imagery about suicide may deliver benefits in an area where treatment innovation is sorely needed.