Some concepts from the threat assessment literature

The pooling of these varied areas of inquiry reveals shared themes that are significant in the stalking of celebrities and public figures.

First, pathological fixation (Mullen Reference Mullen, Pathé and Purcell2009b) – the development of an abnormal, highly focused and all-encompassing interest in a person or cause that comes to dominate the life of the stalker, leading them to believe they are entitled to a relationship with the person whom they are pursuing. This fixation distinguishes them from ordinary fans, who may hope for a relationship, but do not have this sense of ‘entitled reciprocity’ (Meloy Reference Meloy, Sheridan, Hoffmann, Meloy, Sheridan and Hoffmann2008).

Second, grievance and the importance of thinking about the motivation for engaging in this unwanted behaviour, which we explore in more detail below.

Third, serious violence as a possible outcome, and specifically targeted violence, i.e. aimed at a specific person and likely to have been planned in advance. This offers opportunities for earlier discovery, prevention and disruption. It also distinguishes it from the sort of violence we may be more used to in psychiatric settings – namely, impulsive violence where a person in a highly aroused state strikes out, in the moment, at whoever happens to be nearby. Sometimes the original target is never reached – perhaps because they are too well protected and a more accessible one is selected instead. In our experience, when there have been adverse outcomes, it has often been the friends and family of the stalker who have been the victim. Threat management professionals themselves need to be mindful that they may become a part of the grievance (Federal Bureau of Investigation 2017). Violent outcomes are, of course, rare, and we will also be interested in preventing the persistence of the stalking behaviour, the psychosocial damage it causes to the stalker and the victim, the risk of an escalation of the behaviour, and the risks of disruption at high-profile events or locations.

Epidemiology

Stalking in the general population is common – in the UK the lifetime prevalence of being stalked is 15% (20.2% for women and 9.8% for men (Office for National Statistics 2016)). The prevalence of stalking of celebrities and public figures is less researched, but probably much higher. For example, a survey of UK members of parliament (MPs) (James Reference James, Sukhwal and Farnham2016a,Reference James and Farnhamb) suggested that 81% had experienced, at the hands of constituents, at least one of the twelve inappropriate behaviours asked about in the study, and the career prevalence of being stalked was 38%. In addition, 18% had been subjected to an attack or an attempted attack. In The Netherlands, 33% of politicians had been stalked during their lifetime and 46% threatened (Malsch Reference Malsch, Visscher and Blaauw2002), in Sweden 74% of MPs reported having experienced harassment, threats or violence (Staatens Offentliga Utredningar 2006) and 28% of Norwegian MPs had experienced ‘genuine stalking’ (Narud Reference Narud and Dahl2015). In terms of celebrities, a German study reported a 79% career prevalence of stalking among television personalities (Hoffmann Reference Hoffmann, Sheridan, Meloy, Sheridan and Hoffmann2008).

Mental disorder, warning behaviours and attacks

The stalkers of public figures and celebrities come to notice either by their unwanted communications or by inappropriate approaches. The Exceptional Case Study (Fein Reference Fein and Vossekuil1999) looked at 83 people who had attacked, or approached to attack, a prominent public figure in the USA between 1949 and 1996: 61% had a history of psychiatric illness, 43% a history of delusional ideas (88% had acted on them) and 11% experienced command hallucinations. For 77% there was a history of verbal or written communication with the target. In a survey of attacks on European politicians between 1990 and 2004 (James Reference James, Mullen and Meloy2007), 10 attackers (42%) were psychotic at the time, and there was evidence of warning behaviours in 11 cases. Looking at 23 attacks on the British royal family between 1778 and 1994 (James Reference James, Mullen and Pathé2008), 48% of the attackers were psychotic, and 10 out of the 23 exhibited warning behaviours. There are similar findings from the Dutch royal family (van der Meer Reference van der Meer, Bootsma and Meloy2012), Canadian justice officials (Eke Reference Eke, Meloy and Brooks2014), the US Congress (Dietz Reference Dietz, Matthews and Martell1991a) and Hollywood celebrities (Dietz Reference Dietz, Matthews and Van Duyne1991b). This is a group with very high rates of severe mental illness, primarily psychosis, and with high rates of ‘warning behaviours’. Taken together these two clinical features offer the opportunity to intervene to prevent adverse outcomes.

In examining the behaviour of public figure stalkers, several important features to look out for emerge from the literature (Box 1), as well as suggesting close attention be paid to ‘warning behaviours’ (Box 2), indicators that the subject may be moving along a hypothesised ‘pathway to violence’ (Calhoun Reference Calhoun and Weston2016). It is unclear how useful this theoretical ‘pathway model’ is in practice. Experience in this field suggests that individuals rarely move smoothly along a path, can skip certain of the hypothesised steps and can proceed to violence without engaging in the other steps. However, evidence of finality in language and leakage to a third party are well-recognised precursors in many of those who do resort to violence.

Whatever model is adopted, an effective threat management unit needs to capture such ‘knowable’ information, to share it with the right people in a timely fashion and to respond promptly to proactively manage behaviours of concern. This will involve raising awareness of the unit's existence and purpose with a wide range of potential stakeholders, as well as sharing information about the underlying empirical evidence for this work.

Motivational typology

Why would one choose to correspond with or to meet a public figure? Cases cluster into similar motivational groups to those in the wider stalking literature (Box 3), although particular public figures may tend to attract one group more than another. For example, Parliament has the constitutional function of resolving grievances, so it comes as no surprise to learn that MPs often attract those seeking help with pathological grievances. Young, attractive, eligible celebrities or royalty may be a particular focus for erotic interest, and therefore for the pathologies of love, too.

Threats

Threats occupy an odd position in this work and have a complex literature. Sending correspondence which is not threatening is associated with a higher likelihood of approaching (Meloy Reference Meloy2014). There is one group of stalkers – ‘howlers’ (Calhoun Reference Calhoun and Weston2016) or ‘shockers’ (Warren Reference Warren, Mullen, McEwan, Meloy and Hoffmann2014) – whose aim is to cause fear and distress in their target and they therefore send very alarming and threatening correspondence, but are unlikely to approach or attack as they have achieved their purpose with their writings. However, making threats is also associated with a high level of mortality in the longer term, both suicide and homicide (MacDonald Reference MacDonald1968; Warren Reference Warren, Mullen and Thomas2008). The absence of threatening communications should not reassure the clinician that the person does not pose a threat, and if threats are made, they should be taken seriously. One modifying factor may be the ‘intimacy effect’. Violence is much more likely in a stalking case when the two parties have formerly been intimate, and in this group, threats of violence are significantly associated with actual violence (Calhoun Reference Calhoun and Weston2016).

Unusual clinical features

In line with the studies, it is our experience that those who approach or send problematic correspondence to public figures have high levels of severe mental illness. There are also some unusual features. There are high rates of encapsulated delusional disorder (including, of course, erotomania), where the person can appear to maintain a high level of functioning and the presence of mental illness can be missed altogether by the inexperienced. We have also encountered many cases of folie à deux and internet-facilitated folie à plusieurs, where people with paranoid illnesses, particularly those featuring some form of intrusive government surveillance, are able to find others with the same delusional beliefs online. In particular, there are communities of people who describe themselves as ‘targeted individuals’ and as victims of ‘gang stalking’ (Sheridan Reference Sheridan and James2015). Their symptoms often include highly concerning horrific delusions, and several recent US mass shootings have been perpetrated by ‘targeted individuals’, apparently as a direct result of their psychotic experiences – Jiverley Wong in the 2009 Binghamton shootings, Aaron Alexis at the Washington Naval Yard in 2013, Myron May at Florida State University in 2014, and Gavin Long in Baton Rouge in 2016, for example. In other words, not all delusional beliefs are equal when it comes to making a clinical assessment of the patient (see ‘threat/control-override’ symptoms and violence risk in psychosis more generally (Link Reference Link, Stueve and Phelan1998)).

Peripatetic patients are also common. People are attracted to particular public figures (especially the Queen, the US President, the Pope and certain Hollywood celebrities) from around the world, and travel internationally in pursuit of delusional quests. Sometimes people travel around the country or the world in an attempt to flee delusional persecution. Sometimes people move to evade mental health services, often with great success. This can present great difficulties for mental health services that are tied to catchment areas and legal jurisdictions.

A word on celebrities

Much of the threat management research has centred on public figures. The response to managing threats is dependent on the government response in different countries. For example, the existence of an embedded National Health Service, with mental health teams arranged according to geographical location, has been fundamental to the ability of FTAC to operate in the UK to assist with the diversion of those needing psychiatric treatment into health care rather than punishment in the criminal justice system. It is no surprise that the proliferation of stalkers of celebrities in Los Angeles was noted following the murder of Rebecca Schaeffer by a stalker in 1989, and that it led to a change in stalking laws in California and the establishment of the LAPD Threat Management Unit. However, for a celebrity sitting outside of this specialist jurisdiction, it is likely that any management of a stalking case would fall to local police. There are differences in presentation and it is unwise to assume that all lessons learnt with public figures will provide a total description of the circumstances facing a celebrity victim of stalking. It is long noted that celebrities are often reluctant to bring stalkers to the attention of the authorities. There may also be difficulties in distinguishing a stalker from an obsessive ‘superfan’. The importance of education for those in the public eye without the safeguards enjoyed by some public figures should be emphasised, although it remains to be seen how this is best achieved.

Social media

Many of the studies undertaken on stalkers of celebrities and public figures pre-date the dawn of social media. Nowadays, maintaining a large following on social networking sites such as Instagram and Twitter is seen as essential to the business of self-promotion. Celebrities often address their fans in a familiar way, promoting the idea of intimacy. The stalkers among their potentially large audience, some of whom may be psychotic, may misinterpret this familiar tone. Social media can be used as a ‘weapon’ in the stalker's armoury (Suzy Lamplugh Trust 2016), and general guidance on the use of social media and cyberstalking would be of benefit to celebrities. Indeed, minor celebrities with no access to advice or security may be particularly vulnerable.

Threat assessment and risk assessment

Threat (unlike risk) assessment is conducted rapidly, in an actively evolving, operational environment, with limited information, as a means of triaging and prioritising cases. For example, imagine two referrals come into the unit at the same time. The first involves a man who has been corresponding angrily with his MP for many months or years about a grievance. He has sent a new letter with a change in the tone of his language. He says he has reached the end of the line, that there will be no more correspondence, and he will be coming to the MP's surgery that evening to end matters once and for all. The second consists of a large volume of sexually explicit correspondence from a woman to a young prince, insisting that they are destined to be together and asking him to collect her from Heathrow airport when her flight arrives from the USA. Both cases will require some intervention, but it is the former that would be given a higher priority – the sender is angry, demanding, using last-resort language and has given a deadline. However, further inquiries reveal that the correspondent in the former case is serving a lengthy prison sentence and is in no position to visit the MP's surgery, whereas the woman with erotomania is currently on a flight from the USA. The prioritisation of the cases would reverse.

The motivational typologies described, together with warning behaviours and risk factors, can be used to assign cases to prioritisation categories. Structured professional judgement tools from the stalking literature can assist (e.g. the Stalking Risk Profile (MacKenzie Reference MacKenzie, McEwan and Pathé2009)). They can also help in identifying a range of treatment targets in individual cases – for example, reminding one to think not only about the treatment of the psychotic illness, but also to address the patient's substance misuse, social isolation and cognitive distortions related to their perceived relationship with the public figure. They are also useful in maintaining focus on managing risks in a variety of domains, not only violence, but also persistence, escalation or disruption. What they cannot do is predict bad outcomes with certainty. Rare events, such as suicides, are mathematically impossible to predict, however good the risk assessment tool (Nielssen Reference Nielssen, Wallace and Large2017), because the positive predictive value is dependent on the prevalence of the condition being measured. In rare conditions, even if the test used does much better than chance or clinical opinion, it will still be wrong much more often that it is right (a high rate of false positives).

Large & Nielssen (Reference Large and Nielssen2011) advocate focusing on the welfare of the patient rather than the current preoccupation with risks of suicide or violence. This view has some merit, as it is not possible to predict what any individual will do. However, the use of risk factors does allow for the identification of the group of fixated loners most at risk, and from which dangerous behaviour is most likely to arise. Intervening and treating the risk factors in this entire group will prevent adverse outcomes without needing to know which individuals would have gone on to engage in the behaviour in question. If one's response is a reactive intervention (to perceived ‘risk’), one is always in a state of chasing the behaviour or risk being exhibited, hoping to minimise the potential violence or communicated threat. In taking a preventive approach, we endeavour to assist as many people as possible into care, prioritising those in higher-risk groups when they are ill, thereby doing them a health good and preventing adverse outcomes.

Interventions

Threat management units have no additional powers: the focus is on making referrals to the relevant agencies responsible for managing the person, sharing information to enable them to make the best decisions they can and to coordinate their responses. The efficacy of an intervention will also be dependent on the context of the country, its government, laws, health system and sanctions. Given the high prevalence of psychosis among stalkers, the most common interaction in the UK is between the threat management unit and the local health service (the person's family doctor or mental health team). There is no barrier to information sharing in these circumstances – health professionals from two teams involved in the care of a patient (the threat management team and the local team) are sharing information to improve the patient's management (General Medical Council 2017). In general, the direction of information flow is from the threat management team to the local clinical team, providing them with additional information about their patient's behaviour and mental state from their interactions with the public figure.

Issues arising at FTAC when effecting an intervention in the UK

The interface between a threat management team and a mental health service, like all interfaces in psychiatry, often throws up interesting material. We have encountered forms of institutional acting out, including the ‘opt-in letter’ (Wilson Reference Wilson, Farnham and Henley2013) and the ‘consenting service’, two exquisite service designs for excluding individuals with anosognosia as a result of psychosis. This can require clinical escalation to obtain the appropriate response to referrals – expecting someone with grandiose messianic delusions to make an appointment if they'd like to discuss their mental health is plainly an absurd response from a mental health service.

There are genuine clinical complexities, though, in this work. How should one deal with a person with a chronic untreated psychotic illness who is refusing treatment when compulsion under mental health legislation does not appear to be a proportionate response? What do we mean when we say a person in a chronically angry and paranoid mental state is not a risk to themselves or others? They are clearly at much higher relative risk of violence than someone (including themselves) not in that mental state (Coid Reference Coid, Ullrich and Kallis2013). Perhaps we mean that the violence is not imminent? In which case, we are assuming (or hoping) we will find out about any violence shortly before it happens so that we can intervene at that point. Is this reasonable or wise? Perhaps we mean the absolute risk is low, but then the absolute risk of serious violence will always be low. ‘Playing the odds’ is no substitute for clinical judgement. We know that psychiatric thinking about this area, even in experts, is muddled (Large Reference Large, Ryan and Callaghan2012).

These difficulties can often be fruitfully explored in individual cases via professionals' meetings, involving the relevant agencies (perhaps the local mental health team, family doctor, threat management team and local police). Psychotic illnesses require treatment in the usual way with antipsychotic medication, psychological therapy and family support. The importance of occupational therapy cannot be overstated, although is often difficult to obtain, in helping people replace the hole left in their lives when they choose to stop their pursuit of the public figure. Delusional disorders, despite received clinical wisdom, are just as responsive to treatment as other psychotic illnesses. Manschreck & Khan (Reference Manschreck and Khan2006), in their review of 134 published case reports of delusional disorder, report that 89.6% either recovered or improved in response to treatment (mostly with antipsychotic medication). The therapeutic pessimism associated with such cases is not warranted.

In some cases, the criminal justice system may need to be involved, because there is no mental disorder at all, or as an additional means of providing mental healthcare (perhaps via the criminal sections of mental health legislation or by adding mental health treatment to a community order) or to provide the person with a clear message that their behaviour is unwanted and unlawful. The criminal offences most likely to be involved are those of harassment, stalking or malicious communications. A threat management team, often being jointly staffed by mental health clinicians and police officers, may be in a unique position to help oversee and coordinate multi-agency responses like this.

Threat management teams also offer advice and training in relation to this unusual area of practice, giving presentations to police forces and mental health services, sharing best practice and being involved with the planning of large events (Pathé Reference Pathé, Haworth and Lowry2015). Hoffmann & Sheridan (Reference Hoffmann, Sheridan, Meloy, Sheridan and Hoffmann2008) observed that historically there was little sympathy for celebrity victims of stalking, that in some way this behaviour ‘goes with the territory’ and they should endure it. However, the police need to take such reports seriously and to utilise multi-agency forms to achieve best results.